Technology « Shahzad Bhatti

January 1, 2024

Lessons from “Tidy First?”

Filed under: Technology — admin @ 1:59 pm

I just finished reading “Tidy First?” by Kent Beck. I have been following Kent for a long time since early days of XUnit and Extreme programming in late 90s. Kent is one of few influential people who have changed the landscape of software development and I have discussed some of his work in previous blogs such as Tips from Implementation Patterns, Responsibility vs Accountability, Heuristics from “Code That Fits in Your Head”, etc. In this book, Kent describes the concept of Tidy to clean up messy code so that you can make changes easily. The concept of Tidying is very similar to “Refactoring”, and though Refactoring originally meant changing the structure without changing the behavior but it has lost that distinction and now applies to both structure and behavior changes. In addition, the question mark in the book title implies that you may choose to tidy your code based on cost and benefits from the changes. This book is also heavily influenced by Structured Design: Fundamentals of a Discipline of Computer Program and Systems Design by Edward Yourdon and Larry Constantine who introduced the concept of coupling and cohesion, which is very fundamental to the software design. I also read that book in late 80s when I first learned about the structured design and programming with concepts such as a single entry and a single exit, cohesion and coupling, etc.

The “Tidy First?” book is very light and concise, which assumes that you are already familiar with a lot of fundamental concepts of software design but following are a few essential lessons from the book:

1. Guard Clauses

This is part of the first section that describes techniques for tidying that suggests adding guard clauses to routines. The author does not mention structured programming directly and only gives an example of FORTRAN that frowned upon multiple returns because debugging with multiple returns in that environment was hard. Instead the author advocates early return to reduce nested conditional logic but cautions against overuse of too many early returns. This advice is also similar to the Replace Nested Conditional with Guard Clauses technique from the Refactoring book though the author does not references it.

2. Dead Code

The author advocates deleting dead code, which is no-brainer.

3. Normalize Symmetries

The author advocates a consistency especially when different parts of a system are grown organically that might solve same problem differently. For example, you might see different ways of lazy initialization in different parts of the system that can confuse readers.

4. New Interface, Old Implementation

When dealing with a complex interface or implementation, Kent suggests creating a new pass through interface that simplifies the interaction with the old implementation. The original Refactoring book by Martin Fowler also includes a number of similar patterns and techniques such as Composing Methods, Extract Method, Replace Method with Method Object, Extract Class, etc. but the author does not reference those in the book.

5. Reading Order

When working with a long file, you may be tempted to refactor it into multiple smaller modules but the author suggests reordering parts of the file so that it’s easy for a reader to understand it. The author again cautions against mixing behavior changes with structure changes to minimize risks associated with the tidying the code.

6. Cohesion Order

When extending an existing system requires changing several widely dispersed sports in the code, you may be tempted to refactor existing code to make code more cohesive. However, the author suggests just placing cohesive code next to each other first before larger refactoring and recommends decoupling only when:

cost((decoupling) + cost(change) < cost(coupling) + cost(change)

7. Move Declaration and Initialization Together

This is another no-brainer and unlike older languages that required explicit declaration of variables, most modern languages encourage declaration and initialization together.

8. Explaining Variables

Instead of using a complex expression upon initialization a struct or invoking a method, the author recommends using well named variables to extract the sub-expressions. This advice is similar to the Extract Variable technique from the Refactoring book though the author does not cite the source.

9. Explaining Constants

The advice of extracting constants dates back to the structured programming and design and Refactoring book also includes Replace Magic Number with Symbolic Constant for similar advice though the author does not cite those sources. However, the author does caution against adding coupling for constants, e.g., if you have ONE defined as 1 in a module, you don’t need to import that module just to reuse that constant.

10. Explicit Parameters

The author recommends passing explicit parameters instead of passing a map that may include a lot of unrelated properties. Similarly, instead of accessing environment variables deep in the code, extract them at the top-level and then pass them explicitly. This advice is similar to the Stamp coupling (data-structured coupling) that cautions against passing a composite data structure when a routine uses only parts of it though the author does not references it in the book.

Note: other forms of coupling defined in the structured design (not referenced in the book) includes content coupling (high – information hiding), common coupling (sharing global data), external coupling (same format/protocol), control coupling (passing flag to control the flow), and data coupling (sharing parameters).

11. Chunk Statement

This technique simply recommends adding a blank line between parts of the code that are closely related. Though, you may use other refactoring techniques such as Extract Method, Move Method or other techniques but tidying up is less costly.

12. Extract Helper

In this technique, you extract a part of the code that has limited interaction with other code into a helper routine. In this case, the author does cite Extract Method from the Refactoring book. Another thing that you may consider that the author didn’t mention explicitly is to using one level of abstraction to facilitate reading code from top to bottom, which is cited in the Clean Code book.

13. One Pile

When dealing with a system where the code is split into many tiny pieces, the author recommends inlining the code into a single pile and then tidy it up. The author offers a few symptoms such as long, repeated argument lists; repeated cod with conditionals; poor naming of helper routines; shared mutable data structures for the code that can be tidy up. This technique is somewhat similar to Inline Class and Consolidate Duplicate Conditional Fragments techniques from the Refactoring book. In my experience, a lot of emphasis on unit testing to decoupling different parts of the system causes the code to split into tiny pieces but author does not go into specific reasons. Also, it may not be practical to create a single pile if there is a large complex code and author provides no practical examples for this technique.

14. Explaining Comments

This another common advice where the author suggests commenting only what wasn’t obvious from the code from the reader’s perspective. Other uses of comments can be a TODO when a code has a bug or requires changes for other limitations.

15. Delete Redundant Comments

This common advice probably could have combined with previous advice and suggests eliminating any comments that are obvious from the code.

16. Separate Tidying

This is part of the second part of the book that describes how tidying up fits into the software development lifecycle. When making changes to tidy up, you will have to consider how big are those changes, how many PRs you will need, and whether you should combine multiple PRs for structure. Ultimately, this requires some experimentation and customization based on your organization. Finally, author recommends omitting reviews for tidying PRs, which in my opinion is a bit risky unless you have fairly rigorous test harness. I have observed a trivial or innocuous change sometime can cause undefined behavior. For example, you might have removed dead-code but then find out later that some parts of the system still had some dependency on it.

17. Chaining

In this chapter, the author suggests tiny tidying steps and build upon various techniques for tidying up discussed in the first section such as guard clause, dead code, normalize symmetries, etc. The author is a bit apologetic for being anti-comment (or anti-documentation) as the agile movement specially extreme-programming does not explicitly calls for it but commenting (and documentation) is essential part for understanding the code.

18. Batch Size

This chapter focuses on how many batches of structure and behavior related PRs for tidy up should be made before integrating and deploying. This requires evaluating a tradeoff with various number of tidying per batch. For example, the cost and risk increases as the batch size grows due to merge conflicts, interactions with behavior changes, and more speculation due to additional cost.

The author recommends smaller batches to reduce the cost of review to reduce the cost of tidying up but ultimately the teams will have to find the right balance.

19. Rhythm

The author suggests managing rhythm for tidying up in batches where each change takes a few minutes or an hour before enabling desired behavior changes. The author cites Pareto principle and argues that 80% of the changes will occur in 20% of the files so you may end up touching the same code for tidying up, which is why the author suggests tidying up in small increments.

20. Getting Untangled

When making behavior changes, you may see a lot of opportunities for tidy up with structure changes but it may result in combination of structure and behavior changes. In that case, you may be tempted to create a single PR for all changes, which can be error prone; split tidyings into separate PRs, which may add more work; discard your changes and start over with tidying first. Though, sunk cost fallacy may discourage last option but the author recommends experimenting with the last option.

21. First, After, Later, Never

This chapter discusses tidying up with respect to a behavior change in the system. The author suggests Never when the code does not require changing the behavior or an old mantra “if it ain’t broke, don’t fix it.” The Later is generally a fantasy but basically you will have to estimate the work for tidying up. Another benefit is that you may want to learn about how the behavior evolves before tidying it up. The author suggests After when the code is messy and you will need to change the same code again soon so it will be cheaper to tidy now and cost of tidying is proportion to the cost of behavior change. The author suggests First when it will simplify making behavior change with immediate benefits as long as the cost can be amortized with future changes. These tradeoffs in this chapter are very similar to managing technical debt that most teams have to manage but the author does not use this term in the book.

22. Beneficially Relating Elements

This is part of third section of the book that discusses software design, cost of changes, trade-offs with investing in the structure of software, and principles for making changes. The author in this chapter defines software design as “beneficially relating elements,” where elements can be composed and have sub-elements such as atoms->molecules->crystals, and elements can have boundaries. The elements can have relations with other elements such as Invokes, Publishes, Listens, Refers, etc., and elements benefit each other in creating larger systems. The author gives an example of restructuring following code:

caller():
  return box.width() * box.height()

into following for simplification:

caller():
  return box.area()

Box>>area()
  return width() * height()

23. Structure and Behavior

The author describes behavior in terms of input/output pairs and invariants that should be maintained, which in turn creates value for businesses. The software creates value in providing the current behavior and how it can be extended for future changes. The author shows how software can provide greater value by supporting options for extension. This holds specially true for volatile environment and author relates lessons from his Extreme Programming Explained book, which I read back in 2004:

The structure of the system doesn’t matter directly to its behavior but it creates options for allowing behavior changes easily.

24. Economics: Time Value and Optionality

In this chapter, the author shares how he learned about the nature of money through a series of finance-related projects. The major lessons included:

A dollar today is worth more than a dollar tomorrow.
In a chaotic situation, options are better than things.

These strategies may conflict at times when earning now reduces future options, thus these trade-offs must be carefully considered using Net present value (NPV) and Options from the finance world.

25. A Dollar Today > A Dollar Tomorrow

A dollar today is more valuable than a dollar tomorrow because you can’t spend future dollar or invest it. As the software behavior creates value and makes money now so it encourages tidy after over tidy first. However, when cost of tidy-first + behavior-change is less than the cost of behavior change without tidying, then tidy first. In a nutshell, economic value of a software system is the sum of the discounted future cash flows.

26. Options

As discounted future cash flows conflict with optionality, you will need to find a balance in a Goldilocks world. The author shares how he learned about options pricing when working with trading software on Wall Street. These lessons included:

What behavior an I implement next has the value even before it’s implemented.
The more behavior in the portfolio, the more value will be created with optionality.
The more uncertain are the predictions of value are, the greater the value of the option is.

The author then defines basics of options, that allows you option to buy or sell based on underlying price, premium of the option, and duration of the option. Having worked in an options trading company myself, I understand the value of options in risk management though chapter does not cover those topics.

In terms of the software design with optionality:

The more volatile the value of a potential behavior change, the better.
The longer the development duration, the better.
The cheaper the software cost in future, the better.
The less design work to create an option, the better.

27. Options Versus Cash Flow

The discounted cash flow biases towards making money sooner and don’t tidy first and options on the other side biases towards spending money now to make more money later or tidy first/after/later. Author recommends tidy first when

cost(tidying) + cost(behavior-change after tidying) < cost(behavior-change without tidying)

Though, in some cases you may want to tidy firs even if the cost with tidy-first is a bit higher by amortizing the cost against other future changes.

28. Reversible Structure Changes

The structure changes are generally reversible but behavior changes are not easily reversible, thus require rigorous reviews and validation. Also, certain design changes such as extracting a service is not reversible and requires careful planning, testing and feature-flags, etc. At Amazon where I work, these kind of decisions are called “one-way doors” and “two-way doors” decisions, which are considered for any changes to the APIs and implementation.

29. Coupling

This book narrates insights from Structured Design: Fundamentals of a Discipline of Computer Program and Systems Design where authors of the book noticed that all expensive programs had one property in common: changing one element required changing other elements, which was termed as coupling. Coupling drives the cost of software and more elements are coupled, the higher the cost of change will be. In addition, some coupling can be cascading where each element triggers more changes for dependent elements. The author also refers to “connascence” to describe coupling from Meilir Page-Jones that includes (not referenced in the book):

Static connascences such as Connascence of name (CoN), Connascence of type (CoT), Connascence of meaning (CoM) or connascence of convention (CoC), Connascence of position (CoP), Connascence of algorithm (CoA).
Dynamic connascence such as Connascence of execution (CoE), Connascence of timing (CoT), Connascence of timing is when the timing of the execution of multiple components is important., Connascence of values (CoV), Connascence of identity (CoI)

Though, the author does not discuss other forms of coupling but it’s worth knowing common coupling in object-oriented systems such as Subclass-coupling, Dynamic coupling, Semantic coupling, Logical coupling, and Temporal coupling.

In summary, it’s essential to understand dependent changes before tidying up. One thing lacking in this chapter was use of tooling as a lot of static analysis and dependency management tools can be used to gather this kind of data.

30. Constantine’s Equivalence

In Structured Design: Fundamentals of a Discipline of Computer Program and Systems Design, the authors (Larry/Yourdon) postulated that the goal of software design is to minimize the cost of software as most of the software lifecycle goes into maintenance. Kent defines Constantine’s Equivalence as:

cost(software) ~= cost(change)

The author uses power law distribution (bell graph) to graph cost per time that grows slowly, then rapidly, then shrinks. In other words:

cost(change) ~= cost(big change)

cost(big change) ~= coupling

cost(software) ~= coupling

31. Coupling Versus Decoupling

You may choose not to remove coupling due to its cost as defined by discounted cash flow. In other cases, coupling does not become problem until now. Finally, having some coupling is just inevitable. For example, you may try to reduce coupling in one part of the software but it shifts towards other parts. This means that you should try to find a balance by evaluating trades off with cost of decoupling.

32. Cohesion

When decoupling, you may group sub-elements under the same containing element with more cohesion and move unrelated elements out of the group. When sub-elements are changed together, they benefit from cohesion with ease of change and analysis. On the other hand uncoupled elements can be put under their own submodules or sibling modules, which require careful consideration. Other factors that I have found to be useful, which are not mentioned in the book include analyzing interactions and direction of communication. For example, when two elements are more chatty then they should be closer together. Similarly, I learned from Large-Scale C++ Software Design back in 90s that software design should try to remove any cyclic dependency across modules, which generally requires extracting common behavior into a separate module.

33. Conclusion

In this chapter, the author summarizes that “tidy first” is impacted by cost of change, revenue it generates, how coupling can be reduced to make changes faster, and how it will add cohesion so that changes are in smaller scope. The author encourages tidying up the code but cautions against against doing too much. Instead, tidy up to enable the next behavior change. This book is part of a series and though this book focuses on individuals but subsequent series will examine the relationships between changers as shown below:

Who? | When? | What? | How? | Why?
You | Minutes-Hours | Tidying | SB diff | Coupling and Cohesion
You/Team | Days to Weeks | Refactoring | Wkly plan | Power laws
Stakeholders | Months to years| Arch. evolution | Dyn blnce | ?

Overall, this book addresses essential tradeoffs between quality and speed of software delivery that each team has to face. As this book was very concise, I felt it missed a number of related topics. For example, first section many of techniques and patterns that are better explained in other books such as Refactoring book. This book also does not cover testing and use of static analysis tools such as lint, clippy, findbugs, checkstyle, etc. that can help identify code smells, dead code, cyclic dependencies, cyclomatic complexity, etc. Another dimension of software development is non-functional requirements that addresses quality, scalability, reliability, security, etc. For example, in chapter 23, the author claims that if you can send 1,000 notifications, you can almost certainly send 100,000 notifications but this requires architectural support for scaling and isn’t always guaranteed. In addition, with scale you also have to deal with failures such as timeout, retries, unavailability, etc., which is of lesser issue at smaller scale. I also felt that the author avoided some of well-defined terms such as technical-debt and quality metrics that are commonly used when dealing with tidying or refactoring your code. The insight to compare “tidying your code” (or technical-debt) to options trading isn’t new. For example, Steve Freeman wrote blog entry, Bad code isn’t Technical Debt, it’s an unhedged Call Option in 2010 about this and Jay Fields expanded on it with types of technical-debt in 2011. Ultimately, the tradeoffs related to tidying your code depends on your organization. For instance, I have worked a number of startups where they want to be paid now otherwise they won’t exist to earn the dollar tomorrow as well as large financial and technology companies where the quality of the code in terms of structure and non-functional aspects are just as essential as the behavior change. Nevertheless, this tiny book encourages individual contributors to make incremental improvements to the code and I look forward to Kent’s next books in the series.

Comments (0)

August 26, 2023

Modern Software Development Process

Filed under: Methodologies,Project Management,Technology — admin @ 5:20 pm

The technologies, methodologies, and paradigms that have shaped the way software is designed, built, and delivered have evolved from the Waterfall model, proposed in 1970 by Dr. Winston W. Royce to Agile model in early 2000. The Waterfall model was characterized by sequential and rigid phases, detailed documentation, and milestone-based progression and suffered from inflexibility to adapt to changes, longer time-to-market, high risk and uncertainty difficulty in accurately estimating costs and time, silos between teams, and late discovery of issues. The Agile model addresses these shortcomings by adopting iterative and incremental development, continuous delivery, feedback loops, lower risk, and cross-functional teams. In addition, modern software development process over the last few years have adopted DevOps integration, automated testing, microservices and modular architecture, cloud services, infrastructure as code, and data driven decisions to improve the time to market, customer satisfaction, cost efficiency, operational resilience, collaboration, and sustainable development.

1. Goals

The goals of modern software development process includes:

Speed and Agility: Accelerate the delivery of high-quality software products and updates to respond to market demands and user needs.
Quality and Reliability: Ensure that the software meets specified requirements, performs reliably under all conditions, and is maintainable and scalable.
Reduced Complexity: Simplify complex tasks with use of refactoring and microservices.
Customer-Centricity: Develop features and products that solve real problems for users and meet customer expectations for usability and experience.
Scalability: Design software that can grow and adapt to changing user needs and technological advancements.
Cost-Efficiency: Optimize resource utilization and operational costs through automation and efficient practices.
Security: Ensure the integrity, availability, and confidentiality of data and services.
Innovation: Encourage a culture of innovation and learning, allowing teams to experiment, iterate, and adapt.
Collaboration and Transparency: Foster a collaborative environment among cross-functional teams and maintain transparency with stakeholders.
Employee Satisfaction: Focus on collaboration, autonomy, and mastery can lead to more satisfied, motivated teams.
Learning and Growth: Allows teams to reflect on their wins and losses regularly, fostering a culture of continuous improvement.
Transparency: Promote transparency with regular meetings like daily stand-ups and sprint reviews.
Flexibility: Adapt to changes even late in the development process.
Compliance and Governance: Ensure that software meets industry-specific regulations and standards.
Risk Mitigation: Make it easier to course-correct, reducing the risk of late project failure.
Competitive Advantage: Enables companies to adapt to market changes more swiftly, offering a competitive edge.

2. Elements of Software Development Process

Key elements of the Software Development Cycle include:

Discovery & Ideation: Involves brainstorming, feasibility studies, and stakeholder buy-in. Lean startup methodologies might be applied for new products.
Continuous Planning: Agile roadmaps, iterative sprint planning, and backlog refinement are continuous activities.
User-Centric Design: The design process is iterative, human-centered, and closely aligned with development.
Development & Testing: Emphasizes automated testing, code reviews, and incremental development. CI/CD pipelines automate much of the build and testing process.
Deployment & Monitoring: DevOps practices facilitate automated, reliable deployments. Real-time monitoring tools and AIOps (Artificial Intelligence for IT Operations) can proactively manage system health.
Iterative Feedback Loops: Customer feedback, analytics, and KPIs inform future development cycles.

3. Roles in Modern Processes

Critical Roles in Contemporary Software Development include:

Product Owner/Product Manager: This role serves as the liaison between the business and technical teams. They are responsible for defining the product roadmap, prioritizing features, and ensuring that the project aligns with stakeholder needs and business objectives.
Scrum Master/Agile Coach: These professionals are responsible for facilitating Agile methodologies within the team. They help organize and run sprint planning sessions, stand-ups, and retrospectives.
Development Team: This group is made up of software engineers and developers responsible for the design, coding, and initial testing of the product. They collaborate closely with other roles, especially the Product Owner, to ensure that the delivered features meet the defined requirements.
DevOps Engineers: DevOps Engineers act as the bridge between the development and operations teams. They focus on automating the Continuous Integration/Continuous Deployment (CI/CD) pipeline to ensure that code can be safely, efficiently, and reliably moved from development into production.
QA Engineers: Quality Assurance Engineers play a vital role in the software development life cycle by creating automated tests that are integrated into the CI/CD pipeline.
Data Scientists: These individuals use analytical techniques to draw actionable insights from data generated by the application. They may look at user behavior, application performance, or other metrics to provide valuable feedback that can influence future development cycles or business decisions.
Security Engineers: Security Engineers are tasked with ensuring that the application is secure from various types of threats. They are involved from the early stages of development to ensure that security is integrated into the design (“Security by Design”).

Each of these roles plays a critical part in modern software development, contributing to more efficient processes, higher-quality output, and ultimately, greater business value.

4. Phases of Modern SDLC

In the Agile approach to the software development lifecycle, multiple phases are organized not in a rigid, sequential manner as seen in the Waterfall model, but as part of a more fluid, iterative process. This allows for continuous feedback loops and enables incremental delivery of features. Below is an overview of the various phases that constitute the modern Agile-based software development lifecycle:

4.1 Inception Phase

The Inception phase is the initial stage in a software development project, often seen in frameworks like the Rational Unified Process (RUP) or even in Agile methodologies in a less formal manner. It sets the foundation for the entire project by defining its scope, goals, constraints, and overall vision.

4.1.1 Objectives

The objectives of inception phase includes:

Project Vision and Scope: Define what the project is about, its boundaries, and what it aims to achieve.
Feasibility Study: Assess whether the project is feasible in terms of time, budget, and technical constraints.
Stakeholder Identification: Identify all the people or organizations who have an interest in the project, such as clients, end-users, developers, and more.
Risk Assessment: Evaluate potential risks, like technical challenges, resource limitations, and market risks, and consider how they can be mitigated.
Resource Planning: Preliminary estimation of the resources (time, human capital, budget) required for the project.
Initial Requirements Gathering: Collect high-level requirements, often expressed as epics or user stories, which will be refined later.
Project Roadmap: Create a high-level project roadmap that outlines major milestones and timelines.

4.1.2 Practices

The inception phase includes following practices:

Epics Creation: High-level functionalities and features are described as epics.
Feasibility Study: Analyze the technical and financial feasibility of the proposed project.
Requirements: Identify customers and document requirements that were captured from the customers. Review requirements with all stakeholders and get alignment on key features and deliverables. Be careful for a scope creep so that the software change can be delivered incrementally.
Estimated Effort: A high-level effort in terms of estimated man-month, t-shirt size, or story-points to build and deliver the features.
Business Metrics: Define key business metrics that will be tracked such as adoption, utilization and success criteria for the features.
Customer and Dependencies Impact: Understand how the changes will impact customers as well as upstream/downstream dependencies.
Roadmapping: Develop a preliminary roadmap to guide the project’s progression.

4.1.3 Roles and Responsibilities:

The roles and responsibilities in inception phase include:

Stakeholders: Validate initial ideas and provide necessary approvals or feedback.
Product Owner: Responsible for defining the vision and scope of the product, expressed through a set of epics and an initial product backlog. The product manager interacts with customers, defines the customer problems and works with other stakeholders to find solutions. The product manager defines end-to-end customer experience and how the customer will benefit from the proposed solution.
Scrum Master: Facilitates the Inception meetings and ensures that all team members understand the project’s objectives and scope. The Scrum master helps write use-cases, user-stories and functional/non-functional requirements into the Sprint Backlog.
Development Team: Provide feedback on technical feasibility and initial estimates for the project scope.

By carefully executing the Inception phase, the team can ensure that everyone has a clear understanding of the ‘what,’ ‘why,’ and ‘how’ of the project, thereby setting the stage for a more organized and focused development process.

4.2. Planning Phase

The Planning phase is an essential stage in the Software Development Life Cycle (SDLC), especially in methodologies that adopt Agile frameworks like Scrum or Kanban. This phase aims to detail the “how” of the project, which is built upon the “what” and “why” established during the Inception phase. The Planning phase involves specifying objectives, estimating timelines, allocating resources, and setting performance measures, among other activities.

4.2.1 Objectives

The objectives of planning phase includes:

Sprint Planning: Determine the scope of the next sprint, specifying what tasks will be completed.
Release Planning: Define the features and components that will be developed and released in the upcoming iterations.
Resource Allocation: Assign tasks and responsibilities to team members based on skills and availability.
Timeline and Effort Estimation: Define the timeframe within which different tasks and sprints should be completed.
Risk Management: Identify, assess, and plan for risks and contingencies.
Technical Planning: Decide the architectural layout, technologies to be used, and integration points, among other technical aspects.

4.2.2 Practices:

The planning phase includes following practices:

Sprint Planning: A meeting where the team decides what to accomplish in the coming sprint.
Release Planning: High-level planning to set goals and timelines for upcoming releases.
Backlog Refinement: Ongoing process to add details, estimates, and order to items in the Product Backlog.
Effort Estimation: Techniques like story points, t-shirt sizes, or time-based estimates to assess how much work is involved in a given task or story.

4.2.3 Roles and Responsibilities:

The roles and responsibilities in planning phase include:

Product Owner: Prioritizes the Product Backlog and clarifies details of backlog items. Helps the team understand which items are most crucial to the project’s success.
Scrum Master: Facilitates planning meetings and ensures that the team has what they need to complete their tasks. Addresses any impediments that the team might face.
Development Team: Provides effort estimates for backlog items, helps break down stories into tasks, and commits to the amount of work they can accomplish in the next sprint.
Stakeholders: May provide input on feature priority or business requirements, though they typically don’t participate in the detailed planning activities.

4.2.4 Key Deliverables:

Key deliverables in planning phase include:

Sprint Backlog: A list of tasks and user stories that the development team commits to complete in the next sprint.
Execution Plan: A plan outlining the execution of the software development including dependencies from other teams. The plan also identifies key reversible and irreversible decisions.
Deployment and Release Plan: A high-level plan outlining the features and major components to be released over a series of iterations. This plan also includes feature flags and other configuration parameters such as throttling limits.
Resource Allocation Chart: A detailed breakdown of who is doing what.
Risk Management Plan: A documentation outlining identified risks, their severity, and mitigation strategies.

By thoroughly engaging in the Planning phase, teams can have a clear roadmap for what needs to be achieved, how it will be done, who will do it, and when it will be completed. This sets the project on a course that maximizes efficiency, minimizes risks, and aligns closely with the stakeholder requirements and business objectives.

4.3. Design Phase

The Design phase in the Software Development Life Cycle (SDLC) is a critical step that comes after planning and before coding. During this phase, the high-level architecture and detailed design of the software system are developed. This serves as a blueprint for the construction of the system, helping the development team understand how the software components will interact, what data structures will be used, and how user interfaces will be implemented, among other things.

4.3.1 Objectives:

The objectives of design phase includes:

Architectural Design: Define the software’s high-level structure and interactions between components. Understand upstream and downstream dependencies and how architecture decisions affect those dependencies. The architecture will ensure scalability to a minimum of 2x of the peak traffic and will build redundancy to avoid single points of failure. The architecture decisions should be documented in an Architecture Decision Record format and should be defined in terms of reversible and irreversible decisions.
Low-Level Design: Break down the architectural components into detailed design specifications including functional and non-functional considerations. The design documents should include alternative solutions and tradeoffs when selecting a recommended approach. The design document should consider how the software can be delivered incrementally and whether multiple components can be developed in parallel. Other best practices such as not breaking backward compatibility, composable modules, consistency, idempotency, pagination, no unbounded operations, validation, and purposeful design should be applied to the software design.
API Design: Specify the endpoints, request/response models, and the underlying architecture for any APIs that the system will expose.
User Interface Design: Design the visual elements and user experiences.
Data Model Design: Define the data structures and how data will be stored, accessed, and managed.
Functional Design: Elaborate on the algorithms, interactions, and methods that will implement software features.
Security Design: Implement security measures like encryption, authentication, and authorization.
Performance Considerations: Design for scalability, reliability, and performance optimization.
Design for Testability: Ensure elements can be easily tested.
Operational Excellence: Document monitoring, logging, operational and business metrics, alarms, and dashboards to monitor health of the software components.

4.3.2 Practices:

The design phase includes following practices:

Architectural and Low-Level Design Reviews: Conducted to validate the robustness, scalability, and performance of the design. High-level reviews focus on the system architecture, while low-level reviews dig deep into modules, classes, and interfaces.
API Design Review: A specialized review to ensure that the API design adheres to RESTful principles, is consistent, and meets performance and security benchmarks.
Accessibility Review: Review UI for accessibility support.
Design Patterns: Utilize recognized solutions for common design issues.
Wireframing and Prototyping: For UI/UX design.
Sprint Zero: Sometimes used in Agile for initial setup, including design activities.
Backlog Refinement: Further details may be added to backlog items in the form of acceptance criteria and design specs.

4.3.3 Roles and Responsibilities:

The roles and responsibilities in design phase include:

Product Owner: Defines the acceptance criteria for user stories and validates that the design meets business and functional requirements.
Scrum Master: Facilitates design discussions and ensures that design reviews are conducted effectively.
Development Team: Creates the high-level architecture, low-level detailed design, and API design. Chooses design patterns and data models.
UX/UI Designers: Responsible for user interface and user experience design.
System Architects: Involved in high-level design and may also review low-level designs and API specifications.
API Designers: Specialized role focusing on API design aspects such as endpoint definition, rate limiting, and request/response models.
QA Engineers: Participate in design reviews to ensure that the system is designed for testability.
Security Engineers: Involved in both high-level and low-level design reviews to ensure that security measures are adequately implemented.

4.3.4 Key Deliverables:

Key deliverables in design phase include:

High-Level Architecture Document: Describes the architectural layout and high-level components.
Low-Level Design Document: Provides intricate details of each module, including pseudo-code, data structures, and algorithms.
API Design Specification: Comprehensive documentation for the API including endpoints, methods, request/response formats, and error codes.
Database Design: Includes entity-relationship diagrams, schema designs, etc.
UI Mockups: Prototypes or wireframes of the user interface.
Design Review Reports: Summaries of findings from high-level, low-level, and API design reviews.
Proof of Concept and Spikes: Build proof of concepts or execute spikes to learn about high risk design components.
Architecture Decision Records: Documents key decisions that were made including tradeoffs, constraints, stakeholders and final decision.

The Design phase serves as the blueprint for the entire project. Well-thought-out and clear design documentation helps mitigate risks, provides a clear vision for the development team, and sets the stage for efficient and effective coding, testing, and maintenance.

4.4 Development Phase

The Development phase in the Software Development Life Cycle (SDLC) is the stage where the actual code for the software is written, based on the detailed designs and requirements specified in the preceding phases. This phase involves coding, unit testing, integration, and sometimes even preliminary system testing to validate that the implemented features meet the specifications.

4.4.1 Objectives:

The objectives of development phase includes:

Coding: Translate design documentation and specifications into source code.
Modular Development: Build the software in chunks or modules for easier management, testing, and debugging.
Unit Testing: Verify that each unit or module functions as designed.
Integration: Combine separate units and modules into a single, functioning system.
Code Reviews: Validate that code adheres to standards, guidelines, and best practices.
Test Plans: Review test plans for all test-cases that will validate the changes to the software.
Documentation: Produce inline code comments, READMEs, and technical documentation.
Early System Testing: Sometimes, a form of system testing is performed to catch issues early.

4.4.2 Practices:

The development phase includes following practices:

Pair Programming: Two developers work together at one workstation, enhancing code quality.
Test-Driven Development (TDD): Write failing tests before writing code, then write code to pass the tests.
Code Reviews: Peer reviews to maintain code quality.
Continuous Integration: Frequently integrate code into a shared repository and run automated tests.
Version Control: Use of tools like Git to manage changes and history.
Progress Report: Use of burndown charts, risk analysis, blockers and other data to apprise stakeholders up-to-date with the progress of the development effort.

4.4.3 Roles and Responsibilities:

The roles and responsibilities in development phase include:

Product Owner: Prioritizes features and stories in the backlog that are to be developed, approves or rejects completed work based on acceptance criteria.
Scrum Master: Facilitates daily stand-ups, removes impediments, and ensures the team is focused on the tasks at hand.
Software Developers: Write the code, perform unit tests, and integrate modules.
QA Engineers: Often involved early in the development phase to write automated test scripts, and to run unit tests along with developers.
DevOps Engineers: Manage the CI/CD pipeline to automate code integration and preliminary testing.
Technical Leads/Architects: Provide guidance and review code to ensure it aligns with architectural plans.

4.4.4 Key Deliverables:

Key deliverables in development phase include:

Source Code: The actual code files, usually stored in a version control system.
Unit Test Cases and Results: Documentation showing what unit tests have been run and their results.
Code Review Reports: Findings from code reviews.
Technical Documentation: Sometimes known as codebooks or developer guides, these detail how the code works for future maintenance.

4.4.5 Best Practices for Development:

Best practices in development phase include:

Clean Code: Write code that is easy to read, maintain, and extend.
Modularization: Build in a modular fashion for ease of testing, debugging, and scalability.
Commenting and Documentation: In-line comments and external documentation to aid in future maintenance and understanding.
Code Repositories: Use version control systems to keep a historical record of all changes and to facilitate collaboration.

The Development phase is where the design and planning turn into a tangible software product. Following best practices and guidelines during this phase can significantly impact the quality, maintainability, and scalability of the end product.

4.5. Testing Phase

The Testing phase in the Software Development Life Cycle (SDLC) is crucial for validating the system’s functionality, performance, security, and stability. This phase involves various types of testing methodologies to ensure that the software meets all the requirements and can handle expected and unexpected situations gracefully.

4.5.1 Objectives:

The objectives of testing phase includes:

Validate Functionality: Ensure all features work as specified in the requirements.
Ensure Quality: Confirm that the system is reliable, secure, and performs optimally.
Identify Defects: Find issues that need to be addressed before the software is deployed.
Risk Mitigation: Assess and mitigate potential security and performance risks.

4.5.2 Practices:

The testing phase includes following practices:

Test Planning: Create a comprehensive test plan detailing types of tests, scope, schedule, and responsibilities.
Test Case Design: Prepare test cases based on various testing requirements.
Test Automation: Automate repetitive and time-consuming tests.
Continuous Testing: Integrate testing into the CI/CD pipeline for continuous feedback.

4.5.3 Types of Testing and Responsibilities:

The test phase includes following types of testing:

Integration Testing:
Responsibilities: QA Engineers, DevOps Engineers
Objective: Verify that different modules or services work together as expected.
Functional Testing:
Responsibilities: QA Engineers
Objective: Validate that the system performs all the functions as described in the specifications.
Load/Stress Testing:
Responsibilities: Performance Test Engineers, DevOps Engineers
Objective: Check how the system performs under heavy loads.
Security and Penetration Testing:
Responsibilities: Security Engineers, Ethical Hackers
Objective: Identify vulnerabilities that an attacker could exploit.
Canary Testing:
Responsibilities: DevOps Engineers, Product Owners
Objective: Validate that new features or updates will perform well with a subset of the production environment before full deployment.
Other Forms of Testing:
Smoke Testing: Quick tests to verify basic functionality after a new build or deployment.
Regression Testing: Ensure new changes havenâ€™t adversely affected existing functionalities.
User Acceptance Testing (UAT): Final validation that the system meets business needs, usually performed by stakeholders or end-users.

4.5.4 Key Deliverables:

Test Plan: Describes the scope, approach, resources, and schedule for the testing activities.
Test Cases: Detailed description of what to test, how to test, and expected outcomes.
Test Scripts: Automated scripts for testing.
Test Reports: Summary of testing activities, findings, and results.

4.5.5 Best Practices for Testing:

Early Involvement: Involve testers from the early stages of SDLC for better understanding and planning.
Code Reviews: Conduct code reviews to catch potential issues before the testing phase.
Data-Driven Testing: Use different sets of data to evaluate how the application performs with various inputs.
Monitoring and Feedback Loops: Integrate monitoring tools to get real-time insights during testing and improve the test scenarios continuously.

The Testing phase is crucial for delivering a reliable, secure, and robust software product. The comprehensive testing strategy that includes various types of tests ensures that the software is well-vetted before it reaches the end-users.

4.6. Deployment Phase

The Deployment phase in the Software Development Life Cycle (SDLC) involves transferring the well-tested codebase from the staging environment to the production environment, making the application accessible to the end-users. This phase aims to ensure a smooth transition of the application from development to production with minimal disruptions and optimal performance.

4.6.1 Objectives:

The objectives of deployment phase includes:

Release Management: Ensure that the application is packaged and released properly.
Transition: Smoothly transition the application from staging to production.
Scalability: Ensure the infrastructure can handle the production load.
Rollback Plans: Prepare contingencies for reverting changes in case of failure.

4.6.2 Key Components and Practices:

The key components and practices in deployment phase include:

Continuous Deployment (CD):
Responsibilities: DevOps Engineers, QA Engineers, Developers
Objective: Automatically deploy all code changes to the production environment that pass the automated testing phase.
Tools: Jenkins, GitLab CI/CD, GitHub Actions, Spinnaker
Infrastructure as Code (IaC):
Responsibilities: DevOps Engineers, System Administrators
Objective: Automate the provisioning and management of infrastructure using code.
Tools: Terraform, AWS CloudFormation, Ansible
Canary Testing:
Responsibilities: DevOps Engineers, QA Engineers
Objective: Gradually roll out the new features to a subset of users before a full-scale rollout to identify potential issues.
Phased Deployment:
Responsibilities: DevOps Engineers, Product Managers
Objective: Deploy the application in phases to monitor its stability and performance, enabling easier troubleshooting.
Rollback:
Responsibilities: DevOps Engineers
Objective: Be prepared to revert the application to its previous stable state in case of any issues.
Feature Flags:
Responsibilities: Developers, Product Managers
Objective: Enable or disable features in real-time without deploying new code.
Resilience:
Responsibilities: DevOps Engineers, Developers
Objective: Prevent retry storms, throttle the number of client requests, and apply reliability patterns such as Circuit Breakers and BulkHeads. Avoid bimodal behavior and maintain software latency and throughput within the defined SLAs.
Scalability:
Responsibilities: DevOps Engineers, Developers
Objective: Monitor scalability limits and implement elastic scalability. Review scaling limits periodically.
Throttling and Sharding Limits (and other Configurations):
Responsibilities: DevOps Engineers, Developers
Objective: Review configuration and performance configurations such as throttling and sharding limits.
Access Policies:
Responsibilities: DevOps Engineers, Developers
Objective: Review access policies, permissions and roles that can access the software.

4.6.3 Roles and Responsibilities:

The deployment phase includes following roles and responsibilities:

DevOps Engineers: Manage CI/CD pipelines, IaC, and automate deployments.
Product Managers: Approve deployments based on feature completeness and business readiness.
QA Engineers: Ensure the application passes all tests in the staging environment before deployment.
System Administrators: Ensure that the infrastructure is ready and scalable for deployment.

4.6.4 Key Deliverables:

The deployment phase includes following key deliverables:

Deployment Checklist: A comprehensive list of tasks and checks before, during, and after deployment.
CD Pipeline Configuration: The setup details for the Continuous Deployment pipeline.
Automated Test Cases: Essential for validating the application before automatic deployment.
Operational, Security and Compliance Review Documents: These documents will define checklists about operational excellence, security and compliance support in the software.
Deployment Logs: Automated records of what was deployed, when, and by whom (usually by the automation system).
Logging: Review data that will be logged and respective log levels so that data privacy is not violated and excessive logging is avoided.
Deployment Schedule: A timeline for the deployment process.
Observability: Health dashboard to monitor operational and business metrics, and alarms to receive notifications when service-level-objectives (SLO) are violated. The operational metrics will include availability, latency, and error metrics. The health dashboard will monitor utilization of CPU, disk space, memory, and network resources.
Rollback Plan: A detailed plan for reverting to the previous stable version if necessary.

4.6.5 Best Practices:

Following are a few best practices for the deployment phase:

Automated Deployments: Use automation tools for deployment of code and infrastructure to minimize human error.
Monitoring and Alerts: Use monitoring tools to get real-time insights into application performance and set up alerts for anomalies.
Version Control: Ensure that all deployable artifacts are versioned for traceability.
Comprehensive Testing: Given the automated nature of Continuous Deployment, having a comprehensive suite of automated tests is crucial.
Rollback Strategy: Have an automated rollback strategy in case the new changes result in system failures or critical bugs.
Feature Toggles: Use feature flags to control the release of new features, which can be enabled or disabled without having to redeploy.
Audit Trails: Maintain logs and history for compliance and to understand what was deployed and when.
Documentation: Keep detailed records of the deployment process, configurations, and changes.
Stakeholder Communication: Keep all stakeholders in the loop regarding deployment schedules, success, or issues.
Feedback from Early Adopters: If the software is being released to internal or a beta customers, then capture feedback from those early adopters including any bugs report.
Marketing and external communication: The release may need to be coordinated with a marketing campaign so that customers can be notified about new features.

The Deployment phase is critical for ensuring that the software is reliably and securely accessible by the end-users. A well-planned and executed deployment strategy minimizes risks and disruptions, leading to a more dependable software product.

4.7 Maintenance Phase

The Maintenance phase in the Software Development Life Cycle (SDLC) is the ongoing process of ensuring the software’s continued effective operation and performance after its release to the production environment. The objective of this phase is to sustain the software in a reliable state, provide continuous support, and make iterative improvements or patches as needed.

4.7.1 Objectives:

The objectives of maintenance phase includes:

Bug Fixes: Address any issues or defects that arise post-deployment.
Updates & Patches: Release minor or major updates and patches to improve functionality or security.
Optimization: Tune the performance of the application based on metrics and feedback.
Scalability: Ensure that the software can handle growth in terms of users, data, and transaction volume.
Documentation: Update all documents related to software changes and configurations.

4.7.2 Key Components and Practices:

The key components and practices of maintenance phase include:

Incident Management:
Responsibilities: Support Team, DevOps Engineers
Objective: Handle and resolve incidents affecting the production environment.
Technical Debt Management:
Responsibilities: Development Team, Product Managers
Objective: Prioritize and resolve accumulated technical debt to maintain code quality and performance.
Security Updates:
Responsibilities: Security Engineers, DevOps Engineers
Objective: Regularly update and patch the software to safeguard against security vulnerabilities.
Monitoring & Analytics:
Responsibilities: DevOps Engineers, Data Analysts
Objective: Continuously monitor software performance, availability, and usage to inform maintenance tasks.
Documentation and Runbooks:
Responsibilities: Support Team, DevOps Engineers
Objective: Define cookbooks for development processes and operational issues.

4.7.3 Roles and Responsibilities:

The maintenance phase includes following roles and responsibilities:

DevOps Engineers: Monitor system health, handle deployments for updates, and coordinate with the support team for incident resolution.
Support Team: Provide customer support, report bugs, and assist in reproducing issues for the development team.
Development Team: Develop fixes, improvements, and updates based on incident reports, performance metrics, and stakeholder feedback.
Product Managers: Prioritize maintenance tasks based on customer needs, business objectives, and technical requirements.
Security Engineers: Regularly audit the software for vulnerabilities and apply necessary security patches.

4.7.4 Key Deliverables:

Key deliverables for the maintenance phase include:

Maintenance Plan: A detailed plan outlining maintenance activities, schedules, and responsible parties.
Patch Notes: Documentation describing what has been fixed or updated in each new release.
Performance Reports: Regular reports detailing the operational performance of the software.

4.7.5 Best Practices for Maintenance:

Following are a few best practices for the maintenance phase:

Automated Monitoring: Use tools like Grafana, Prometheus, or Zabbix for real-time monitoring of system health.
Feedback Loops: Use customer feedback and analytics to prioritize maintenance activities.
Version Control: Always version updates and patches for better tracking and rollback capabilities.
Knowledge Base: Maintain a repository of common issues and resolutions to accelerate incident management.
Scheduled Maintenance: Inform users about planned downtimes for updates and maintenance activities.

The Maintenance phase is crucial for ensuring that the software continues to meet user needs and operate effectively in the ever-changing technical and business landscape. Proper maintenance ensures the longevity, reliability, and continual improvement of the software product. Also, note that though maintenance is defined as a separate phase above but it will include other phases from inception to deployment and each change will be developed and deployed incrementally in iterative agile process.

5. Best Practices for Ensuring Reliability, Quality, and Incremental Delivery

Following best practices ensure incremental delivery with higher reliability and quality:

5.1. Iterative Development:

Embrace the Agile principle of delivering functional software frequently. The focus should be on breaking down the product into small, manageable pieces and improving it in regular iterations, usually two to four-week sprints.

Tools & Techniques: Feature decomposition, Sprint planning, Short development cycles.
Benefits: Faster time to market, easier bug isolation, and tracking, ability to incorporate user feedback quickly.

5.2. Automated Testing:

Implement Test-Driven Development (TDD) or Behavior-Driven Development (BDD) to script tests before the actual coding begins. Maintain these tests to run automatically every time a change is made.

Tools & Techniques: JUnit, Selenium, Cucumber.
Benefits: Instant feedback on code quality, regression testing, enhanced code reliability.

5.3. Design Review:

Detailed Explanation: A formal process where architects and developers evaluate high-level and low-level design documents to ensure they meet the project requirements, are scalable, and adhere to best practices.
Tools & Techniques: Design diagrams, UML, Peer Reviews, Design Review Checklists.
Benefits: Early identification of design flaws, alignment with stakeholders, consistency across the system architecture.

5.4 Code Reviews:

Before any code gets merged into the main repository, it should be rigorously reviewed by other developers to ensure it adheres to coding standards, is optimized, and is free of bugs.

Tools & Techniques: Git Pull Requests, Code Review Checklists, Pair Programming.
Benefits: Team-wide code consistency, early detection of anti-patterns, and a secondary check for overlooked issues.

5.5 Security Review:

A comprehensive evaluation of the security aspects of the application, involving both static and dynamic analyses, is conducted to identify potential vulnerabilities.

Tools & Techniques: OWASP Top 10, Security Scanners like Nessus or Qualys, Code Review tools like Fortify, Penetration Testing.
Benefits: Proactive identification of security vulnerabilities, adherence to security best practices, and compliance with legal requirements.

5.6 Operational Review:

Before deploying any new features or services, assess the readiness of the operational environment, including infrastructure, data backup, monitoring, and support plans.

Tools & Techniques: Infrastructure as Code tools like Terraform, Monitoring tools like Grafana, Documentation.
Benefits: Ensures the system is ready for production, mitigates operational risks, confirms that deployment and rollback strategies are in place.

5.7 CI/CD (Continuous Integration and Continuous Deployment):

Integrate all development work frequently and deliver changes to the end-users reliably and rapidly using automated pipelines.

Tools & Techniques: Jenkins, GitLab CI/CD, Docker, Kubernetes.
Benefits: Quicker discovery of integration bugs, reduced lead time for feature releases, and improved deployability.

5.8 Monitoring:

Implement sophisticated monitoring tools that continuously observe system performance, user activity, and operational health.

Tools & Techniques: Grafana, Prometheus, New Relic.
Benefits: Real-time data analytics, early identification of issues, and rich metrics for performance tuning.

5.9 Retrospectives:

At the end of every sprint or project phase, the team should convene to discuss what worked well, what didn’t, and how processes can be improved.

Tools & Techniques: Post-its for brainstorming, Sprint Retrospective templates, Voting mechanisms.
Benefits: Continuous process improvement, team alignment, and reflection on the project’s success and failures.

5.10 Product Backlog Management:

A live document containing all known requirements, ranked by priority and constantly refined to reflect changes and learnings.

Tools & Techniques: JIRA, Asana, Scrum boards.
Benefits: Focused development on high-impact features, adaptability to market or user needs.

5.11. Kanban for Maintenance:

For ongoing maintenance work and technical debt management, utilize a Kanban system to visualize work, limit work-in-progress, and maximize efficiency.

Tools & Techniques: Kanban boards, JIRA Kanban, Trello.
Benefits: Dynamic prioritization, quicker task completion, and efficient resource utilization.

5.12 Feature Flags:

Feature flags allow developers to toggle the availability of certain functionalities without deploying new versions.

Tools & Techniques: LaunchDarkly, Config files, Custom-built feature toggles.
Benefits: Risk mitigation during deployments, simpler rollbacks, and fine-grained control over feature releases.

5.13 Documentation:

Create comprehensive documentation, ranging from code comments and API docs to high-level architecture guides and FAQs for users.

Tools & Techniques: Wiki platforms, OpenAPI/Swagger for API documentation, Code comments.
Benefits: Streamlined onboarding for new developers, easier troubleshooting, and long-term code maintainability.

By diligently applying these multi-layered Agile best practices and clearly defined roles, your SDLC will be a well-oiled machineâ€”more capable of rapid iterations, quality deliverables, and high adaptability to market or user changes.

6. Conclusion

The Agile process combined with modern software development practices offers an integrated and robust framework for building software that is adaptable, scalable, and high-quality. This approach is geared towards achieving excellence at every phase of the Software Development Life Cycle (SDLC), from inception to deployment and maintenance. The key benefits of this modern software development process includes flexibility and adaptability, reduced time-to-market, enhanced quality, operational efficiency, risk mitigation, continuous feedback, transparency, collaboration, cost-effectiveness, compliance and governance, and documentation for sustainment. By leveraging these Agile and modern software development practices, organizations can produce software solutions that are not only high-quality and reliable but also flexible enough to adapt to ever-changing requirements and market conditions.

Comments (0)

June 24, 2023

Systems Performance

Filed under: Computing,Technology — admin @ 3:33 pm

This blog summarizes learnings from Systems Performance by Brendan Gregg that covers concepts, tools, and performance tuning for operating systems and applications.

1. Introduction

This chapter introduces Systems Performance including all major software and hardware components with goals to improve the end-user experience by reducing latency and computing cost. The author describes challenges for performance engineering such as subjectivity without clear goals, complexity that requires holistic approach, multiple causes, and multiple performance issues. The author defines key concepts for performance such as:

Latency: measurement of time spent waiting and it allows maximum speedup to be estimated.
Observability: refers to understanding a system through observation and includes that use counters, profiling, and tracing. This relies on counters, statistics, and metrics, which can be used to trigger alerts by the monitoring software. The profiling performs sampling to paint the picture of target. Tracing is event-based recording, where event data is captured and saved for later analysis using static instrumentation or dynamic instrumentation. The latest dynamic tools use BPF (Berkley Packet Filter) to build dynamic tracing, which is also referred as eBPF.
Experimentation tools: benchmark tools that test a specific component, e.g. following example performs a TCP network throughput:

iperf -c 192.168.1.101 -i 1 -t 10

The chapter also describes common Linux tools for analysis such as:

dmesg -T | tail
vmstat -SM 1
mpstat -P ALL 1
pidstat 1
iostat -sxz 1
free -m
sar -n DEV 1
sar -n TCP,ETCP 1

2. Methodologies

This chapter introduces common performance concepts such as IOPS, throughput, response-time, latency, utilization, saturation, bottleneck, workload, and cache. It defines following models of system performance like System under Test and Queuing System shown below:

The chapter continues with defining concepts such as latency, time-scales, trade-offs, tuning efforts, load vs architecture, scalability, and metrics. It defines utilization based on time as:

U = B / T
where U = utilization, B = total-busy-time, T = observation period

and in terms of capacity, e.g.

U = % used capacity

The chapter defines saturation as the degree to which a resource has queued work it cannot service and defines caching considerations to improve performance such as hit ratio, cold-cache, warm-cache, and hot-cache.

The author then describes analysis perspectives and defines resource analysis that begins with analysis of the system resources for investigating performance issues and capacity planning, and workload analysis that examines the performance of application for identifying and confirming latency issues. Next, author describes following methodologies for analysis:

Streetlight Anti-Method: It is an absence of a deliberate methodology and user analyzes a familiar tool but can be a hit or miss.
Random Change Anti-Method: In this approach, user randomly guesses where the problems may be and then changes things until it goes away.
Blame-Someone-Else Anti-Method: In this approach, user blames someone else and redirects the issue to another team.
Ad Hoc Checklist Method: It’s a common methodology where a user uses an ad hoc list built from recent experience.
Problem Statement: It defines the problem statement based on if there has been a performance issue; what was changed recently; who are being affected; etc.
Scientific Method: This approach is summarized as: Question -> Hypothesis -> Prediction -> Test -> Analysis.
Diagnostic Cycle: This is defined as hypothesis -> instrumentation -> data -> hypothesis.
Tools Method: This method lists available performance tools; gather metrics from each tool; and then interpret the metrics collected.
Utilization, Saturation, and Errors (USE) Method: This method focuses on system resources and checks utilization, saturation and errors for each resource.

RED Method: This approach checks request rate, errors, and duration for every service.
Workload Characteristics: This approach answers questions about Who is causing the load; why is the load being called; what are the load characteristics?
Drill-Down Analysis: This approach defines three-stage drill-down analysis methodology for system resource: Monitoring, Identification, Analysis.
Latency Analysis: This approach examines the time taken to complete an operation and then breaks it into smaller components, continuing to subdivide the components.
Method R: This methodology developed for Oracle database that focuses on finding the origin of latency.

Modeling

The chapter then defines analytical modeling of a system using following techniques:

Enterprise vs Cloud
Visual Identification uses graphs to identify patterns for linear scalability, contention, coherence (propagation of changes), knee-point (where performance stops scaling linearly), scalability ceiling.
Admdahl’s Law of Scalability describes content for the serial resource:

C(N) = N / (1 + a(N - 1))
where C(N) is relative capacity, N is scaling dimension such as number of CPU, and a is degree of seriality.

Universal Scalability Law is described as:

C(N) = N / (1 + a(N - 1) + bN(N - 1))
where b is the coherence parameter and when b = 0, it becomes Amdahl's Law

Queuing Theory describes Little’s Law as:

L = lambda * W
where L is average number of requests in the system, lambda is average arrival rate, and W is average request time.

Capacity Planning

This section describes Capacity Planning for examining how system will handle load and will scale as load scales. It searches for a resource that will become the bottleneck under load including hardware and software components. It then applies factor analysis to determine what factors to change to achieve the desired performance.

Statistics

This section reviews how to use statistics for analysis such as:

Quantifying Performance Gains: using observation-based and experimentation-based techniques to compare performance improvements.
Averages: including geometric-mean (nth root of multiplied values), harmonic-mean (count of values divided by sum of their reciprocals), average over time, decayed average (recent time weighed more).
Standard Deviation, Percentile, Median
Coefficient of Variations
Multimodal Distribution

Monitoring

Monitoring records performance statistics over time for comparison and identification using various time-based patterns such as hourly, daily, weekly, and yearly.

Visualization

This section examines various visualization graphs such as line-chart, scatter-plots, heat-maps, timeline-charts, and surface plot.

3. Operating Systems

This chapter examines operating system and kernel for system performance analysis and defines concepts such as:

Background

Kernel

Kernel is the core of operating system and though Linux and BSED have a monolithic kernel but other kernel models include micokernels, unikernels and hybrid kernels. In addition, new Linux versions include extended BPF for enabling secure kernel-mode applications.

Kernel and User Modes

The kernel runs in a kernel mode to access devices and execution of the privileged instructions. User applications run in a user mode where they can request privileged operations using system calls such as ioctl, mmap, brk, and futex.

Interrupts

An interrupt is a single to the processor that some event has occurred that needs processing, and interrupts the current execution of the processor and runs interrupt service routine to handle the event. The interrupts can be asynchronous for handling interrupt service requests (IRQ) from hardware devices or synchronous generated by software instructions such as traps, exceptions, and faults.

Clocks and Idle

In old kernel implementations, tick latency and tick overhead caused some performance issues but modern implementations have moved much functionality out of the clock routine to on-demand interrupts to create tickless kernel for improving power efficiency.

Processes

A process is an environment for executing a user-level program and consists of memory address space, file descriptors, thread stacks, and registers. A process contains one or more threads, where each thread has a stack, registers and an instruction pointer (PC). Processes are normally created using fork system call that wraps around clone and exec/execve system call.

Stack

A stack is a memory storage area for temporary data, organized as last-in, first-out (FIFO) list. It is used to store the return address when calling a function and passing parameters to function, which is also referred as a stack frame. The call path can be seen by examining saved return addresses across all the stack frames, which is called stack trace or stack back trace. While executing a system call, a process thread has two stacks: a user-level stack and a kernel-level stack.

Virtual Memory

Virtual memory is an abstraction of main memory, providing processes and the kernel with their own private view of main memory. It supports multitasking of threads and over-subscription of main memory. The kernel manages memory using process-swapping and paging schemes.

Schedulers

The scheduler schedules processes on processors by dividing CPU time among the active processes and threads. The scheduler tracks all threads in the read-to-run state in run priority-queues where process priority can be modified to improve performance of the workload. Workloads are categorized as either CPU-bound or I/O bound and scheduler may decrease the priority of CPU-bound processes to allow I/O-bound workloads to run sooner.

File Systems

File systems are an organization of data as files and directories. The virtual file system (VFS) abstracts file system types so that multiple file systems may coexist.

Kernels

This section discusses Unix-like kernel implementations with a focus on performance such as Unix, BSD, and Solaris. In the context of Linux, it describes systemd, which is a commonly used service manager to replace original UNIX init system and extended BPF that can be used for networking, observability, and security. BPF programs run in kernel model and are configured to run on events like USDT probes, kprobes, uprobes, and perf_events.

4. Observability Tools

This chapter identifies static performance and crises tools including their overhead like counters, profiling, and tracing.

Tools Coverage

This section describes static performance tools like sysctl, dmesg, lsblk, mdadm, ldd, tc, etc., and crisis tools like vmstat, ps, dmesg, lscpu, iostat, mpstat, pidstat, sar, and more.

Tools Type

The observability tool can categorized as a system-wide, per-process observability or counters/events based, e.g. top shows system-wide summary; ps, pmap are maintained per-process; and profilers/tracers are event-based tools. Kernel maintains various counters that are incremented when events occur such as network packets received, disk I/O, etc. Profiling collects a set of samples such as CPU usage at fixed-rate or based on untimed hardware events, e.g., perf, profile are system-wide profilers, and gprof and cachegrind are per-process profilers. Tracing instruments occurrence of an event, and can store event-based details for later analysis. Examples of system-wide tracing tools include tcpdump, biosnoop, execsnoop, perf, ftrace and bpftrace, and examples of per-process tracing tools include strace and gdb. Monitoring records statistics continuously for later analysis, e.g. sar, prometheus, collectd are common tools for monitoring.

Observability Sources

The main sources of system performance include /proc and /sys. The /proc is a file system for kernel statistics and is created dynamically by the kernel. For example, ls -F /proc/123 lists per-process statistics for process with PID 123 like limits, maps, sched, smaps, stat, status, cgroup and task. Similarly, ls -Fd /proc/[a-z]* lists system-wide statistics like cpuinfo, diskstats, loadavg, meminfo, schedstats, and zoneinfo. The /sys was originally designed for device-driver statistics but has been extended for other types of statistics, e.g. find /sys/devices/system/cpu/cpu0 -type f provides information about CPU caches. Tracepoints are Linux kernel event source based on static instrumentation and provide insight into kernel behavior. For example, perf list tracepoint lists available tracepoints and perf trace -e block:block_rq_issue traces events. Tracing tools can also use tracepoints, e.g., strace -e openat ~/iosnoop and strace -e perf_event_open ~/biosnoop. A kernel event source based on dynamic instrumentation includes kprobes that can trace entry to functions and instructions, e.g. bpftrace -e 'kprobe:do_nanosleep { printf("sleep %s\n", comm); }'. A user-space event-source for dynamic instrumentation includes uprobes, e.g., bpftrace -l 'uprob:/bin/bash:*'. User-level statically-defined tracing (UsDT) is the user-space version of tracepoint and some libraries/apps have added USDt probes, e.g., bpftrace -lv 'usdt:/openjdk/libjvm.so:*'. Hardware counters (PMC) are used for observing activity by devices, e.g. perf stat gzip words instruments the architectural PMCs.

Sar

Sar is a key monitoring tool that is provided via the sysstat package, e.g., sar -u -n TCP,ETCP reports CPU and TCP statistics.

5. Applications

This chapter describes performance tuning objectives, application basics, fundamentals for application performance, and strategies for application performance analysis.

Application Basics

This section defines performance goals including lowering latency, increasing throughput, improving resource utilization, and lowering computing costs. Some companies use a target application performance index (ApDex) as an objective and as a metric to monitor:

Apdex = (satisfactory + 0.5 x tolerable + 0 x frustrating) / total-events

Application Performance Techniques

This section describes common techniques for improving application performance such as increasing I/O size to improve throughput, caching results of commonly performed operations, using ring buffer for continuous transfer between components, and using event notifications instead of polling. This section describes concurrency for loading multiple runnable programs and their execution that may overlap, and recommends parallelism via multiple processes or threads to take advantage of multiprocessor systems. These multiprocess or multithreaded applications use CPU scheduler with the cost of context-switch overhead. Alternatively, use-mode applications may implement their own scheduling mechanisms like fibers (lightweight threads), co-routines (more lightweight than fiber, e.g. Goroutine in Golang), and event-based concurrency (as in Node.js). The common models of user-mode multithreaded programming are: using service thread-pool, CPU thread-pool, and staged event-driven architecture (SEDA). In order to protect integrity of shared memory space when accessing from multiple threads, applications can use mutex, spin locks, RW locks, and semaphores. The implementation of these synchronization primitives may use fastpath (using cmpxchg to set owner), midpath (optimistic spinning), slowpath (blocks and deschedules thread), or read-copy-update (RCU) mechanisms based on the concurrency use-cases. In order to avoid the cost of creation and destruction of mutex-locks, the implementations may use hashtable to store a set of mutex locks instead of using a global mutex lock for all data structures or a mutex lock for every data structure. Further, non-blocking allows issuing I/O operations asynchronously without blocking the thread using O_ASYNC flag in open, io_submit, sendfile and io_uring_enter.

Programming Languages

This section describes compiled languages, compiler optimization flags, interpreted languages, virtual machines, and garbage collection.

Methodology

This section describes methodologies for application analysis and tuning using:

CPU profiling and visualizing via CPU flame graphs.
off-CPU analysis using sampling, scheduler tracing, and application instrumentation, which may be difficult to interpret due to wait time in the flame graphs so zooming or kernel-filtering may be required.
Syscall analysis can be instrumented to study resource based performance issues where the target for syscall analysis include new process tracing, I/O profiling, and kernel time analysis.
USE method
Thread state analysis like user, kernel, runnable, swapping, disk I/O, net I/O, sleeping, lock, idle.
Lock analysis
Static performance tuning
Distributed tracing

Observability Tools

This section introduces application performance observability tools:

perf is standard Linux profiler with many uses such as:
- CPU Profiling: perf record -F 49 -a -g -- sleep 30 && per script --header > out.stack
- CPU Flagme Graphs: ./stackcollapse-perf.pl < out.stacks | ./flamgraph.ol --hash > out.svg
- Syscall Tracing: perf trace -p $(pgrep mysqld)
- Kernel Time Analysis: perf trace -s -p $(pgrep mysqld)
porfile is timer-based CPU profiler from BCC, e.g.,
- profile -F 49 10
offcputime and bpftrace to summarize time spent by thrads blocked and off-CPU, e.g.,
- offcputime 5
strace is the Linux system call tracer
- strace -ttt -T -p 123
execsnoop traces new process execution system-wide.
syscount to count system call system wide.
bpftrace is a BPF-based tracer for high-level programming languages, e.g.,
- bpftrace -e 't:syscalls:sys_enter_kill { time("%H:%M:%S "); }'

6. CPU

This chapter provides basis for CPU analysis such as:

Models

This section describes CPU architecture and memory caches like CPU registers, L1, L2 and L3. It then describes run-queue that queue software threads that are ready to run and time spent waiting on CPU run-queue is called run-queue latency or dispatcher-queue latency.

Concepts

This section describes concepts regarding CPU performance including:

Clock Rate: Each CPU instruction may take one or more cycles of the clock to execute.
Instructions: CPU execute instructions chosen from their instruction set.
Instruction Pipeline: This allows multiple instructions in parallel by executing different components of different instructions at the same time. Modern processors may implement branch prediction to perform out-of-order execution of the pipeline.
Instruction Width: Superscalar CPU architecture allows more instructions can make progress with each clock cycle based on the width of instruction.
SMT: Simultaneous multithreading makes use of a superscalar architecture and hardware multi-threading support to improve parallelism.
IPC, CP: Instructions per cycle ((IPC) describe how CPU is spending its clock cycles.
Utilization: CPU utilization is measured by the time a CPU instance is busy performing work during an interval.
User Time/Kernel Time: The CPU time spent executing user-level software is called user time, and kernel-time software is kernel time.
Saturation: A CPU at 100% utilization is saturated, and threads will encounter scheduler lateny as they wait to run on CPU.
Priority Inversion: It occurs when a lower-priority threshold holds a resource and blocks a high priority thread from running.
Multiprocess, Multithreading: Multithreading is generally considered superior.

Architecture

This section describes CPU architecture and implementation:

Hardware

CPU hardware include processor and its subsystems:

Processor: The processor components include P-cache (prefetch-cache), W-cache (write-cache), Clock, Timestamp counter, Microcode ROM, Temperature sensor, and network interfaces.
P-States and C-States: The advanced configuration and power interface (ACPI) defines P-states, which provides different levels of performance during execution, and C-states, which provides different idle states for when execution is halted, saving power.
CPU caches: This include levels of caches like level-1 instruction cache, level-1 data cache, translation lookaside buffer (TLB), level-2 cache, and level-3 cache. Multiple levels of cache are used to deliver the optimum configuration of size and latency.
Associativity: It describes constraint for locating new entries like full-associative (e.g. LRU), direct-mapped where each entry has only one valid location in cache, set-associative where a subset of the cache is identified by mapping, e.g., four-way associative maps an address to four possible location.
Cache Line: Cache line size is a range of bytes that are stored and transferred as a unit.
Cache Coherence: Cache coherence ensures that CPUs are always accessing the correct state of memory.
MMU: The memory management unit (MMU) is responsible for virtual-to-physical address translation.
Hardware Counters (PMC): PMCs are processor registers implemented in hardware and include CPU cycles, CPU instructions, Level 1, 2, 3 cache accesses, floating-point unit, memory I/O, and resource I/O.
GPU: GPU support graphical displays.
Software: Kernel software include scheduler that performs time-sharing, preemption, and load balancing. The scheduler uses scheduling classes to manage the behavior of runnable threads like priorities and scheduling policies. The scheduling classes for Linux kernels include RT (fixed and high-priorities for real-time workloads), O(1) for reduced latency, CFS (Completely fair scheduling), Idle, and Deadline. Scheduler policies include RR (round-robin), FIFO, NORMAL, BATCH, IDLE, and DEADLINE.

Methodology

This section describes various methodologies for CPU analysis and tuning such as:

Tools Method: iterate over available tools like uptime, vmstat, mpstat, perf/profile, showboost/turboost, and dmesg.
USE Method: It checks for utilization, saturation, and errors for each CPU.
Workload Charcterization like CPU load average, user-time to system-time ratio, syscall rate, voluntary context switch rate, and interrupt rate.
Profiling: CPU profiler can be performed by time-based sampling or function tracing.
Cycle Analysis: Using performance monitor counter ((PMC) to understand CPU utilization at the cycle level.
Performance Monitoring: identifies active issues and patterns over time using metrics for CPU like utilization and saturation.
Static Performance Tuning
Priority Tuning
CPU Binding

Observability Tools

This section introduces CPU performance observability tools such as:

uptime
load average – exponentially damped moving average for load including current resource usage plus queued requests (saturation).
pressure stall information (PSI)
vmstat, e.g. vmstat 1
mpstat, e.g. mpstat -P ALL 1
sar
ps
top
pidstat
time, ptime
turbostat
showboost
pmcash
tlbstat
perf, e.g., perf record -F 99 command, perf stat gzip ubuntu.iso, and perf stat -a -- sleep 10
profile
cpudist, e.g., cpudist 10 1
runqlat, e.g., runqlat 10 1
runqlen, e.g., runqlen 10 1
softirqs, e.g., softirqs 10 1
hardirqs, e.g., hardirqs 10 1
bpftrace, e.g., bpftrace -l 'tracepoint:sched:*'

Visualization

This section introduces CPU utilization heat maps, CPU subsecond-offset heat maps, flame graphs, and FlameScope.

Tuning

The tuning may use scheduling priority, power stats, and CPU binding.

7. Memory

This chapter provides basis for memory analysis including background information on key concepts and architecture of hardware and software memory.

Concepts

Virtual Memory

Virtual memory is an abstraction that provides each process its own private address space. The process address space is mapped by the virtual memory subsystem to main memory and the physical swap device.

Paging

Paging is the movement of pages in and out of main memory. File system paging (good paging) is caused by the reading and writing of pages in memory-mapped files. Anonymous paging (bad paging) involves data that is private to process: the process heap and stacks.

Demand Paging

Demand paging map pages of virtual memory to physical memory on demand and defers the CPU overhead of creating the mapping until they are needed.

Utilization and Saturation

If demands for the memory exceeds the amount of main memory, main memory becomes saturated and operating system may employ paging or OOM killer to free it.

Architecture

This section introduces memory architecture:

Hardware Main Memory

The common type of the main memory is dynamic random-access memory (DRAM) and column address strobe (CAS) latency for DDR4 is around 10-20ns. The main memory architecture can be uniform memory access (UMA) or non-uniform memory access (NUMA). The main memory may use a shared system-bus to connect a single or multiprocessors, directly attached memory, or interconnected memory bus. The MMU (memory management unit) translates virtual addresses to physical addresses for each page and offset within a page. The MMU uses a TLB (translation lookaside buffer) as a first-level cache for addresses in the page tables.

Software

The kernel tracks free memory in free list of pages that are available for immediate allocation. The kernel may use swapping, reap any memory that can be freed, or use OOM killer to free memory when memory is low.

Process Virtual Address Space

The process virtual address space is a range of virtual pages that are mapped to physical pages and addresses are split into segments like executable text, executable data, heap, and stack. There are a variety of user- and kernel-level allocators for memory with simple APIs (malloc/free), effcient memory usage, performance, and observability.

Methodology

This section describes various methodlogies for memory analysis:

Tools Method: involves checking page scanning, pressure stall information (PSI), swapping, vmstat, OOM killer, and perf.
USE Method: utilization of memory, the degree of page scanning, swapping, OOM killer, and hardware errors.
Characterizing usage
Performance monitoring
Leak detection
Static performance tuning

Observability Tools

This section includes memory observability tools including:

vmstat
PSI, e.g., cat /proc/pressure/memory
swapon
sar
slabtop, e.g., slabtop -sc
numastat
ps
pmap, e.g. pmap -x 123
perf
drsnoop
wss
bpftrace

Tuning

This section describes tunable parameters for Linux kernels such as:

vm.dirty_background_bytes
vm.dirty_ratio
kernel.numa_balancing

8. File Systems

This chapter provides basic for file system analysis:

Models

File System Interfaces: File system interfaces include read, write, open, and more.
File System Cache: The file system cache may cache reads or buffer writes.
Second-Level Cache: It can be any memory type like level-1, level2, RAM, and disk.

Concepts

File System Latency: primary metric of file system performance for time spent in the file system and disk I/O subsystem.
Cache: The file system will use main memory as a cache to improve performance.
Random vs Sequential I/O: A series of logical file system I/O can be either random or sequential based on the file offset of each I/O.
Prefetch/Read-Ahead: Prefetch detects sequential read workload and issue disk reads before the application request it.
Write-Back Caching: It marks write completed after transfering to main memory and writes to disk asynchronously.
Synchronous Writes: using O_SYNC, O_DSYNC or O_RSYNC flags.
Raw and Direct I/O
Non-Blocking I/O
Memory-Mapped Files
Metadata including information about logical and physical that is read and written to the file system.
Logical vs Physical I/O
Access Timestamps
Capacity

Architecture

This section introduces generic and specific file system architecture:

File System I/O Stack

VFS (virtual file system) common interface for different file system types.
File System Caches
- Buffer Cache
- Page Cache
- Dentry Cache
- Inode Cache
File System Features
- Block (fixed-size) vs Extent (pre-allocated contiguous space)
- Journaling
- Copy-On-Write
- Scrubbing
File System Types
- FFS – Berkley fast file system
- ext3/ext4
- XFS
- ZFS
Volume and Pools

Methodology

This section describes various methodologies for file system analysis and tuning:

Disk Analysis
Latency Analysis
- Transaction Cost
Workload Characterization
- cache hit ratio
- cache capacity and utilization
- distribution of I/O arrival times
- errors
Performance monitoring (operation rate and latency)
Static Performance Tuning
Cache Tuning
Workload Separation
Micro-Benchmaking
- Operation types (read/write)
- I/O size
- File offset pattern
- Write type
- Working set size
- Concurency
- Memory mapping
- Cache state
- Tuning

Observability Tools

mount
free
vmstat
sar
slabtop, e.g., slabtop -a
strace, e.g., strace -ttT -p 123
fatrace
opensnoop, e.g., opensnoop -T
filetop
cachestat, e.g., cachestat -T 1
bpftrace

9. Disks

This chapter provides basis for disk I/O analysis. The parts are as follows:

Models

Simple Disk: includes an on-disk queue for I/O requests
Caching Disk: on-disk cache
Controller: HBA (host-bus adapter) bridges CPU I/O transport with the storage transport and attached disk devices.

Concepts

Measuring Time: I/O request time = I/O wait-time + I/O service-time
- disk service time = utilization / IOPS
Time Scales
Caching
Random vs Sequential I/O
Read/Write Ratio
I/O size
Utilization
Saturation
I/O Wait
Synchronous vs Asynchronous
Disk vs Application I/O

Architecture

Disk Types
- Magnetic Rotational
  - max throughput = max sector per track x sector-size x rpm / 60 s
  - Short-Stroking
  - Sector Zoning
  - On-Disk Cache
- Solid-State Drives
  - Flash Memory
  - Persistent Memory
Interfaces
- SCSI
- SAS
- SATA
- NVMe
Storage Type
- Disk Devices
- RAID
Operating System Disk I/O Stack

Methodology

Tools Method
- iostat
- iotop/biotop
- biolatency
- biosnoop
USE Method
Performance Monitoring
Workload Characterization
- I/O rate
- I/O throughput
- I/O size
- Read/write ratio
- Random vs sequential
Latency Analysis
Static Performance Tuning
Cache Tuning
Micro-Benchmarking
Scaling

Observability Tools

iostat
pressure stall information (PSI)
perf
biolatency, e.g., biolatency 10 1
biosnoop
biotop
ioping

10. Network

This chapter introduces network analysis. The parts are as follows:

Models

Network Interface

Controller
Protocol Stack
- TCP/IP
- OSI Model

Concepts

Network and Routing
Protocols
Encapsulation
Packet Size
Latency
- Connection Latency
- First-Byte Latency
- Round-Trip Time
Buffering
Connection Backlog
Congestion Avoidance
Utilization
Local Connection

Architecture

Protocols
IP
TCP
- Sliding Window
- Congestion avoidance
- TCP SYN cookies
- 3-way Handshake
- Duplicate Ack Detection
- Congestion Controls
- Nagle algorithm
- Delayed Acks
UDP
- QUIC and HTTP/3
Hardware
- Interfaces
- Controller
- Switches and Routers
- Firewalls
Software
- Network Stack

Linux Stack
- TCP Connection queues
- TCP Buffering
- Segmentation Offload: GSO and TSO
- Network Device Drivers
- CPU Scaling
- Kernel Bypass

Methodology

Tools Method
- netstat -s
- ip -s link
- ss -tiepm
- nicstat
- tcplife
- tcptop
- tcpdump
USE Method
Workload Characterization
- Network interface throughput
- Network interface IOPS
- TCP connection rate
Latency Analysis
Performance Monitoring
- Throughput
- Connections
- Errors
- TCP retransmits
- TCP out-of-order pack
Packet Sniffing
- tcpdump -ni eth4
TCP Analysis
Static Performance Tuning
Resource Controls
Micro-Benchmarking

Observability Tools

ss, e.g., ss -tiepm
strace, e.g., strace -e sendmesg,recvmsg ss -t
ip, e.g., ip -s link
ifconfig
netstat
sar
nicstat
ethtool, e.g., ethtool -S eth0
tcplife, tcptop
tcpretrans
bpftrace
tcpdump
wireshark
pathchar
iperf
netperf
tc

Tuning

sysctl -a | grep tcp

11. Cloud Computing

This chapter introduces cloud performance analysis with following parts:

Background

Instance Types: m5 (general-purpose), c5 (compute optimized)
Scalable Architecture: horizontal scalability with load balancers, web servers, application servers, and databases.
Capacity Planning: Dynamic sizing (auto-scaling) using auto-scaling-group and and Scalability testing.
Storage: File store, block store, object store
Multitenancy
Orchestration (Kubernetes)

Hardware Virtualization

Type 1: execute directly on the processors using native hypervisor or bare-metal hypervisor (e.g., Xen)
Type 2: execute within a host OS and hypervisor is scheduled by the host kernel
Implementation: Xen, Hyper-V, KVM, Nitro
Overhead:
- CPU overhead (binary translation, paravirtualization, hardware assisted)
- Memory Mapping
- Memory Size
- I/O
- MultiTenant Contention
- Resource Controls
Resource Controls
- CPUs – Borrowed virtual time, Simple earliest deadline, Credit based schedulers
- CPU Caches
- Memory Capacity
- File System Capacity
- Device I/O
Observability
- xentop
- perf kvm stat live
- bpftrace -lv t:kvm:kvm_exit
- mpstat -P ALL 1

OS Virtualization

Implementation: Linux supports namespaces and cgroups that are used to create containers. Kubernetes uses following architecture for Pods, Kube Proxy and CNI.

Namespaces
- lsns
Control Groups (cgroups) limit the usage of resources
Overhead – CPU, Memory Mapping, Memory Sie, I/O, and Multi-Tenant Contention
Resource Controls – throttle access to resources so they can be shared more fairly
- CPU
- Shares and Bandwidth
- CPU Cache
- Memory Capacity
- Swap Capacity
- File System Capacity
- File System Cache
- Disk I/O
- Network I/O
Observability
- from Host
  - kubectl get pod
  - docker ps
  - kubectl top nodes
  - kubectl top pods
  - docke stats
  - cgroup stats (cpuacct.usage and cpuacct.usage_percpu)
  - system-cgtop
  - nsenter -t 123 -m -p top
  - Resource Controls (throttled time, non-voluntary context switches, idle CPU, busy, all other tenants idle)
- from guest (container)
  - iostat -sxz 1
Lightweight Virtualization

Lightweight hypervisor based on process virtualization (Amazon Firecracker)
- Implementation – Intel Clear, Kata, Google gVisor, Amazon Firecracker
Overhead
Resource Controls
Observability
- From Host
- From Guest
  - mpstat -P ALL 1

12. Benchmarking

This chapter discusses benchmarks and provides advice with methodologies. The parts of this chapter include:

Background

Reasons
- System design
- Proof of concept
- Tuning
- Development
- Capacity planning
- Troubleshooting
- Marketing
Effective Bencharmking
- Repeatable
- Observable
- Portable
- Easily presented
- Realistic
- Runnable
- Bencharmk Analysis
Bencharm Failures
- Causal Bencharmking – you benchmark A, but measure B, and conclude you measured C, e.g. disk vs file system (buffering/caching may affect measurements).
- Blind Faith
- Numbers without Analysis – include description of the benchmark and analysis.
- Complex Benchmark Tools
- Testing the wrong thing
- Ignoring the Environment (not tuning same as production)
- Ignoring Errors
- Ignoring Variance
- Ignoring Perturbations
- Changing Multiple Factors
- Friendly Fire
Benchmarking Types
- Micro-Benchmarking
- Simulation – simulate customer application workload (macro-bencharmking)
- Replay
- Industry Standards – TPC, SPEC

Methodology

This section describes methologies for performing becharmking:

Passive Bencharmking (anti-methodology)
- pick a BM tool
- run with a variety of options
- Make a slide of results and share it with management
- Problems
  - Invalid due to software bugs
  - Limited by benchmark software (e.g., single thread)
  - Limited by a component that is unrelated to the benchmark (congested network)
  - Limited by configuration
  - Subject to perturbation
  - Benchmarking the wrong the thing entirely
Active Benchmarking
- Analyze performance while benchmarking is running
- bonie++
- iostat -sxz 1
CPU Profiling
USE Method
Workload Characterization
Custom Benchmarks
Ramping Load
Statistical Analysis
- Selection of benchmark tool, its configuration
- Execution of the benchmark
- Interpretation of the data
Benchmarking Checklist
- Why not double
- Did it beak limit
- Did it error
- Did it reproduce
- Does it matter
- Did it even happen

13. perf

This chapter introduces perf tool:

Subcommands Overview
- perf record -F 99 -a -- sleep 30

perf Events
- perf list
Hardware Events (PMCs)
- Frequency Sampling
- perf record -vve cycles -a sleep 1
Software Events
- perf record -vve context-switches -a -- sleep
Tracepoint Events
- perf record -e block:block_rq_issue -a sleep 10; perf script
Probe Events
- kprobes, e.g., perf prob --add do_nanosleep
- uprobes, e.g., perf probe -x /lib.so.6 --add fopen
- USDT
perf stat
- Interval Statistics
- Per-CPU Balance
- Event Filters
- Shadow Statistics

perf record
- CPU Profiling, e.g., perf record -F 99 -a -g -- sleep 30
- Stack Walking
perf report
- TUI
- STDIO
perf script
- Flame graphs
perf trac

12. Ftrace

This chapter introduces Ftrace tool. The sections are:

Capabilities
tracefs
- tracefs contents, e.g., ls -F /sys/kernel/debug/tracing
Ftrace Function Profiler
Ftrace Function Tracing
Tracepoints
- Filter
- Trigger
kprobes
- Event Tracing
- Argument and Return Values
- Filters and Triggers
uprobes
- Event Tracing
- Argument and Return Values
Ftrace function_graph
Ftrace hwlat
Ftrace Hist Triggers
perf ftrace

15. BPF

This chapter introduces BPF tool. The sections are:

BCC – BPF COmpiler Collection, e.g., biolatency.py -mF
bpftrace – open-source tracer bult upon BPF and BC
Programming

16. Case Study

This chapter describes the story of a real-world performance issue.

Problem Statement – Java application in AWS EC2 Cloud
Analysis Strategy
- Checklist
- USE method
Statistics
- uptime
- mpstat 10
Configuration
- cat /proc/cpuinfo
PMCs
- ./pmarch -p 123 10
Software Events
- perf stat -e cs -a -I 1000
Tracing
- cpudist -p 123 10 1
Conclusion
- No container neighbors
- LLC size and workload diffeence
- CPU difference

Comments (0)

May 21, 2023

Heuristics from “Code That Fits in Your Head”

Filed under: Methodologies,Technology,Uncategorized — admin @ 5:00 pm

The code maintenance and readability are important aspects of writing software systems and the “Code That Fits in Your Head” comes with a lot of practical advice for writing maintainable code. Following are a few important heuristics from the book:

1. Art or Science

In the first chapter, the author compares software development with other fields such as Civil engineering that deals with design, construction, and maintenance of components. Though, software development has these phases among others but the design and construction phases in it are intimately connected and requires continuous iteration. Another metaphor discussed in the book is thinking software development as a living organism like garden, which makes more sense as like pruning weeds in garden, you have to refactor the code base and manage technical debt. Another metaphor described in the book is software craftsmanship and software developer may progress from apprentice, journeyman to master. Though, these perspectives help but software doesn’t quite fit the art metaphor and author suggests heuristics and guidelines for programming. The author introduces software engineering that allows a structured framework for development activities.

2. Checklists

A lot of professions such as airplane pilots and doctors follow a checklist for accomplishing a complex task. You may use similar checklist for setting up a new code-base such as using Git, automating the build, enabling all compiler error messages, using linters, static analysis, etc. Though, software engineering is more than following a checklist but these measures help make small improvements.

3. Tackling Complexity

This chapter defines sustainability and quality as the purpose for the book as the software may exists for decades and it needs to sustain its organization. The software exists to provide a value, though in some cases the value may not be immediate. This means at times, worse technology succeeds because it provides faster path to the value and companies which are too focus on perfection can run out of business. Richard Gabriel coined the aphorism that worse is better. The sustainability chooses middle ground by moving in right direction with checklists and balanced software development practices. The author compares computer with human brain and though this comparison is not fair and working memory of humans is much smaller that can hold from four to seven pieces of information. This number is important when writing a code as you spend more time reading the code and a code with large number of variables or conditional logic can make it harder to understand. The author refers to the work of Daniel Kahneman who suggested model of thoughts comprising two systems: System 1 and System 2. When a programmer is in the zone or in a flow, the system 1 always active and try to understand the code. This means that writing modular code with a fewer dependencies, variables and decisions is easier to understand and maintain. The human brain can deal with limited memory and if the code handles more than seven things at once then it will lead to the complexity.

4. Vertical Slice and Walking Skeleton

This chapter recommends starting and deploying a vertical slice of the application to get to the working software. A vertical slice may consists of multiple layers but it gives an early feedback and is a working software. A number of software development methodologies such as Test-driven development, Behavioral-driven development, Domain-driven design, Type-driven development and Property-driven development help building fine-grained implementations with tests. For example, if you don’t have tests then you can use characterization test to describe the behavior of existing software. The tests generally follow Arranage-Act-Assert phases where the arrange phase prepares the test, the act phase invokes the operation under test and the assert phase verifies the actual outcome. The documentation can further explain why decisions in the code were made. The walking skeleton helps vertical slice by using acceptance-test-driven development or outside-in-test-driven development. For example, you can pick a simplest feature to implement that aims for the happy path to demonstrate that the system has a specific capability. The unit-tests will test this feature by using Fake Object, data-transfer-object (DTO) and interfaces (e.g. RepositoryInterface). The dependencies are injected into tests with this mock behavior. The real objects that are difficult tests can use a Humble Object pattern and drain the object of branching logic. Making small improvements that are continuously delivered also keep stakeholders updated so that they know when you will be done.

5. Encapsulation

The encapsulation hides details by defining a contract that describes the valid interactions between objects and callers. The parameterized tests can capture the desired behavior and assert the invariants. The incremental changes can be added using test-driven development that uses red-green-refactor where you first write a failing test, then make the test pass and then refactor to improve the code. When using a contract to capture the interactions, you can use Postel’s law to build resilient systems, i.e.,

Be conservative in what you send, be liberal in what you accept.

The encapsulation guarantees that an object is always valid, e.g. you can use a constructor to validate all invariants including pre-conditions and post-conditions.

6. Triangulation

As the working memory for humans is very small, you have to decompose and compartmentalize the code structure into smaller chunks that can be easily understood. The author describes a devil’s advocate technique for validating behavior in the unit tests where you try to pass the tests with incomplete implementation, which tells you that you need more test cases. This process can be treated as kind of triangulation:

As the tests get more specific, the code gets more generic

7. Decomposition

The code rot occurs because no one pays attention to the overall quality when making small changes. You can use metrics to track gradual decay such as cyclomatic complexity should be below seven. In order to improve the code readability, the author suggests using 80/24 rule where you limit a method size to be no more than 24 lines and width of each line to be no more than 80 characters. The author also suggests hex flower rule:

No more than seven things should be going on in a single piece of code.

The author defines abstraction to capture essence of an object, i.e.,

Abstraction is the elimination of the irrelevant and the amplification of the essential.

Another factor that influences decomposition is cohesion so that code that works on the same data structure or all of its attributes is defined in the same module or class. The author cautions against the feature envy to decrease the complexity and you may need to refactor the code to another method or class. The author refers to a technique “parse, don’t validate” when validating an object so that the validate method takes less-structured input and produces more-structured output. Next, author describes fractal architecture where a large system is decomposed into smaller chunks and each chunk hides details but can be zoomed in to see the structure. The fractal architecture helps organize the code so that lower-level details are captured in a single abstract chunk and can easily fit in your brain.

8. API Design

This chapter describes principles of API design such as affordance, which uses encapsulation to preserve the invariants of objects involved in the API. The affordance allows a caller to invoke an API only when preconditions are met. The author strengthen the affordance with a poka-yoke (mistake proof) analogy, which means a good interface design should be hard to misuse. Other techniques in the chapter includes: write code for the readers; favor well-named code over comments; and X out names. The X out names replaces API name with xxx and sees if a reader can guess what the API does. For example, you may identify APIs for command-query separation where a method structure like void xxx() can be considered as command with a side effect. In order to communicate the intent of an API, the author describes a hierarchy of communication such as using API’s distinct types, helpful names, good comments, automated tests, helpful commit messages and good documentation.

9. Teamwork

In this chapter, the author provides tips for teamwork and communication with other team mates such as writing good Git commit messages using 50/72 rule where you first write a summary no wider than 50 characters, followed by a blank line and then detailed text with no wider than 72 characters. Other techniques include Continuous Integration that generally use trunk or main branch for all commits and developers make small changes optionally with feature-flags that are frequently merged. The developers are encouraged to make small commits and the code ownership is collective to decrease the bus factor. The author refers to pair programming and mob programming for collaboration within the team. In order to facilitate the collaboration, the author suggests reducing code review latency and rejecting any large change set. The reviewers should be asking whether they can maintain the code, is the code intent clear and could it be further simplified, etc. You can also pull down the code and test it locally to further gain the insight.

10. Augmenting Code

This chapter focuses on refactoring existing code for adding new functionality, enhancing existing behavior and bug fixes. The author suggests using feature-flags when deploying incomplete code. The author describes the strangler pattern for refactoring with incremental changes and suggests:

For any significant change, don’t make it in-place; make it side-by-side.

The strangler pattern can be applied at method-level where you may add a new method instead of making in-place change to an existing method and then remove the original method. Similarly, you can use class-level strangler to introduce new data structure and then remove old references. The author suggests using semantic versioning so that you can support backward compatible or breaking changes.

11. Editing Unit Tests

Though, with an automated test suite, you can refactor production code safely but there is no safety net when making changes to the test code. You can add additional tests, supplement new assertions to existing tests or change unit tests to parametersized tests without affecting existing behavior. Though, some programmers follow a single assertion per test and consider multiple assertions an Assertion Roulette but author suggests strengthening the postconditions in unit tests with additional assertions, which is somewhat similar to the Liskov Substitution Principle that says that subtypes may weaken precondition and strengthen postconditions. The author suggests separating refactoring of test and production code and use IDE’s supported refactoring tools such as rename, extract or move method when possible.

12. Troubleshooting

When troubleshooting, you first have to understand what’s going on. This chapter suggests using scientific method to make a hypothesis, performing the experiment and comparing the outcome to prediction. The author also suggests simplifying and removing the code to check if a problem goes away. Other ways to simplify the code include composing an object graph in code instead of using complex dependency injection; using pure functions instead of using mock objects; merging often instead of using complex diff tools; learning SQL instead of using convoluted object-relational mapping, etc. Another powerful technique for troubleshooting is rubber ducking where you try to explain the problem and gain a new insight in the process. In order to build quality, you should aim to reduce defects to zero. The tests also help with troubleshooting by writing an automated test to reproduce defects before fixing so that they serve as a regression test. The author cautions against slow tests and non-deterministic defects due race conditions. Finally, the author suggests using bisection that uses a binary search for finding the root cause where you reproduce the defect in half of the code and continue until you find the problem. You can also use bisection feature of Git to find the commit that introduced the defect.

13. Separation of Concerns

The author describes Kent Beck’s aphorism:

Things that change at the same rate belong together. Things that change at different rates belong apart.

The principle of separation of concerns can be used for decomposing working software into smaller parts, which can be decomposed further with nested composition. The author suggests using command query separation principle to keep side effects separated from the query operations. Object-oriented composition tends to focus on composing side effects together such as Composite design pattern, which lead to complex code. The author describes Sequential Composition that chains methods together and Referential Transparency to define a deterministic method without side effects. Next, the author describes cross cutting concerns such as logging, performance monitoring, auditing, metering, instrumentation, caching, fault tolerance, and security. The author finally describes Decorator pattern to enhance functionality, e.g., you can add logging to existing code without changing it and log actions from impure functions.

14. Rhythm

This chapter describes daily and recurring practices that software development teams follow such as daily stand-ups. The personal rhythm includes time-boxing or using Pomodoro technique; taking a break; using time deliberately; and touch type. The team rhythm includes updating dependencies regularly, scheduling other things such as checking certificates. The author describes Conway’s law:

Any organization that design a system […] will inevitably produce a design whose structure is a copy of the organization’s communication structure.

You can use this law to organize the work that impacts the code base.

15. The Usual Suspects

This chapter covers usual suspects of software engineering: architecture, algorithms, performance, security and other approaches. For example, performance is often a critical aspect but premature optimization can be wasteful. Instead correctness, an effort to reduce complexity and defects should be priority. In order to implement security, the author suggests STRIDE threat modelling, which includes Spoofing, Tempering, Repudiation, Information disclosure, Denial of service and Elevation of privilege. Other techniques include property-based testing and Behavioral code analysis can be used to extract information from Git to identify patterns and problems.

16. Tour

In this last chapter, the author shows tips on understanding an unfamiliar code by navigating to the main method and finding the way around. You can check if the application uses broader patterns such as Fractal architecture, Model-View-Controller and understands authentication, authorization, routing, etc. The author provides a few suggestions about code structure and file organization such as putting files in one directory though it’s a contestable advice. The author refers to the Hex flower and fractal architecture where you can zoom in to see more details. When using a monolithic architecture, the entire production code compiles to a single executable file that makes it harder to reuse parts of the code in new ways. Another drawback of monolithic architecture is that dependencies can be hard to manage and abstraction can be coupled with implementation, which violates the Dependency Inversion Principle. Further in order to prevent cyclic dependencies, you will need to detect and prevent Acyclic Dependency Principle. Finally, you can use test suite to learn about the system.

Summary

The book is full of practical advice on writing maintainable code such as:

50/72 Rule for Git commit messages
80/24 Rule for writing small blocks of code
Tests based on Arrange-Act-Assert and Red-Green Refactor
Bisection for troubleshooting
Checklists for a new codebase
Command Query Separation
Cyclomatic Complexity and Counting the Variables
Decorators for cross-cutting concerns
Devil’s advocate for improving assertions
Feature flags
Functional core and imperative shell
Hierarchy of communication
Parse, don’t validate
Postel’s law to maintain invariants
Regularly update dependencies
Reproduce defects as Tests
Review code
Semantic Versioning
Separate refactoring of test and production code
Strangler pattern
Threat modeling using STRIDE
Transformation priority premise to make small changes and keeping the code in working condition
X-driven development by using unit-tests, static code analysis, etc.
X out of Names

These heuristics help make the software development sustainable so that the team can make incremental changes to the code while maintaining high quality.

Comments (0)

December 20, 2022

Property-based and Generative testing for Microservices

Filed under: REST,Technology,Testing — Tags: distributed — admin @ 1:26 pm

The software development cycle for microservices generally include unit testing during the development where mock implementation for the dependent services are injected with the desired behavior to test various test-scenarios and failure conditions. However, the development teams often use real dependent services for integration testing of a microservice in a local environment. This poses a considerable challenge as each dependent service may be keeping its own state that makes it harder to reliably validate the regression behavior or simulate certain error response. Further, as the number of request parameters to the service or downstream services grow, the combinatorial explosion for test cases become unmanageable. This is where property-based testing offers a relief as it allows testing against automatically generated input fuzz-data, which is why this form of testing is also referred as a generative testing. A generator defines a function that generate random data based on type of input and constraints on the range of input values. The property-based test driver then iteratively calls the system under test to validate the result and assert the desired behavior, e.g.

def pre_condition_test_input_param(kind):
  ### assert pre-condition based on type of parameter and range of input values it may take

def generate_test_input_param(kind):
  ### generate data meeting pre-condition for the type
    
def generate_test_input_params(kinds):
  return [generate_test_input_param(kind) for kind in kinds]  
  
for i in range(max_attempts):
  [a, b, c, ...] = generate_test_input_params(type1, type2, type3, ...)
  output = function_under_test(a, b, c, ...)
  assert property1(output)
  assert property2(output)
  ...

In above example, the input parameters are randomly generated based on a precondition. The generated parameters are passed to the function under test and the test driver validates result based on property assertions. This entire process is also referred as fuzzing, which is repeated based on a fixed range to identify any input parameters where the property assertions fail. There are a lot of libraries for property-based testing in various languages such as QuickCheck, fast-check, junit-quickcheck, ScalaCheck, etc. but we will use the api-mock-service library to demonstrate these capabilities for testing microservice APIs.

Following sections describe how the api-mock-service library can be used for testing microservice with fuzzing/property-based approaches and for mocking dependent services to produce the desired behavior:

Sample Microservices Under Test

A sample eCommerce application will be used to demonstrate property-based and generative testing. The application will use various microservices to implement online shopping experience. The primary purpose of this example is to show how different parameters can be passed to microservices, where microservice APIs will validate the input parameters, perform a simple business logic and then generate a valid result or an error condition. You can view the Open-API specifications for this sample app here.

Customer APIs

The customer APIs define operations to manage customers who shop online, e.g.:

Product APIs

The product APIs define operations to manage products that can be shopped online, e.g.:

Payment APIs

The payment APIs define operations to charge credit card and pay for online shopping, e.g.:

Order APIs

The order APIs define operations to purchase a product from the online store and it will use above APIs to validate customers, check product inventory, charge payments and then store record of orders, e.g.:

Defining Test Scenarios with Open-API Specifications

In this example, test scenarios will be generated by api-mock-service based on open-api specifications ecommerce-api.json by starting the mock service first as follows:

docker pull plexobject/api-mock-service:latest
docker run -p 8000:8000 -p 9000:9000 -e HTTP_PORT=8000 -e PROXY_PORT=9000 \
	-e DATA_DIR=/tmp/mocks -e ASSET_DIR=/tmp/assets api-mock-service

And then uploading open-API specifications for ecommerce-api.json:

curl -H "Content-Type: application/yaml" --data-binary @ecommerce-api.json \
	http://localhost:8000/_oapi

It will generate mock APIs for all microservices, e.g. you can produce result of products APIs, e.g.:

curl http://localhost:8000/products

to produce:

[
  {
    "id": "fd6a5ddb-35bc-47a9-aacb-9694ff5f8a32",
    "category": "TOYS",
    "inventory": 13,
    "name": "Se nota.",
    "price":{
      "amount":2,
      "currency": "USD"
    }
  },
  {
    "id": "47aab7d9-ecd2-4593-b1a6-c34bb5ca02bc",
    "category": "MUSIC",
    "inventory": 30,
    "name": "Proferuntur mortem.",
    "price":{
      "amount":23,
      "currency": "CAD"
    }
  },
  {
    "id": "ae649ae7-23e3-4709-b665-b1b0f436c97a",
    "category": "BOOKS",
    "inventory": 8,
    "name": "Cor.",
    "price":{
      "amount":13,
      "currency": "USD"
    }
  },
  {
    "id": "a3bd8426-e26d-4f66-8ee8-f55798440dc3",
    "category": "MUSIC",
    "inventory": 43,
    "name": "E diutius.",
    "price":{
      "amount":22,
      "currency": "USD"
    }
  },
  {
    "id": "7f328a53-1b64-4e4f-b6a6-7a69aed1b183",
    "category": "BOOKS",
    "inventory": 54,
    "name": "Dici utroque.",
    "price":{
      "amount":23,
      "currency": "USD"
    }
  }
]

Above response is randomly generated based on the properties defined in Open-API and calling this API will automatically generate all valid and error responses, e.g. calling “curl http://localhost:8000/products” again will return:

< HTTP/1.1 400 Bad Request
< Content-Type:
< Vary: Origin
< X-Mock-Path: /products
< X-Mock-Request-Count: 1
< X-Mock-Scenario: getProductByCategory-07ef44df0d38389ca9d589faaab9e458bd79e8abe7d2e1149e56c00820fac1fb
< Date: Tue, 20 Dec 2022 04:54:58 GMT
< Content-Length: 122
<
{ [122 bytes data]

* Connection #0 to host localhost left intact
{
  "errors": [
    "category_gym_bargain",
    "expand_tuna_stomach",
    "cage_enroll_between",
    "bulk_choice_category",
    "trend_agree_purse"
  ]
}

Applying Property-based/Generative Testing for Clients of Microservices

Upon uploading the Open-API specifications of microservices, the api-mock-service automatically generated templates for producing mock responses and error conditions, which can be customized for property-based and generative testing of microservice clients by defining constraints for generating input/output data and assertions for request/response validation.

Client-side Testing for Listing Products

You can find generated mock scenarios for listing products on the mock service using:

curl -v http://localhost:8000/_scenarios|jq '.'|grep "GET.getProductByCategory"

which returns:

"/_scenarios/GET/getProductByCategory-1a6d4d84e4a8a1ad706d671a26e66c419833b3a99f95cc442942f96d0d8f43f8/products": {
"/_scenarios/GET/getProductByCategory-6e522e565bb669ab3d9b09cc2e16b9d636220ec28a860a1cc30e9c5104e41f53/products": {
"/_scenarios/GET/getProductByCategory-7ede8f15af851323576a0c864841e859408525632eb002a1571676a0d835a0e1/products": {
"/_scenarios/GET/getProductByCategory-9ed14ecd11bbeb9f7bfde885d00efcbf168661354e4c48fe876c545e9a778302/products": {

and then invoking above URL paths, e.g.

curl -v http://localhost:8000/_scenarios/GET/getProductByCategory-7ede8f15af851323576a0c864841e859408525632eb002a1571676a0d835a0e1/products

which will return randomly generated response such as:

method: GET
name: getProductByCategory-7ede8f15af851323576a0c864841e859408525632eb002a1571676a0d835a0e1
path: /products
description: ""
order: 1
group: products
predicate: ""
request:
    match_query_params: {}
    match_headers: {}
    match_contents: '{}'
    path_params: {}
    query_params:
        category: '[\x20-\x7F]{1,128}'
    headers:
        "Content-Type": "application/json"
    contents: ""
response:
    headers: {}
    contents: '[{"category":"{{EnumString `BOOKS MUSIC TOYS`}}","id":"{{RandStringMinMax 0 0}}","inventory":"{{RandNumMinMax 10000 10000}}","name":"{{RandStringMinMax 2 50}}","price":{"amount":{{RandNumMinMax 0 0}},"currency":"{{RandStringMinMax 0 0}}"}}]'
    contents_file: ""
    status_code: 200
    match_headers: {}
    match_contents: '{"category":".+","id":"(__string__\\w+)","inventory":".+","name":"(__string__\\w+)","price.amount":".+","price.currency":"(__string__\\w+)"}'
wait_before_reply: 0s

We can customize above response contents using builtin template functions in the api-mock-service library to generate fuzz response, e.g.

    headers:
        "Content-Type":
          - "application/json"
    contents: >
      [
{{- range $val := Iterate 5}}
        {
          "id": "{{UUID}}",
          "category": "{{EnumString `BOOKS MUSIC TOYS`}}",
          "inventory": {{RandNumMinMax 1 100}},
          "name": "{{RandSentence 1 3}}",
          "price":{
            "amount":{{RandNumMinMax 1 25}},
            "currency": "{{EnumString `USD CAD`}}"
          }
        }{{if lt $val 4}},{{end}}
{{ end }}
      ]
    status_code: 200

In above example, we slightly improved the test template by generating product entries in a loop and using built-in functions to randomize the data. You can upload this scenario using:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/get_products.yaml \
	http://localhost:8000/_scenarios

You can also generate a template for returning an error response similarly, i.e.,

method: GET
name: error-products
path: /products
description: ""
order: 2
group: products
predicate: '{{NthRequest 2}}'
request:
    headers:
        "Content-Type": "application/json"
    query_params:
        category: '[\x20-\x7F]{1,128}'
response:
    headers: {}
    contents: '{"errors":["{{RandSentence 5 10}}"]}'
    contents_file: ""
    status_code: {{EnumInt 400 415 500}}
    match_contents: '{"errors":"(__string__\\w+)"}'
wait_before_reply: 0s

Invoking “curl -v http://localhost:8000/products“ will randomly return both of those test scenarios so that client code can test for various conditions.

Client-side Testing for Creating Products

You can find mock scenarios for creating products that were generated from above Open-API specifications using:

curl -v http://localhost:8000/_scenarios|jq '.'|grep "POST.saveProduct"

You can then customize scenarios as follows and then upload it:

method: POST
name: saveProduct
path: /products
description: ""
order: 0
group: products
request:
    match_query_params: {}
    match_headers: {}
    match_contents: '{"category":"(__string__(BOOKS|MUSIC|TOYS))","id":"(__string__\\w+)","inventory":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","name":"(__string__\\w+)","price.amount":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","price.currency":"(USD|CAD)"}'
    path_params: {}
    query_params: {}
    headers:
        "Content-Type": "application/json"
    contents: |
        category: MUSIC
        id: suavitas
        inventory: 5408.89695278641
        name: leporem
        price:
            amount: 7373.800941656166
            currency: cordis
response:
    headers: {}
    contents: '{"category":"{{EnumString `BOOKS MUSIC TOYS`}}","id":"{{RandStringMinMax 0 0}}","inventory":"{{RandNumMinMax 5 500}}","name":"{{RandStringMinMax 2 50}}","price":{"amount":{{RandNumMinMax 0 0}},"currency":"{{RandStringMinMax 0 0}}"}}'
    contents_file: ""
    status_code: 200
    match_headers: {}
    match_contents: '{"category":"(__string__(BOOKS|MUSIC|TOYS))","id":"(__string__\\w+)","inventory":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","name":"(__string__\\w+)","price.amount":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","price.currency":"(USD|CAD)"}'
    pipe_properties:
      - id
      - name
    assertions: []
wait_before_reply: 0s

And then invoke above POST /products API using:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/save_product.yaml http://localhost:8000/_scenarios

curl  http://localhost:8000/products -d \
  '{"category":"BOOKS","id":"123","inventory":"10","name":"toy 1","price":{"amount":12,"currency":"USD"}}'

The client code can test for product properties and other error scenarios can be added to simulate failure conditions.

Applying Property-based/Generative Testing for Microservices

The api-mock-service test scenarios defined above can also be used to test against the microservice implementations. You can start your service, e.g. we will use sample-openapi for testing purpose and then invoke test request for server-side testing using:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/get_products.yaml \
	http://localhost:8000/_scenarios
curl -H "Content-Type: application/yaml" --data-binary @fixtures/save_product.yaml \
	http://localhost:8000/_scenarios

curl -k -v -X POST http://localhost:8000/_contracts/products -d \
	'{"base_url": "http://localhost:8080", "execution_times": 5, "verbose": true}'

Above command will submit request to execute all scenarios belonging to products group five times and then return:

{
  "results": {
    "getProducts_0": {},
    "getProducts_1": {},
    "getProducts_2": {},
    "getProducts_3": {},
    "getProducts_4": {},
    "saveProduct_0": {
      "id": "895f584b-dc65-4950-982e-167680bcd133",
      "name": "Opificiis misera dei."
    },
    "saveProduct_1": {
      "id": "d89b6c16-549c-4baa-9dca-4dd9bb4b3ecf",
      "name": "Ea sumus aula teneant."
    },
    "saveProduct_2": {
      "id": "15dd54eb-fe89-4de8-9570-59fca20b9969",
      "name": "Vim odor et respondi."
    },
    "saveProduct_3": {
      "id": "e3769044-2a19-4e86-b0aa-9724378a0113",
      "name": "Me tua timeo an."
    },
    "saveProduct_4": {
      "id": "07ee20b9-df9a-487d-9ff9-cf76bef09a8f",
      "name": "Ruminando latinae omnibus."
    }
  },
  "metrics": {
    "getProducts_counts": 5,
    "getProducts_duration_seconds": 0.007,
    "saveProduct_counts": 5,
    "saveProduct_duration_seconds": 0.005
  },  
  "errors": {},
  "succeeded": 10,
  "failed": 0
}

You can also add custom assertions to validate the response in the save-product scenario:

method: POST
name: saveProduct
path: /products
description: ""
order: 0
group: products
predicate: ""
request:
    match_query_params: {}
    match_headers: {}
    match_contents: '{"category":"(BOOKS|MUSIC|TOYS)","id":"(__string__\\w+)","inventory":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","name":"(__string__\\w+)","price.amount":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","price.currency":"(USD|CAD)"}'
    path_params: {}
    query_params: {}
    headers:
        "Content-Type": "application/json"
    contents: |
        category: TOYS
        id: tempus
        inventory: 3890.9145609093966
        name: pleno
        price:
            amount: 5539.183583809511
            currency: "{{EnumString `USD CAD`}}"
response:
    headers: {}
    contents: '{"category":"{{EnumString `BOOKS MUSIC TOYS`}}","id":"{{RandStringMinMax 0 0}}","inventory":"{{RandNumMinMax 5 500}}","name":"{{RandStringMinMax 2 50}}","price":{"amount":{{RandNumMinMax 0 0}},"currency":"$"}}'
    contents_file: ""
    status_code: 200
    match_headers: {}
    match_contents: '{"category":"(__string__(BOOKS|MUSIC|TOYS))","id":"(__string__\\w+)","inventory":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","name":"(__string__\\w+)","price.amount":"(__number__[+-]?(([0-9]{1,10}(\\.[0-9]{1,5})?)|(\\.[0-9]{1,10})))","price.currency":"(USD|CAD)"}'
    pipe_properties:
      - id
      - name
    assertions:
        - VariableGE contents.inventory 5
        - VariableContains contents.category S
        - VariableContains contents.category X
wait_before_reply: 0s

If you try to run it again, the execution will fail with following error because none of the categories include X:

{
  "results": {
    "getProducts_0": {},
    "getProducts_1": {},
    "getProducts_2": {},
    "getProducts_3": {},
    "getProducts_4": {}
  },
  "errors": {
    "saveProduct_0": "failed to assert '{{VariableContains \"contents.category\" \"X\"}}' with value 'false'",
    "saveProduct_1": "failed to assert '{{VariableContains \"contents.category\" \"X\"}}' with value 'false'",
    "saveProduct_2": "failed to assert '{{VariableContains \"contents.category\" \"X\"}}' with value 'false'",
    "saveProduct_3": "failed to assert '{{VariableContains \"contents.category\" \"X\"}}' with value 'false'",
    "saveProduct_4": "failed to assert '{{VariableContains \"contents.category\" \"X\"}}' with value 'false'"
  },
  "succeeded": 5,
  "failed": 5
}

Summary

Using unit-testing and other forms of testing methodologies donâ€™t rule out presence of the bugs but they can greatly reduce the probability of bugs. However, with large sized test suites, the maintenance of tests incur a high development cost especially if those tests are brittle that requires frequent changes. The property-based/generative testing can help fill in gaps in unit testing while keeping size of the tests suite small. The api-mock-service tool is designed to mock and test microservices using fuzzing and property-based testing techniques. This mocking library can be used to test both clients and server side implementation and can also be used to generate error conditions that are not easily reproducible. This library can be a powerful tool in your toolbox when developing distributed systems with a large number services, which can be difficult to deploy and test locally. You can read more about the api-mock-library at â€œMocking and Fuzz Testing Distributed Micro Services with Record/Play, Templates and OpenAPI Specificationsâ€ and download it freely from https://github.com/bhatti/api-mock-service.

Comments (0)

December 4, 2022

Evolving Software Development Model for Building Distributed Systems

Filed under: Business,Technology — Tags: distributed, services — admin @ 5:19 pm

1. Overview

Over the last few decades, the software systems has evolved from the mainframe and client-server models to distributed systems and service oriented architecture. In early 2000, Amazon and Netflix forged ahead the industry to adopt microservices by applying Conwayâ€™s law and self-organizing teams structure where a small 2-pizza team own entire lifecycle of the microservices including operational responsibilities. The microservice architecture with small, cross-functional and independent structure has helped team agility to develop, test and deploy microservices independently. However, the software systems are becoming increasingly complex with the Cambrian explosion of microservices and the ecosystem of microservices is reaching a boiling point where building new features, releasing the enhancements and operational load from maintaining high availability, scalability, resilience, security, observability, etc are slowing down the development teams and raising the artificial complexity as a result of mixing different concerns.

Following diagram shows difference between monolithic architecture and microservice architecture:

As you can see in above diagram, each microservice is responsible for managing a number of cross-cutting concerns such as AuthN/AuthZ, monitoring, rate-limit, configuration, secret-management, etc., which adds a scope of work for development of each service. Following sections dive into fundamental causes of the complexity that comes with the microservices and a path forward for evolving the development methodology to build these microservices more effectively.

2. Perils in Building Distributed Systems

Following is a list of major pitfalls faced by the development teams when building distributed systems and services:

2.1 Coordinating Feature Development

The feature development becomes more convoluted with increase in dependencies of downstream services and any large change in a single service often requires the API changes or additional capabilities from multiple dependent services. This creates numerous challenges such as prioritization of features development among different teams, coordinating release timelines and making progress in absence of the dependent functionalities in the development environment. This is often tackled with additional personnel for project management and managing dependencies with Gantt charts or other project management tools but it still leads to unexpected delays and miscommunication among team members about the deliverables.

2.2 Low-level Concurrency Controls

The development teams often use imperative languages and apply low-level abstractions to implement distributed services where each request is served by a native thread in a web server. Due to extensive overhead of native threads such as stack size limits, the web server becomes constrained with the maximum number of concurrent requests it can support. In addition, these native threads in the web server often share a common state, which must be protected with a mutex, semaphore or lock to avoid data corruption. These low-level and primitive abstractions couple business logic with the concurrency logic, which add accidental complexity when safeguarding the shared state or communicating between different threads. This problem worsens with the time as the code size increases that results in subtle concurrency related heisenbugs where these bugs may produce incorrect results in the production environment.

2.3 Security

Each microservice requires implementing authentication, authorization, secure communication, key management and other aspects of the security. The development teams generally have to support these security aspects for each service they own and they have to be on top of any security patches and security vulnerabilities. For example, a zero-day log4j vulnerability in December 2021 created a havoc in most organizations as multiple services were affected by the bug and needed a patch immediately. This resulted in large effort by each development team to patch their services and deploy the patched services as soon as possible. Worst, the development teams had to apply patches multiple times because initial bug fixes from the log4j team didnâ€™t fully work, thus further multiplying the work by each development team. With growth of the dependency stack or bill of material for third party libraries in modern applications, the development teams face an overwhelming operational burden and enormous security risk to support their services safely.

2.4 Web Server

In general, each microservice requires a web server, which adds additional computing and administration overhead for deploying and running the service stack. The web server must be running all the time whether the service is receiving requests or not, thus wasting CPU, memory and storage resources needlessly.

2.5 Colocating Multiple Services

The development teams often start with a monolithic style applications that hosts multiple services on a single web server or with segregated application servers hosting multiple services on the same web server to lessen the development and deployment effort. The monolithic and service colocation architecture hinders speed, agility and extensibility as the code becomes complicated, harder to maintain and reasoned due to lack of isolation. In this style of deployment, computing resources can be entirely consumed by a single component or a bug in one service can crash the entire system. As each service may have unique runtime or usage characteristics, itâ€™s also arduous to scale a single service, to plan service capacity or to isolate service failures in a colocated runtime environment.

2.6 Cross Cutting Concerns

Building distributed systems require managing a lot of horizontal concerns such as security, resilience, business continuity, and availability but coupling these common concerns with the business logic results in inconsistencies and higher complexity by different implementations in microservices. Mixing these different concerns with business logic in microservices means each development team will have to solve those concerns independently and any omission or divergence may cause miserable user experience, faulty results, poor protection against load spikes or a security breach in the system.

2.7 Service Discovery

Though, microservices use various synchronous and asynchronous protocols for communicating with other services but they often store the endpoints of other services locally in the service configurations. This adds maintenance and operational burden for maintaining the endpoints for all dependent services in each development, test and production environment. In addition, services may not be able to apply certain containment or access control policies such as not invoking cross-region service to maintain lower latency or sustain a service for disaster recovery.

2.8 Architectural Quality Attributes

The architecture quality attributes include performance, availability sustainability, security, scalability, fault tolerance, performance, resilience, recovery and usability, etc. Each development team not only has to manage these attributes for each service but often requires coordination with other teams when scaling their services so that downstream services can handle additional load or meet the availability/reliability guarantees. Thus, the availability, fault tolerance, capacity management or other architectural concerns become tied with downstream services as an outage in any of those services directly affect upstream services. Thus, improving availability, scalability, fault tolerance or other architecture quality attributes often requires changes from the dependent services, which adds scope of the development work.

2.9 Superfluous Development Work

When a developing a microservice, a development team owns end-to-end development and release process that includes a full software development lifecycle support such as :

maintaining build scripts for CI/CD pipelines
building automation tools for integration/functional/load/canary tests
defining access policies related to throttling/rate-limits
implementing consistent error handling, idempotency behavior, contextual information across services
adding alarms/metrics/monitoring/observability/notification/logs
providing customized personal and system dashboards
supporting data encryption and managing secret keys
defining security policies related to AuthN/AuthZ/Permissons/ACL
defining network policies related to VPN, firewall, load-balancer, network/gateway/routing configuration
adding compression, caching, and any other common pre/post processing for services

As a result, any deviations or bugs in implementation of these processes or misconfiguration in underlying infrastructure can lead to conflicting user experience, security gaps and outages. Some organizations maintain lengthy checklists and hefty review processes for applying best practices before the software release but they often miss key learnings from other teams and slow down the development process due to cumbersome release process.

2.10 Yak shaving when developing and testing features

Due to enormous artificial complexity of microservices with abysmal dependency stack, the development teams have to spend inordinate amount of time in setting up a development environment when building a new feature. The feature development requires testing the features using unit tests with a mock behavior of dependent services and using integration tests with real dependent services in a local development environment. However, itâ€™s not always possible to run complete stack locally and fully test the changes for new features, thus the developers are encumbered with finding alternative integration environment where other developers may also be testing their features. All this yak shaving makes the software development awfully tedious and error prone because developers canâ€™t test their features in isolation with a high confidence. This means that development teams find bugs later in phases of the release process, which may require a rollback of feature changes and block additional releases until bugs are fixed in the main/release branch.

3. Path to the Enlightenment

Following are a few recommendations to remedy above pitfalls in the development of distributed systems and microservices:

3.1 Higher level of Development Abstraction

Instead of using low-level imperative languages or low-level concurrency controls, high-level abstractions can be applied to simplify the development of the microservices as follows:

3.1.1 Actor Model

The Actor model was first introduced in 1973 by Carl Hewitt and it provides a high-level abstraction for concurrent computation. An actor uses a mailbox or a queue to store incoming messages and processes one message at a time using local state in a single green thread or a coroutine. An actor can create other actors or send messages without using any lock-based synchronization and blocking. Actors are reactive so they cannot initiate any action on their own, instead they simply react to external stimuli in the form of message passing. Actors provide much better error handling where applications can define a hierarchy of actors with parent/child relationships and a child actor may crash when encountering system errors, which are monitored and supervised by parent actors for failure recovery.

Actor model is supported natively in many languages such Erlang/Elixir, Scala, Swift and Pony and itâ€™s available as a library in many other languages. In these languages, actors generally use green threads, coroutines or a preemptive scheduler to schedule actors with non-blocking I/O operations. As the actors incur much lower overhead compare to native threads, they can be used to implement microservices with much greater scalability and performance. In addition, message passing circumvents the need to guard the shared state as each actors only maintains a local state, which provides more robust and reliable implementation of microservices. Here is an example of actor model in Erlang language:

-module(sum).
-export([init/0, add/1, get/0]).

init() ->
    Pid = spawn(fun() -> loop(0) end),
    register(sumActor, Pid).

loop(N) ->
    receive
        {add, X} -> loop(N+X);
        {Client, get} ->
            Client ! N,
            loop(N)
    end.

add(X) ->
    sumActor ! {add, X}.

get() ->
    sumActor ! {self(), get},
    receive Result -> Result end.

In above example, an actor is spawned to run loop function, which uses a tail recursion to receive next message from the queue and then processes it based on the tag of the message such as add or get. The client code uses a symbol sumActor to send a message, which is registered with a local registry. As an actor only maintains local state, microservices may use use external data store and manage a state machine using orchestration based SAGA pattern to trigger next action.

3.1.2 Function as a service (FaaS) and Serverless Computing

The FaaS and serverless computing can be used to develop microservices where business logic can be embedded within a serverless functions but any platform services such as storage, messaging, caching, SMS can be exposed via Backend as a Service (BaaS). The Backend as a Service (BaaS) adds additional business logic on top of Platform as a Service (PaaS).

3.1.3 Agent based computing

Microservices architecture decouples data access from the business logic and microservices fetch data from the data-store, which incurs a higher overhead if the business logic needs to fetch a lot of data for processing or filtering before generating results. As opposed, agent style computing allow migrating business logic remotely where data or computing resources reside, thus it can process data more efficiently. In a simplest example, an agent may behave like a stored procedure where a function is passed to a data store for executing a business logic, which is executed within the database but other kind of agents may support additional capabilities to gather data from different data stores or sources and then produces desired results after processing the data remotely.

3.2 Service and Schema Registry

The service registry allows microservices to register the endpoints so that other services can look up the endpoints for communication with them instead of storing the endpoints locally. This allows service registry to enforce any authorization and access policies for communication based on geographic location or other constraints. A service registry may also allow registering mock services for testing in a local development environment to facilitate feature development. In addition, the registry may store schema definitions for the API models so that services can validate requests/responses easily or support multiple versions of the API contracts.

3.3 API Router, Reverse-Proxy or Gateway

API router, reverse-proxy or an API gateway are common patterns with microservices for routing, monitoring, versioning, securing and throttling APIs. These patterns can also be used with FaaS architecture where an API gateway may provide these capabilities and eliminate the need to have a web server for each service function. Thus, API gateway or router can result in lowering computing cost for each service and reducing the complexity for maintaining non-functional capabilities or -ilities.

3.4 Virtualization

Virtualization abstracts computer hardware and uses a hypervisor to create multiple virtual computers with different operating systems and applications on top of a single physical computer.

3.4.1 Virtual Machines

The initial implementation of virtualization was based on Virtual Machines for building virtualized computing environments and emulating a physical computer. The virtual machines use a hypervisor to communicate with the physical computer.

3.4.2 Containers

Containers implement virtualization using host operating system instead of a hypervisor, thus provide more light-weight and faster provisioning of computing resources. The containers use platforms such as Docker and Kubernetes to execute applications and services, which are bundled into images based on Open Container Initiative (OCI) standard.

3.4.3 MicroVM

MicroVMs such as Firecracker and crosVM are based on kernel-based VM (KVM) and use hostOS acting as a hypervisor to provide isolation and security. As MicroVMs only include essential features for network, storage, throttling and metadata so they are quick to start and can scale to support multiple VMs with minimal overhead. A number of serverless platforms such as AWS Lambda, appfleet, containerd, Fly.io, Kata, Koyeb, OpenNebula, Qovery, UniK, and Weave FireKube have adopted Firecracker VM, which offers low overhead for starting a new virtual machine or executing a serverless workload.

3.4.4 WebAssembly

The WebAssembly is a stack-based virtual machine that can run at the edge or in cloud. The applications written Go, C, Rust, AssemblyScript, etc. are compiled into WebAssembly binary and are then executed on a WebAssembly runtime such as extism, faasm, wasmtime, wamr, wasmr and wagi. The WebAssembly supports WebAssembly System Interface (WASI) standard, which provides access to the systems APIs for different operating systems similar to POSIX standard. There is also an active development of WebAssembly Component Model with proposals such as WebIDL bindings and Interface Types. This allows you to write microservices in any supported language, compile the code into WASM binary and then deploy in a managed platform with all support for security, traffic management, observability, etc.

A number of WebAssembly platforms such as teaclave, wasmCloud, fermyon and Lunatic have also adopted Actor model to build a platform for writing distributed applications.

“So will wasm replace Docker?” No, but imagine a future where Docker runs linux containers, windows containers and wasm containers side by side. Over time wasm might become the most popular container type. Docker will love them all equally, and run it all 🙂 https://t.co/qVq3fomv9d
— Solomon Hykes (@solomonstre) March 28, 2019

If WASM+WASI existed in 2008, we wouldn’t have needed to created Docker. That’s how important it is. Webassembly on the server is the future of computing. A standardized system interface was the missing link. Let’s hope WASI is up to the task!

3.5 Instrumentation and Service Binding

In order to reduce service code that deals specifically with non-functional capabilities such as authentication, authorization, logging, monitoring, etc., the business service can be instrumented to provide those capabilities at compile or deployment time. This means that the development team can largely focus on the business requirements and instrumentation takes care of adding metrics, failure reporting, diagnostics, monitoring, etc. without any development work. In addition, any external platform dependencies such as messaging service, orchestration, database, key/value store, caching can be injected into the service dynamically at runtime. The runtime can be configured to provide different implementation for the platform services, e.g. it may use a local Redis server for key/value store in a hosted environment or AWS/Azure’s implementation in a cloud environment.

When deploying services with WebAssembly support, the instrumentation may use WebAssembly libraries for extending the services to support authentication, rate-limiting, observability, monitoring, state management, and other non-functional capabilities, so that the development work can be shortened as shown below:

3.6 Orchestration and Choreography

The orchestration and choreography allows writing microservices that can be easily composed to provide a higher abstractions for business services. In orchestration design, a coordinator manages synchronous communication among different services whereas choreography uses event-driven architecture to communicate asynchronously. An actor model fits naturally with event based architecture for communicating with other actors and external services. However, orchestration services can be used to model complex business processes where a SAGA pattern is used to manage state transitions for different activities. Microservices may also use staged event-driven architecture (SEDA) to decompose complex event-driven services into a set of stages that are connected by different queues, which supports better modularity and code-reuse. SEDA allows enforcing admission control on each event queue and grants flexible scheduling for processing events based on adaptive workload controls and load shedding policies.

3.7 Automation, Continuous Deployment and Infrastructure as a code

Automation is a key to remove any drudgery work during the development process and consolidate common build processes such as continuous integration and deployment for improving the productivity of a development team. The development teams can employ continuous delivery to deploy small and frequent changes by developers. The continuous deployment often uses rolling updates, blue/green deployments or canary deployments to minimize disruption to end users. The monitoring system watches for error rates at each stage of the deployment and automatically rollbacks changes if a problem occurs.

Infrastructure as code (IaC) uses a declarative language to define development, test and production environment, which is managed by the source code management software. These provisioning and configuration logic can be used by CI/CD pipelines to automatically deploy and test environments. Many cloud vendors provide support for IaC such as Azure Resource Manager (ARM), AWS Cloud Development Kit (CDK), Hashicorp Terraform etc to deploy computing resources.

3.8 Proxy Patterns

Following sections shows implementing cross cutting concerns using proxy patterns for building microservices:

3.8.1 Sidecar Model

The sidecar model helps modularity and reusability where an application requires two containers: application container and sidebar container where sidebar container provides additional functionality such as adding SSL proxy for the service, observability, collecting metrics for the application container.

The Sidecar pattern generally uses another container to proxy off all traffic, which enforces security, access control, throttling before forwarding the incoming requests to the microservice.

3.8.2 Service Mesh

The service mesh uses a mesh of sidecar proxies to enable:

Dynamic request routing for blue-green deployments, canaries, and A/B testing
Load balancing based on latency, geographic locations or health checks
Service discovery based on a version of the service in an environment
TLS/mTLS encryption
Authentication and Authorization
Keys, certificates and secrets management
Rate limiting and throttling
State management and PubSub
Observability, metrics, monitoring, logging
Distributed tracing
Traffic management and traffic splitting
Circuit breaker and retries using libraries like Finagle, Stubby, and Hysterix to isolate unhealthy instances and gradually adding them back after successful health checks
Error handling and fault tolerance
Control plane to manage routing tables, service discovery, load balancer and other service configuration

The service mesh pattern uses a data plane to host microservices and all incoming and outgoing requests go through a sidecar proxy that implements cross cutting concerns such as security, routing, throttling, etc. The control-plan in mesh network allows administrators to change the behavior of data plane proxies or configuration for data access. Popular service mesh frameworks include Consul, Distributed Application Runtime (Dapr), Envoy, Istio, Linkerd, Kong, Koyeb and Kuma for providing control-plane and data-plane with builtin support for networking, observability, traffic management and security. Dapr mesh network also supports Actor-model using the Orleans Virtual Actor pattern, which leverages the scalability and reliability guarantees of the underlying platform.

3.9 Cell-Based Architecture

A Cell-based Architecture (CBA) allows grouping business functions into single units called “cells”, which provides better scalability, modularity, composibility, disaster-recovery and governance for building microservices. It reduces the number of deployment units for microservices, thus simplifying the artificial complexity in building distributed systems.

A cell is an immutable collection of components and services that is independently deployable, manageable, and observable. The components and services in a cell communicate with each other using local network protocols and other cells via network endpoints through a cell gateway or brokers. A cell defines an API interface as part of its definition, which provides better isolation, short radius blast and agility because releases can be rolled out cell by cell. However, in order to maintain the isolation, a cell should be either self contained for all service dependencies or dependent cells should be on the same region so that there is no cross region dependency.

3.10 12-Factor app

The 12-factor app is a set of best practices from Heroku for deploying applications and services in a virtualized environment. It recommends using declarative configuration for automation, enabling continuous deployment and other best practices. Though, these recommendations are a bit dated but most of them still holds except you may consider storing credentials or keys in a secret store or in an encrypted files instead of simply using environment variables.

4. New Software Development Workflow

The architecture patterns, virtual machines, WebAssembly and serverless computing described in above section can be used to simplify the development of microservices and easily support both functional and non-functional requirements of the business needs. Following sections describe how these patterns can be integrated with the software development lifecycle:

4.1 Development

4.1. Adopting WebAssembly as the Lingua Franca

WebAssembly along with its component model and WASI standard has been gaining support and many popular languages now can be compiled into wasm binary. Microservices can be developed in supported languages and ubiquitous WebAssembly support in production environment can reduce the development and maintenance drudgery for the application development teams. In addition, teams can leverage a number of serverless platforms that support WebAssembly such as extism, faasm, netlify, vercel, wasmcloud and wasmedge to reduce the operational load.

4.1.2 Instrumentation and Service Binding

The compiled WASM binary for microservices can be instrumented similar to aspects so that the generated code automatically supports horizontal concerns such as circuit breaker, diagnostics, failure reporting, metrics, monitoring, retries, tracing, etc. without any additional endeavor by the application developer.

In addition, the runtime can use service mesh features to bind external platform services such as event bus, data store, caching, service registry or dependent business services. This simplifies the development effort as development team can use the shared services for developing microservices instead of configuring and deploying infrastructure for each service. The shared infrastructure services support multi-tenancy and distinct namespaces for each microservice so that it can manage its state independently.

4.1.3 Adopting Actor Model for Microservices

As discussed above, the actor model offers a light-weight and highly concurrent model to implement a microservice APIs. The actors based microservices can be coded in any supported high-level language but then compiled into WebAssembly with a WASI support. A number of WebAssembly serverless platforms including Lunatic and wasmCloud already support Actor model while other platforms such as fermyon use http based request handlers, which are invoked for each request similar to actors based message passing. For example, here is a sample actor model in wasmCloud in Rust language though any language with wit-bindgen is supported as well:

#[derive(Debug, Default, Actor, HealthResponder)]
#[services(Actor, HttpServer)]
struct HelloActor {}

#[async_trait]
impl HttpServer for HelloActor {
    async fn handle_request(
        &self,
        _ctx: &Context,
        req: &HttpRequest,
    ) -> std::result::Result<HttpResponse, RpcError> {
        let text=form_urlencoded::parse(req.query_string.as_bytes())
            .find(|(n, _)| n == "name")
            .map(|(_, v)| v.to_string())
            .unwrap_or_else(|| "World".to_string());
        Ok(HttpResponse {
            body: format!("Hello {}", text).as_bytes().to_vec(),
            ..Default::default()
        })
    }
}

The wasmCloud supports Contract-driven design and development (CDD) using Wasmcloud interfaces based on smithy IDL for building microservices and composable systems. There is also a pending work to support OpenFaas with wasmCloud to invoke functions on capability providers with appropriate privileges.

Following example demonstrates similar capability with fermyon, which can be deployed to Fermyon Cloud:

use anyhow::Result;
use spin_sdk::{
    http::{Request, Response},
    http_component,
};
#[http_component]
fn hello_rust(req: Request) -> Result<Response> {
    println!("{:?}", req.headers());
    Ok(http::Response::builder()
        .status(200)
        .header("foo", "bar")
        .body(Some("Hello, Fermyon".into()))?)
}

Following example shows how Dapr and WasmEdge work together to support lightweight WebAssembly-based microservices in a cloud-native environment:

fn main() -> std::io::Result<()> {
    let port = std::env::var("PORT").unwrap_or(9005.to_string());
    println!("new connection at {}", port);
    let listener = TcpListener::bind(format!("127.0.0.1:{}", port))?;
    loop {
        let _ = handle_client(listener.accept()?.0);
    }
}

fn handle_client(mut stream: TcpStream) -> std::io::Result<()> {
  ... ...
}

fn handle_http(req: Request<Vec<u8>>) -> bytecodec::Result<Response<String>> {
  ... ...
}

The WasmEdge can also be used with other serverless platforms such as Vercel, Netlify, AWS Lambda, SecondState and Tencent.

4.1.4 Service Composition with Orchestration and Choreography

As described above, actors based microservices can be extended with the orchestration patterns such as SAGA and choreography/event driven architecture patterns such as SEDA to build composable services. These design patterns can be used to build loosely coupled and extensible systems where additional actors and components can be added without changing existing code.

4.2 Deployment and Runtime

4.2.1 Virtual Machines and Serverless Platform

Following diagram shows the evolution of virtualized environments for hosting applications, services, serverless functions and actors:

In this architecture, the microservices are compiled into wasm binary, instrumented and then deployed in a micro virtual machine.

4.2.2 Sidecar Proxy and Service Mesh

Though, the service code will be instrumented with additional support for error handling, metrics, alerts, monitoring, tracing, etc. before the deployment but we can further enforce access policies, rate limiting, key management, etc. using a sidecar proxy or service mesh patterns:

For example, WasmEdge can be integrated with Dapr service mesh for adding building blocks for state management, event bus, orchestration, observability, traffic routing, and bindings to external services. Similarly, wasmCloud can be extended with additional non-functional capabilities by implementing capability provider. wasmCloud also provides a lattice, self-healing mesh network for simplifying communication between actors and capability providers.

4.2.3 Cellular Deployment

As described above, Cell-based Architecture (CBA) provides better scalability, modularity, composibility and business continuity for building microservices. Following diagram shows how above design with virtual machines, service-mesh and WebAssembly can be extended to support cell-based architecture:

In above architecture, each cell deploys a set of related microservices for an application that persists state in a replicated data store and communicate with other cells with an event-bus. In this model, separate cells are employed to access data-plane services and control-plane services for configuration and administration purpose.

5. Conclusion

The transition of monolithic services towards microservices architecture over last many years has helped development teams to be more agile in building modular and reusable code. However, as teams are building a growing number new microservices, they are also tasked with supporting non-functional requirements for each service such as high availability, capacity planning, scalability, performance, recovery, resilience, security, observability, etc. In addition, each microservice may depend on numerous other microservices, thus a microservice becomes susceptible to scaling/availability limits or security breaches in any of the downstream services. Such tight coupling of horizontal concerns escalates complexity, development undertaking and operational load by each development team resulting in larger time to market for new features and larger risk for outages due to divergence in implementing non-functional concerns. Though, serverless computing, function as a service (Faas) and event-driven compute services have emerged to solve many of these problems but they remain limited in the range of capabilities they offer and lack a common standards across vendors. The advancements in micro virtual machines and containers have created a boon to the serverless platforms such as appfleet, containerd, Fly.io, Kata, Koyeb, OpenNebula, Qovery, UniK, and Weave FireKube. In addition, widespread adoption of WebAssembly along with its component model and WASI standard are helping these serverless platforms such as extism, faasm, netlify, vercel, wasmcloud and wasmedge to build more modular and reusable components. These serverless platforms allow the development teams to primarily focus on building business features and offload all non-functional concerns to the underlying platforms. Many of these serverless platforms also support service mesh and sidecar patterns so that they can bind platform and dependent services and automatically handle concerns such as throttling, security, state management, secrets, key management, etc. Though, cell-based architecture is still relatively new and is only supported by more matured serverless and cloud platforms, but it further raises scalability, modularity, composibility, business continuity and governance of microservices. As each cell is isolated, it adds agility to deploy code changes to a single cell and use canary tests to validate the changes before deploying code to all cells. Due to such isolated deployment, cell-based architecture reduces the blast radius if a bug is found during validation or other production issues are discovered. Finally, automating continuous deployment processes and applying Infrastructure as code (IaC) can simplify local development and deployment so that developers use the same infrastructure setup for local testing as the production environment. This means that the services can be deployed in any environment consistently, thus reduces any manual configuration or subtle bugs due to misconfigurations.

In summary, the development teams will greatly benefit from the architecture patterns, virtualized environments, WebAssembly and serverless platforms described above so that application developers are not burdened with maintaining horizontal concerns and instead they focus on building core product features, which will be the differentiating factors in the competing markets. These serverless and managed platform not only boosts developer productivity but also lowers the infrastructure cost, operational overhead, cruft development work and the time to market for releasing new features.

Comments (0)

October 18, 2022

Mocking and Fuzz Testing Distributed Micro Services with Record/Play, Templates and OpenAPI Specifications

Filed under: GO,REST,Technology — admin @ 11:36 am

Building large distributed systems often requires integrating with multiple distributed micro-services that makes development a particularly difficult as it’s not always easy to deploy and test all dependent services in a local environment with constrained resources. In addition, you might be working on a large system with multiple teams where you may have received new API specs from another team but the API changes are not available yet. Though, you can use mocking frameworks based on API specs when writing a unit tests but integration or functional testing requires an access to the network service. A common solution that I have used in past projects is to configure a mock service that can simulate different API operations. I wrote a JVM based mock-service many years ago with following use-cases:

Use-Cases

As a service owner, I need to mock remote dependent service(s) by capturing/recording request/responses through an HTTP proxy so that I can play it back when testing the remote service(s) without connecting with them.

As a service owner, I need to mock remote dependent service(s) based on a open-api/swagger specifications so that my service client can test all service behavior per specifications for the remote service(s) even when remote service is not fully implemented or accessible.

As a service owner, I need to mock remote dependent service(s) based on a mock scenario defined in a template so that my service client can test service behavior per expected request/response in the template even when remote service is not fully implemented or accessible.

As a service owner, I need to inject various response behavior and faults to the output of a remote service so that I can build a robust client that prevents cascading failures and is more resilient to unexpected faults.

As a service owner, I need to define test cases with faulty or fuzz responses to test my own service so that I can predict how it will behave with various input data and assert the service response based on expected behavior.

Features

API mock service for REST/HTTP based services with following features:

Record API request/response by working as a HTTP proxy server (native http/https or via API) between client and remote service.
Playback API response that were previously recorded based on request parameters.
Define API behavior manually by specifying request parameters and response contents using static data or dynamic data based on GO templating language.
Generate API behavior from open standards such as Open API/Swagger and automatically create constraints and regex based on the specification.
Customize API behavior using a GO template language so that users can generate dynamic contents based on input parameters or other configuration.
Generate large responses using the template language with dynamic loops so that you can test performance of your system.
Define multiple test scenarios for the API based on different input parameters or simulating various error cases that are difficult to reproduce with real services.
Store API request/responses locally as files so that itâ€™s easy to port stubbed request/responses to any machine.
Allow users to define API request/response with various formats such as XML/JSON/YAML and upload them to the mock service.
Support test fixtures that can be uploaded to the mock service and can be used to generate mock responses.
Define a collection of helper methods to generate different kind of random data such as UDID, dates, URI, Regex, text and numeric data.
Ability to playback all test scenarios or a specific scenario and change API behavior dynamically with different input parameters.
Support multiple mock scenarios for the same API that can be selected either using round-robin order, custom predicates based on parameters or based on scenario name.
Inject error conditions and artificial delays so that you can test how your system handles error conditions that are difficult to reproduce or use for game days/chaos testing.
Generate client requests for a remote API for chaos and stochastic testing where a set of requests are sent with a dynamic data generated based on regex or other constraints.

I used this service in many past projects, however I felt it needed a bit fresh approach to meet above goals so I rewrote it in GO language, which has a robust support for writing network services. You can download the new version from https://github.com/bhatti/api-mock-service. As, it’s written in GO, you can either download GO runtime environment or use Docker to install it locally. If you haven’t installed docker, you can download the community version from https://docs.docker.com/engine/installation/ or find installer for your OS on https://docs.docker.com/get-docker/.

docker build -t api-mock-service .
docker run -p 8000:8080 -p 8081:8081 -e HTTP_PORT=8080 PROXY_PORT=8081 \
	-e DATA_DIR=/tmp/mocks -e ASSET_DIR=/tmp/assets api-mock-service

or pull an image from docker hub (https://hub.docker.com/r/plexobject/api-mock-service), e.g.

docker pull plexobject/api-mock-service:latest
docker run -p 8000:8080 -p 8081:8081 -e HTTP_PORT=8080 PROXY_PORT=8081 -e DATA_DIR=/tmp/mocks \
	-e ASSET_DIR=/tmp/assets plexobject/api-mock-service:latest

Alternatively, you can run it locally with GO environment, e.g.,

make && ./out/bin/api-mock-service

For full command line options, execute api-mock-service -h that will show you command line options such as:

./out/bin/api-mock-service -h
Starts mock service

Usage:
  api-mock-service [flags]
  api-mock-service [command]

Available Commands:
  chaos       Executes chaos client
  completion  Generate the autocompletion script for the specified shell
  help        Help about any command
  version     Version will output the current build information

Flags:
      --assetDir string   asset dir to store static assets/fixtures
      --config string     config file
      --dataDir string    data dir to store mock scenarios
  -h, --help              help for api-mock-service
      --httpPort int      HTTP port to listen
      --proxyPort int     Proxy port to listen

Recording a Mock Scenario via HTTP/HTTPS Proxy

Once you have the API mock service running, the mock service will start two ports on startup, first port (default 8080) will be used to record/play mock scenarios, updating templates or uploading OpenAPIs. The second port (default 8081) will setup an HTTP/HTTPS proxy server that you can point to record your scenarios, e.g.

export http_proxy="http://localhost:8081"
export https_proxy="http://localhost:8081"

curl -k -v -H "Authorization: Bearer sk_test_xxxx" \
	https://api.stripe.com/v1/customers/cus_xxx/cash_balance

Above curl command will automatically record all requests and responses and create mock scenario to play it back. For example, if you call the same API again, it will return a local response instead of contacting the server. You can customize the proxy behavior for record by adding X-Mock-Record: true header to your request.

Recording a Mock Scenario via API

Alternatively, you can use invoke an internal API as a pass through to invoke a remote API so that you can automatically record API behavior and play it back later, e.g.

% curl -H "X-Mock-Url: https://api.stripe.com/v1/customers/cus_**/cash_balance" \
	-H "Authorization: Bearer sk_test_***" http://localhost:8080/_proxy

In above example, the curl command is passing the URL of real service as an HTTP header Mock-Url. In addition, you can pass other authorization headers as needed.

Viewing the Recorded Mock Scenario

The API mock-service will store the request/response in a YAML file under a data directory that you can specify. For example, you may see a file under:

default_mocks_data/v1/customers/cus_***/cash_balance/GET/recorded-scenario-***.scr

Note: the sensitive authentication or customer keys are masked in above example but you will see following contents in the captured data file:

method: GET
name: recorded-v1-customers-cus
path: /v1/customers/cus_**/cash_balance
description: recorded at 2022-10-29 04:26:17.24776 +0000 UTC
request:
     match_query_params: {}
     match_headers: {}
     match_content_type: ""
     match_contents: ""
     example_path_params: {}
     example_query_params: {}
     example_headers:
         Accept: '*/*'
         Authorization: Bearer sk_test_xxx
         User-Agent: curl/7.65.2
         X-Mock-Url: https://api.stripe.com/v1/customers/cus_/cash_balance
     example_contents: ""
response:
    headers:
        Access-Control-Allow-Credentials:
            - "true"
        Access-Control-Allow-Methods:
            - GET, POST, HEAD, OPTIONS, DELETE
        Access-Control-Allow-Origin:
            - '*'
        Access-Control-Expose-Headers:
            - Request-Id, Stripe-Manage-Version, X-Stripe-External-Auth-Required, X-Stripe-Privileged-Session-Required
        Access-Control-Max-Age:
            - "300"
        Cache-Control:
            - no-cache, no-store
        Content-Length:
            - "168"
        Content-Type:
            - application/json
        Date:
            - Sat, 29 Oct 2022 04:26:17 GMT
        Request-Id:
            - req_lOP4bCsPIi5hQC
        Server:
            - nginx
        Strict-Transport-Security:
            - max-age=63072000; includeSubDomains; preload
        Stripe-Version:
            - "2018-09-06"
    content_type: application/json
    contents: |-
        {
          "object": "cash_balance",
          "available": null,
          "customer": "cus_",
          "livemode": false,
          "settings": {
            "reconciliation_mode": "automatic"
          }
        }
    contents_file: ""
    status_code: 200
wait_before_reply: 0s

Above example defines a mock scenario for testing /v1/customers/cus_**/cash_balance path. A test scenario includes:

Predicate

This is a boolean condition if you need to enable or disable a scenario test based on dynamic parameters or request count.

Group

This specifies the group for related test scenarios.

Request Matching Parameters:

The matching request parameters will be used to select the mock scenario to execute and you can use regular expressions to validate:

URL Query Parameters
URL Request Headers
Request Body

You can use these parameters so that test scenario is executed only when the parameters match, e.g.

    match_query_params:
      name: [a-z0-9]{1,50}
    match_headers:
      Content-Type: "application/json"

The matching request parameters will be used to select the mock scenario to execute and you can use regular expressions to validate, e.g. above example will be matched if content-type is application/json and it will validate that name query parameter is alphanumeric from 1-50 size.

Example Request Parameters:

The example request parameters show the contents captured from the record/play so that you can use and customize to define matching parameters:

URL Query Parameters
URL Request Headers
Request Body

Response Properties

The response properties will include:

Response Headers
Response Body statically defined or loaded from a test fixture
Response can also be loaded from a test fixture file
Status Code
Matching header and contents
Assertions You can copy recorded scenario to another folder and use templates to customize it and then upload it for playback.

The matching header and contents use match_headers and match_contents similar to request to validate response in case you want to test response from a real service for chaos testing. Similarly, assertions defines a set of predicates to test against response from a real service:

    assertions:
        - VariableContains contents.id 10
        - VariableContains contents.title illo
        - VariableContains headers.Pragma no-cache

Above example will check API response and verify that id property contains 10, title contains illo and result headers include Pragma: no-cache header.

Playback the Mock API Scenario

You can playback the recorded response from above example as follows:

% curl http://localhost:8080/v1/customers/cus_***/cash_balance

Which will return captured response such as:

{
  "object": "cash_balance",
  "available": null,
  "customer": "cus_***",
  "livemode": false,
  "settings": {
    "reconciliation_mode": "automatic"
  }
}%

Though, you can customize your template with dynamic properties or conditional logic but you can also send HTTP headers for X-Mock-Response-Status to override HTTP status to return or X-Mock-Wait-Before-Reply to add artificial latency using duration syntax.

Debug Headers from Playback

The playback request will return mock-headers to indicate the selected mock scenario, path and request count, e.g.

X-Mock-Path: /v1/jobs/{jobId}/state
X-Mock-Request-Count: 13
X-Mock-Scenario: setDefaultState-bfb86eb288c9abf2988822938ef6d4aa3bd654a15e77158b89f17b9319d6f4e4

Upload Mock API Scenario

You can customize above recorded scenario, e.g. you can add path variables to above API as follows:

method: GET
name: stripe-cash-balance
path: /v1/customers/:customer/cash_balance
request:
    match_headers:
        Authorization: Bearer sk_test_[0-9a-fA-F]{10}$
response:
    headers:
        Access-Control-Allow-Credentials:
            - "true"
        Access-Control-Allow-Methods:
            - GET, POST, HEAD, OPTIONS, DELETE
        Access-Control-Allow-Origin:
            - '*'
        Access-Control-Expose-Headers:
            - Request-Id, Stripe-Manage-Version, X-Stripe-External-Auth-Required, X-Stripe-Privileged-Session-Required
        Access-Control-Max-Age:
            - "300"
        Cache-Control:
            - no-cache, no-store
        Content-Type:
            - application/json
        Request-Id:
            - req_2
        Server:
            - nginx
        Strict-Transport-Security:
            - max-age=63072000; includeSubDomains; preload
        Stripe-Version:
            - "2018-09-06"
    content_type: application/json
    contents: |-
        {
          "object": "cash_balance",
          "available": null,
          "customer": {{.customer}}
          "livemode": false,
          "page": {{.page}}
          "pageSize": {{.pageSize}}
          "settings": {
            "reconciliation_mode": "automatic"
          }
        }
    status_code: 200
wait_before_reply: 1s

In above example, I assigned a name stripe-cash-balance to the mock scenario and changed API path to /v1/customers/:customer/cash_balance so that it can capture customer-id as a path variable. I added a regular expression to ensure that the HTTP request includes an Authorization header matching Bearer sk_test_[0-9a-fA-F]{10}$ and defined dynamic properties such as {{.customer}}, {{.page}} and {{.pageSize}} so that they will be replaced at runtime.

The mock scenario uses builtin template syntax of GO. You can then upload it as follows:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/stripe-customer.yaml \
	http://localhost:8080/_scenarios

and then play it back as follows:

curl -v -H "Authorization: Bearer sk_test_0123456789" \
	"http://localhost:8080/v1/customers/123/cash_balance?page=2&pageSize=55"

and it will generate:

 HTTP/1.1 200 OK
< Content-Type: application/json
< X-Mock-Request-Count: 1
< X-Mock-Scenario: stripe-cash-balance
< Request-Id: req_2
< Server: nginx
< Strict-Transport-Security: max-age=63072000; includeSubDomains; preload
< Stripe-Version: 2018-09-06
< Date: Sat, 29 Oct 2022 17:29:12 GMT
< Content-Length: 179
<
{
  "object": "cash_balance",
  "available": null,
  "customer": 123
  "livemode": false,
  "page": 2
  "pageSize": 55
  "settings": {
    "reconciliation_mode": "automatic"
  }

As you can see, the values of customer, page and pageSize are dynamically updated and the response header includes name of mock scenario with request counts. You can upload multiple mock scenarios for the same API and the mock API service will play it back sequentially. For example, you can upload another scenario for above API as follows:

method: GET
name: stripe-customer-failure
path: /v1/customers/:customer/cash_balance
request:
    match_headers:
        Authorization: Bearer sk_test_[0-9a-fA-F]{10}$
response:
    headers:
        Stripe-Version:
            - "2018-09-06"
    content_type: application/json
    contents: My custom error
    status_code: 500
wait_before_reply: 1s

And then play it back as before:

curl -v -H "Authorization: Bearer sk_test_0123456789" \
	"http://localhost:8080/v1/customers/123/cash_balance?page=2&pageSize=55"

which will return response with following error response

> GET /v1/customers/123/cash_balance?page=2&pageSize=55 HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/7.65.2
> Accept: */*
> Authorization: Bearer sk_test_0123456789
>
* Mark bundle as not supporting multiuse
< HTTP/1.1 500 Internal Server Error
< Content-Type: application/json
< Mock-Request-Count: 1
< X-Mock-Scenario: stripe-customer-failure
< Stripe-Version: 2018-09-06
< Vary: Origin
< Date: Sat, 29 Oct 2022 17:29:15 GMT
< Content-Length: 15

Dynamic Templates with Mock API Scenarios

You can use loops and conditional primitives of template language and custom functions provided by the API mock library to generate dynamic responses as follows:

method: GET
name: get_devices
path: /devices
description: ""
request:
    match_content_type: "application/json; charset=utf-8"
response:
    headers:
        "Server":
            - "SampleAPI"
        "Connection":
            - "keep-alive"
    content_type: application/json
    contents: >
     {
     "Devices": [
{{- range $val := Iterate .pageSize }}
      {
        "Udid": "{{SeededUdid $val}}",
        "Line": { {{SeededFileLine "lines.txt" $val}}, "Type": "Public", "IsManaged": false },
        "Amount": {{JSONFileProperty "props.yaml" "amount"}},        
        "SerialNumber": "{{Udid}}",
        "MacAddress": "{{Udid}}",
        "Imei": "{{Udid}}",
        "AssetNumber": "{{RandString 20}}",
        "LocationGroupId": {
         "Id": {
           "Value": {{RandNumMax 1000}},
         },
         "Name": "{{SeededCity $val}}",
         "Udid": "{{Udid}}"
        },
        "DeviceFriendlyName": "Device for {{SeededName $val}}",
        "LastSeen": "{{Time}}",
        "Email": "{{RandEmail}}",
        "Phone": "{{RandPhone}}",        
        "EnrollmentStatus": {{SeededBool $val}}
        "ComplianceStatus": {{RandRegex "^AC[0-9a-fA-F]{32}$"}}
        "Group": {{RandCity}},
        "Date": {{TimeFormat "3:04PM"}},
        "BatteryLevel": "{{RandNumMax 100}}%",
        "StrEnum": {{EnumString "ONE TWO THREE"}},
        "IntEnum": {{EnumInt 10 20 30}},
        "ProcessorArchitecture": {{RandNumMax 1000}},
        "TotalPhysicalMemory": {{RandNumMax 1000000}},
        "VirtualMemory": {{RandNumMax 1000000}},
        "AvailablePhysicalMemory": {{RandNumMax 1000000}},
        "CompromisedStatus": {{RandBool}},
        "Add": {{Add 2 1}},
      }{{if LastIter $val $.PageSize}}{{else}},  {{end}}
{{ end }}
     ],
     "Page": {{.page}},
     "PageSize": {{.pageSize}},
     "Total": {{.pageSize}}
     }
    {{if NthRequest 10 }}
    status_code: {{EnumInt 500 501}}
    {{else}}
    status_code: {{EnumInt 200 400}}
    {{end}}
wait_before_reply: {{.page}}s

Above example includes a number of template primitives and custom functions to generate dynamic contents such as:

Loops

GO template support loops that can be used to generate multiple data entries in the response, e.g.

{{- range $val := Iterate .pageSize }}

Builtin functions

GO template supports custom functions that you can add to your templates. The mock service includes a number of helper functions to generate random data such as:

Add numbers

  "Num": "{{Add 1 2}}",

Date/Time

  "LastSeen": "{{Time}}",
  "Date": {{Date}},
  "DateFormatted": {{TimeFormat "3:04PM"}},
  "LastSeen": "{{Time}}",

Comparison

  {{if EQ .MyVariable 10 }}
  {{if GE .MyVariable 10 }}
  {{if GT .MyVariable 10 }}
  {{if LE .MyVariable 10 }}
  {{if LT .MyVariable 10 }}
  {{if Nth .MyVariable 10 }}

Enums

  "StrEnum": {{EnumString "ONE TWO THREE"}},
  "IntEnum": {{EnumInt 10 20 30}},

Random Data

  "SerialNumber": "{{Udid}}",
  "AssetNumber": "{{RandString 20}}",
  "LastSeen": "{{Time}}",
  "Host": "{{RandHost}}",
  "Email": "{{RandEmail}}",
  "Phone": "{{RandPhone}}",
  "URL": "{{RandURL}}",
  "EnrollmentStatus": {{SeededBool $val}}
  "ComplianceStatus": {{RandRegex "^AC[0-9a-fA-F]{32}$"}}
  "City": {{RandCity}},
  "Country": {{RandCountry}},
  "CountryCode": {{RandCountryCode}},
  "Completed": {{RandBool}},
  "Date": {{TimeFormat "3:04PM"}},
  "BatteryLevel": "{{RandNumMax 100}}%",
  "Object": "{{RandDict}}",
  "IntHistory": {{RandIntArrayMinMax 1 10}},
  "StringHistory": {{RandStringArrayMinMax 1 10}},
  "FirstName": "{{SeededName 1 10}}",
  "LastName": "{{RandName}}",
  "Score": "{{RandNumMinMax 1 100}}",
  "Paragraph": "{{RandParagraph 1 10}}",
  "Word": "{{RandWord 1 1}}",
  "Sentence": "{{RandSentence 1 10}}",
  "Colony": "{{RandString}}",

Request count and Conditional Logic

{{if NthRequest 10 }}   -- for every 10th request
{{if GERequest 10 }}    -- if number of requests made to API so far are >= 10
{{if LTRequest 10 }}    -- if number of requests made to API so far are < 10

The template syntax allows you to define a conditional logic such as:

    {{if NthRequest 10 }}
    status_code: {{AnyInt 500 501}}
    {{else}}
    status_code: {{AnyInt 200 400}}
    {{end}}

In above example, the mock API will return HTTP status 500 or 501 for every 10th request and 200 or 400 for other requests. You can use conditional syntax to simulate different error status or customize response.

Loops

  {{- range $val := Iterate 10}}

     {{if LastIter $val 10}}{{else}},{{end}}
  {{ end }}

Variables

     {{if VariableContains "contents" "blah"}}
     {{if VariableEquals "contents" "blah"}}
     {{if VariableSizeEQ "contents" "blah"}}
     {{if VariableSizeGE "contents" "blah"}}
     {{if VariableSizeLE "contents" "blah"}}

Test fixtures

The mock service allows you to upload a test fixture that you can refer in your template, e.g.

  "Line": { {{SeededFileLine "lines.txt" $val}}, "Type": "Public", "IsManaged": false },

Above example loads a random line from a lines.txt fixture. As you may need to generate a deterministic random data in some cases, you can use Seeded functions to generate predictable data so that the service returns same data. Following example will read a text fixture to load a property from a file:

  "Amount": {{JSONFileProperty "props.yaml" "amount"}},

This template file will generate content as follows:

{ "Devices": [
 {
   "Udid": "fe49b338-4593-43c9-b1e9-67581d000000",
   "Line": { "ApplicationName": "Chase", "Version": "3.80", "ApplicationIdentifier": "com.chase.sig.android", "Type": "Public", "IsManaged": false },
   "Amount": {"currency":"$","value":100},
   "SerialNumber": "47c2d7c3-c930-4194-b560-f7b89b33bc2a",
   "MacAddress": "1e015eac-68d2-42ee-9e8f-73fb80958019",
   "Imei": "5f8cae1b-c5e3-4234-a238-1c38d296f73a",
   "AssetNumber": "9z0CZSA03ZbUNiQw2aiF",
   "LocationGroupId": {
    "Id": {
      "Value": 980
    },
    "Name": "Houston",
    "Udid": "3bde6570-c0d4-488f-8407-10f35902cd99"
   },
   "DeviceFriendlyName": "Device for Alexander",
   "LastSeen": "2022-10-29T11:25:25-07:00",
   "Email": "john.smith@abc.com",
   "Phone": "1-408-454-1507",
   "EnrollmentStatus": true,
   "ComplianceStatus": "ACa3E07B0F2cA00d0fbFe88f5c6DbC6a9e",
   "Group": "Chicago",
   "Date": "11:25AM",
   "BatteryLevel": "43%",
   "StrEnum": "ONE",
   "IntEnum": 20,
   "ProcessorArchitecture": 243,
   "TotalPhysicalMemory": 320177,
   "VirtualMemory": 768345,
   "AvailablePhysicalMemory": 596326,
   "CompromisedStatus": false,
   "Add": 3
 },
...
 ], "Page": 2, "PageSize": 55, "Total": 55 }

Artificial Delays

You can specify artificial delay for the API request as follows:

wait_before_reply: {{.page}}s

Above example shows delay based on page number but you can use any parameter to customize this behavior.

Conditional Logic

The template syntax allows you to define a conditional logic such as:

    {{if NthRequest 10 }}
    status_code: {{AnyInt 500 501}}
    {{else}}
    status_code: {{AnyInt 200 400}}
    {{end}}

Test fixtures

The mock service allows you to upload a test fixture that you can refer in your template, e.g.

"Line": { {{SeededFileLine "lines.txt" $val}}, "Type": "Public", "IsManaged": false },

"Amount": {{JSONFileProperty "props.yaml" "amount"}},

This template file will generate content as follows:

{ "Devices": [
 {
   "Udid": "fe49b338-4593-43c9-b1e9-67581d000000",
   "Line": { "ApplicationName": "Chase", "Version": "3.80", "ApplicationIdentifier": "com.chase.sig.android", "Type": "Public", "IsManaged": false },
   "Amount": {"currency":"$","value":100},   
   "SerialNumber": "47c2d7c3-c930-4194-b560-f7b89b33bc2a",
   "MacAddress": "1e015eac-68d2-42ee-9e8f-73fb80958019",
   "Imei": "5f8cae1b-c5e3-4234-a238-1c38d296f73a",
   "AssetNumber": "9z0CZSA03ZbUNiQw2aiF",
   "LocationGroupId": {
    "Id": {
      "Value": 980,
    },
    "Name": "Houston",
    "Udid": "3bde6570-c0d4-488f-8407-10f35902cd99"
   },
   "DeviceFriendlyName": "Device for Alexander",
   "LastSeen": "2022-10-29T11:25:25-07:00",
   "Email": "anthony.christian@abblhfgkpd.edu",
   "Phone": "1-573-993-7542",   
   "EnrollmentStatus": true
   "ComplianceStatus": ACa3E07B0F2cA00d0fbFe88f5c6DbC6a9e
   "Group": Chicago,
   "Date": 11:25AM,
   "BatteryLevel": "43%",
   "StrEnum": ONE,
   "IntEnum": 20,
   "ProcessorArchitecture": 243,
   "TotalPhysicalMemory": 320177,
   "VirtualMemory": 768345,
   "AvailablePhysicalMemory": 596326,
   "CompromisedStatus": false,
   "Add": 3,
   "Dict": map[one:1 three:3 two:2]
 },
...
 ], "Page": 2, "PageSize": 55, "Total": 55 }

Playing back a specific mock scenario

You can pass a header for X-Mock-Scenario to specify the name of scenario if you have multiple scenarios for the same API, e.g.

curl -v -H "X-Mock-Scenario: stripe-cash-balance" -H "Authorization: Bearer sk_test_0123456789" \
	"http://localhost:8080/v1/customers/123/cash_balance?page=2&pageSize=55"

You can also customize response status by overriding the request header with X-Mock-Response-Status and delay before return by overriding X-Mock-Wait-Before-Reply header.

Using Test Fixtures

You can define a test data in your test fixtures and then upload as follows:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/lines.txt \
	http://localhost:8080/_fixtures/GET/lines.txt/devices

curl -v -H "Content-Type: application/yaml" --data-binary @fixtures/props.yaml \
    http://localhost:8080/_fixtures/GET/props.yaml/devices

In above example, test fixtures for lines.txt and props.yaml will be uploaded and will be available for all GET requests under /devices URL path. You can then refer to above fixture in your templates. You can also use this to serve any binary files, e.g. you can define an image template file as follows:

method: GET
name: test-image
path: /images/mock_image
description: ""
request:
response:
    headers:
      "Last-Modified":
        - {{Time}}
      "ETag":
        - {{RandString 10}}
      "Cache-Control":
        - max-age={{RandNumMinMax 1000 5000}}
    content_type: image/png
    contents_file: mockup.png
    status_code: 200

Then upload a binary image using:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/mockup.png \
	http://localhost:8080/_fixtures/GET/mockup.png/images/mock_image

And then serve the image using:

curl -v "http://localhost:8080/images/mock_image"

Custom Functions

The API mock service defines following custom functions that can be used to generate test data:

Numeric Random Data

Following functions can be used to generate numeric data within a range or with a seed to always generate deterministic test data:

Random
SeededRandom
RandNumMinMax
RandIntArrayMinMax

Text Random Data

Following functions can be used to generate numeric data within a range or with a seed to always generate deterministic test data:

RandStringMinMax
RandStringArrayMinMax
RandRegex
RandEmail
RandPhone
RandDict
RandCity
RandName
RandParagraph
RandPhone
RandSentence
RandString
RandStringMinMax
RandWord

Email/Host/URL

RandURL
RandEmail
RandHost

Boolean

Following functions can be used to generate boolean data:

RandBool
SeededBool

UDID

Following functions can be used to generate UDIDs:

Udid
SeededUdid

String Enums

Following functions can be used to generate a string from a set of Enum values:

EnumString

Integer Enums

Following functions can be used to generate an integer from a set of Enum values:

EnumInt

Random Names

Following functions can be used to generate random names:

RandName
SeededName

City Names

Following functions can be used to generate random city names:

RandCity
SeededCity

Country Names or Codes

Following functions can be used to generate random country names or codes:

RandCountry
SeededCountry
RandCountryCode
SeededCountryCode

File Fixture

Following functions can be used to generate random data from a fixture file:

RandFileLine
SeededFileLine
FileProperty
JSONFileProperty
YAMLFileProperty

Generate Mock API Behavior from OpenAPI or Swagger Specifications

If you are using Open API or Swagger for API specifications, you can simply upload a YAML based API specification. For example, here is a sample Open API specification from Twilio:

openapi: 3.0.1
paths:
  /v1/AuthTokens/Promote:
    servers:
    - url: https://accounts.twilio.com
    description: Auth Token promotion
    x-twilio:
      defaultOutputProperties:
      - account_sid
      - auth_token
      - date_created
      pathType: instance
      mountName: auth_token_promotion
    post:
      description: Promote the secondary Auth Token to primary. After promoting the
        new token, all requests to Twilio using your old primary Auth Token will result
        in an error.
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/accounts.v1.auth_token_promotion'
          description: OK
      security:

...


   schemas:
     accounts.v1.auth_token_promotion:
       type: object
       properties:
         account_sid:
           type: string
           minLength: 34
           maxLength: 34
           pattern: ^AC[0-9a-fA-F]{32}$
           nullable: true
           description: The SID of the Account that the secondary Auth Token was created
             for
         auth_token:
           type: string
           nullable: true
           description: The promoted Auth Token
         date_created:
           type: string
           format: date-time
           nullable: true
           description: The ISO 8601 formatted date and time in UTC when the resource
             was created
         date_updated:
           type: string
           format: date-time
           nullable: true
           description: The ISO 8601 formatted date and time in UTC when the resource
             was last updated
         url:
           type: string
           format: uri
           nullable: true
           description: The URI for this resource, relative to `https://accounts.twilio.com`
...

You can then upload the API specification as:

curl -H "Content-Type: application/yaml" --data-binary @fixtures/oapi/twilio_accounts_v1.yaml \
		http://localhost:8080/_oapi

It will generate a mock scenarios for each API based on mime-type, status-code, parameter formats, regex, data ranges, e.g.,

name: UpdateAuthTokenPromotion-xx
path: /v1/AuthTokens/Promote
description: Promote the secondary Auth Token to primary. After promoting the new token, all requests to Twilio using your old primary Auth Token will result in an error.
request:
    match_query_params: {}
    match_headers: {}
    match_content_type: ""
    match_contents: ""
    example_path_params: {}
    example_query_params: {}
    example_headers: {}
    example_contents: ""
response:
    headers: {}
    content_type: application/json
    contents: '{"account_sid":"{{RandRegex `^AC[0-9a-fA-F]{32}$`}}",\
    "auth_token":"{{RandStringMinMax 0 0}}","date_created":"{{Time}}",\
    "date_updated":"{{Time}}","url":"https://{{RandName}}.com"}'
    contents_file: ""
    status_code: 200
wait_before_reply: 0s

In above example, the account_sid uses regex to generate data and URI format to generate URL. Then invoke the mock API as:

curl -v -X POST http://localhost:8080/v1/AuthTokens/Promote

Which will generate dynamic response as follows:

{
  "account_sid": "ACF3A7ea7f5c90f6482CEcA77BED07Fb91",
  "auth_token": "PaC7rKdGER73rXNi6rVKZMN1Jw0QYxPFeEkqyvnM7Ojw2nziOER7SMWkIV6N2hXYTKxAfDMfS9t0",
  "date_created": "2022-10-29T11:54:46-07:00",
  "date_updated": "2022-10-29T11:54:46-07:00",
  "url": "https://Billy.com"
}

Listing all Mock Scenarios

You can list all available mock APIs using:

curl -v http://localhost:8080/_scenarios

Which will return summary of APIs such as:

{
  "/_scenarios/GET/FetchCredentialAws-8b2fcf02dfb7dc190fb735a469e1bbaa3ccb5fd1a24726976d110374b13403c6/v1/Credentials/AWS/{Sid}": {
    "method": "GET",
    "name": "FetchCredentialAws-8b2fcf02dfb7dc190fb735a469e1bbaa3ccb5fd1a24726976d110374b13403c6",
    "path": "/v1/Credentials/AWS/{Sid}",
    "match_query_params": {},
    "match_headers": {},
    "match_content_type": "",
    "match_contents": "",
    "LastUsageTime": 0,
    "RequestCount": 0
  },
  "/_scenarios/GET/FetchCredentialPublicKey-60a01dcea5290e6d429ce604c7acf5bd59606045fc32c0bc835e57ac2b1b8eb6/v1/Credentials/PublicKeys/{Sid}": {
    "method": "GET",
    "name": "FetchCredentialPublicKey-60a01dcea5290e6d429ce604c7acf5bd59606045fc32c0bc835e57ac2b1b8eb6",
    "path": "/v1/Credentials/PublicKeys/{Sid}",
    "match_query_params": {},
    "match_headers": {},
    "match_content_type": "",
    "match_contents": "",
    "LastUsageTime": 0,
    "RequestCount": 0
  },
  "/_scenarios/GET/ListCredentialAws-28717701f05de4374a09ec002066d308043e73e30f25fec2dcd4c3d3c001d300/v1/Credentials/AWS": {
    "method": "GET",
    "name": "ListCredentialAws-28717701f05de4374a09ec002066d308043e73e30f25fec2dcd4c3d3c001d300",
    "path": "/v1/Credentials/AWS",
    "match_query_params": {
      "PageSize": "\\d+"
    },
    "match_headers": {},
    "match_content_type": "",
    "match_contents": "",
    "LastUsageTime": 0,
    "RequestCount": 0
  },
...

Chaos Testing

In addition to serving a mock service, you can also use a builtin chaos client to test remote services for stochastic testing by generating random data based on regex or API specifications. For example, you may capture a test scenario for a remote API using http proxy such as:

export http_proxy="http://localhost:8081"
export https_proxy="http://localhost:8081"
curl -k https://jsonplaceholder.typicode.com/todos

This will capture a mock scenario such as:

method: GET
name: recorded-todos-ff9a8e133347f7f05273f15394f722a9bcc68bb0e734af05ba3dd98a6f2248d1
path: /todos
description: recorded at 2022-12-12 02:23:42.845176 +0000 UTC for https://jsonplaceholder.typicode.com:443/todos
group: todos
predicate: ""
request:
    match_query_params: {}
    match_headers:
        Content-Type: ""
    match_contents: '{}'
    example_path_params: {}
    example_query_params: {}
    example_headers:
        Accept: '*/*'
        User-Agent: curl/7.65.2
    example_contents: ""
response:
    headers:
        Access-Control-Allow-Credentials:
            - "true"
        Age:
            - "19075"
        Alt-Svc:
            - h3=":443"; ma=86400, h3-29=":443"; ma=86400
        Cache-Control:
            - max-age=43200
        Cf-Cache-Status:
            - HIT
        Cf-Ray:
            - 7782ffe4bd6bc62c-SEA
        Connection:
            - keep-alive
        Content-Type:
            - application/json; charset=utf-8
        Date:
            - Mon, 12 Dec 2022 02:23:42 GMT
        Etag:
            - W/"5ef7-4Ad6/n39KWY9q6Ykm/ULNQ2F5IM"
        Expires:
            - "-1"
        Nel:
            - '{"success_fraction":0,"report_to":"cf-nel","max_age":604800}'
        Pragma:
            - no-cache
    contents: |-
      [
        {
          "userId": 1,
          "id": 1,
          "title": "delectus aut autem",
          "completed": false
        },
        {
          "userId": 1,
          "id": 2,
          "title": "quis ut nam facilis et officia qui",
          "completed": false
        },
      ...
        ]
    contents_file: ""
    status_code: 200
    match_headers: {}
    match_contents: '{"completed":"__string__.+","id":"(__number__[+-]?[0-9]{1,10})","title":"(__string__\\w+)","userId":"(__number__[+-]?[0-9]{1,10})"}'
    assertions: []

You can then customize this scenario with additional assertions and you may remove all response contents as they won’t be used. Note that above scenario is defined with group todos. You can then submit a request for chaos testing as follows:

curl -k -v -X POST http://localhost:8080/_chaos/todos -d '{"base_url": "https://jsonplaceholder.typicode.com", "execution_times": 10}'

Above request will submit 10 requests to the todo server with random data and return response such as:

{"errors":null,"failed":0,"succeeded":10}

If you have a local captured data, you can also run chaos client with a command line without running mock server, e.g.:

go run main.go chaos --base_url https://jsonplaceholder.typicode.com --group todos --times 10

Static Assets

The mock service can serve any static assets from a user-defined folder and then serve it as follows:

cp static-file default_assets

# execute the API mock server
make && ./out/bin/api-mock-service

# access assets
curl http://localhost:8080/_assets/default_assets

API Reference

The API specification for the mock library defines details for managing mock scenarios and customizing the mocking behavior.

Summary

Building and testing distributed systems often requires deploying a deep stack of dependent services, which makes development hard on a local environment with limited resources. Ideally, you should be able to deploy and test entire stack without using network or requiring a remote access so that you can spend more time on building features instead of configuring your local environment. Above examples show how you use the https://github.com/bhatti/api-mock-service to mock APIs for testing purpose and define test scenarios for simulating both happy and error cases as well as injecting faults or network delays in your testing processes so that you can test for fault tolerance. This mock library can be used to define the API mock behavior using record/play, template language or API specification standards. I have found a great use of tools like this when developing micro services and hopefully you find it useful. Feel free to connect with your feedback or suggestions.

Comments (0)

May 12, 2022

Applying Laws of Scalability to Technology and People

Filed under: Algorithms,Concurrency,Methodologies,Project Management,Technology — admin @ 10:16 pm

As businesses grow with larger customers size and hire more employees, they face challenges to meet the customer demands in terms of scaling their systems and maintaining rapid product development with bigger teams. The businesses aim to scale systems linearly with additional computing and human resources. However, systems architecture such as monolithic or ball of mud makes scaling systems linearly onerous. Similarly, teams become less efficient as they grow their size and become silos. A general solution to solve scaling business or technical problems is to use divide & conquer and partition it into multiple sub-problems. A number of factors affect scalability of software architecture and organizations such as the interactions among system components or communication between teams. For example, the coordination, communication and data/knowledge coherence among the system components and teams become disproportionately expensive with the growth in size. The software systems and business management have developed a number of laws and principles that can used to evaluate constraints and trade offs related to the scalability challenges. Following is a list of a few laws from the technology and business domain for scaling software architectures and business organizations:

Amdhal’s Law

Amdahl’s Law is named after Gene Amdahl that is used to predict speed up of a task execution time when it’s scaled to run on multiple processors. It simply states that the maximum speed up will be limited by the serial fraction of the task execution as it will create resource contention:

Speed up (P, N) = 1 / [ (1 - P) + P / N ]

Where P is the fraction of task that can run in parallel on N processors. When N becomes large, P / N approaches 0 so speed up is restricted to 1 / (1 – P) where the serial fraction (1 – P) becomes a source of contention due to data coherence, state synchronization, memory access, I/O or other shared resources.

Amdahl’s law can also be described in terms of throughput using:

N / [ 1 + a (N - 1) ]

Where a is the serial fraction between 0 and 1. In parallel computing, a class of problems known as embarrassingly parallel workload where the parallel tasks have a little or no dependency among tasks so their value for a will be 0 because they don’t require any inter-task communication overhead.

Amdah’s law can be used to scale teams as an organization grows where the teams can be organized as small and cross-functional groups to parallelize the feature work for different product lines or business domains, however the maximum speed up will still be limited by the serial fraction of the work. The serial work can be: build and deployment pipelines; reviewing and merging changes; communication and coordination between teams; and dependencies for deliverables from other teams. Fred Brooks described in his book The Mythical Man-Month how adding people to a highly divisible task can reduce overall task duration but other tasks are not so easily divisible: while it takes one woman nine months to make one baby, “nine women can’t make a baby in one month”.

The theoretical speedup of the latency of the execution of a program according to Amdahl’s law (credit wikipedia).

Brooksâ€™s Law

Brooks’s law was coined by Fred Brooks that states that adding manpower to a late software project makes it later due to ramp up time. As the size of team increases, the ramp up time for new employees also increases due to quadratic communication overhead among team members, e.g.

Number of communication channels = N x (N - 1) / 2

The challenge with adding more engineers to a project. Just moving from 3 developers to 4 doubles the number of lines of communication. pic.twitter.com/TkI2NcHVT2
— Rich Rogers (@RichRogers_) August 9, 2019

The organizations can build small teams such as two-pizza/single-threaded teams where communication channels within each team does not explode and the cross-functional nature of the teams require less communication and dependencies from other teams. The Brook’s law can be equally applied to technology when designing distributed services or components so that each service is designed as a loosely coupled module around a business domain to minimize communication with other services and services only communicate using a well designed interfaces.

Universal Scalability Law

The Universal Scalability Law is used for capacity planning and was derived from Amdahl’s law by Dr. Neil Gunther. It describes relative capacity in terms of concurrency, contention and coherency:

C(N) = N / [1 + a(N – 1) + B.N (N – 1) ]

Where C(N) is the relative capacity, a is the serial fraction between 0 and 1 due to resource contention and B is delay for data coherency or consistency. As data coherency (B) is quadratic in N so it becomes more expensive as size of N increases, e.g. using a consensus algorithm such as Paxos is impractical to reach state consistency among large set of servers because it requires additional communication between all servers. Instead, large scale distributed storage services generally use sharding/partitioning and gossip protocol with a leader-based consensus algorithm to minimize peer to peer communication.

The Universal Scalability Law can be applied to scale teams similar to Amdahl’s law where a is modeled for serial work or dependency between teams and B is modeled for communication and consistent understanding among the team members. The cost of B can be minimized by building cross-functional small teams so that teams can make progress independently. You can also apply this model for any decision making progress by keeping the size of stake holders or decision makers small so that they can easily reach the agreement without grinding to halt.

The gossip protocols also applies to people and it can be used along with a writing culture, lunch & learn and osmotic communication to spread knowledge and learnings from one team to other teams.

Little’s Law

Little’s Law was developed by John Little to predict number of items in a queue for stable stable and non-preemptive. It is part of queueing theory and is described mathematically as:

L = A W

Where L is the average number of items within the system or queue, A is the average arrival time of items and W is the average time an item spends in the system. The Little’s law and queuing theory can be used for capacity planning for computing servers and minimizing waiting time in the queue (L).

The Little’s law can be applied for predicting task completion rate in an agile process where L represents work-in-progress (WIP) for a sprint; A represents arrival and departure rate or throughput/capacity of tasks; W represents lead-time or an average amount of time in the system.

WIP = Throughput x Lead-Time

Lead-Time = WIP / Throughput

You can use this relationship to reduce the work in progress or lead time and improve throughput of tasks completion. Little’s law observes that you can accomplish more by keeping work-in-progress or inventory small. You will be able to better respond to unpredictable delays if you keep a buffer in your capacity and avoid 100% utilization.

King’s formula

The King’s formula expands Little’s law by adding utilization and variability for predicting wait time before serving of requests:

{\displaystyle \mathbb {E} (W_{q})\approx \left({\frac {\rho }{1-\rho }}\right)\left({\frac {c_{a}^{2}+c_{s}^{2}}{2}}\right)\tau } — (credit wikipedia)

where T is the mean service time, m (1/T) is the service rate, A is the mean arrival rate, p = A/m is the utilization, c_a is the coefficient of variation for arrivals and c_s is the coefficient of variation for service times. The King’s formula shows that the queue sizes increases to infinity as you reach 100% utilization and you will have longer queues with greater variability of work. These insights can be applied to both technical and business processes so that you can build systems with a greater predictability of processing time, smaller wait time E(W) and higher throughput ?.

Note: See Erlang analysis for serving requests in a system without a queue where new requests are blocked or rejected if there is not sufficient capacity in the system.

Gustafson’s Law

Gustafson’s law improves Amdahl’s law with a keen observation that parallel computing enables solving larger problems by computations on very large data sets in a fixed amount of time. It is defined as:

S = s + p x N

S = (1 – s) x N

S = N + (1 – N) x s

where S is the theoretical speed up with parallelism, N is the number of processors, s is the serial fraction and p is the parallel part such that s + p = 1.

Gustafson’s law shows that limitations imposed by the sequential fraction of a program may be countered by increasing the total amount of computation. This allows solving bigger technical and business problems with a greater computing and human resources.

Conway’s Law

Conway’s law states that an organization that designs a system will produce a design whose structure is a copy of the organization’s communication structure. It means that the architecture of a system is derived from the team structures of an organization, however you can also use the architecture to derive the team structures. This allows defining building teams along the architecture boundaries so that each team is a small, cross functional and cohesive. A study by the Harvard Business School found that the often large co-located teams tended to produce more tightly-coupled and monolithic codebases whereas small distributed teams produce more modular codebases. These lessons can be applied to scaling teams and architecture so that teams and system modules are built around organizational boundaries and independent concerns to promote autonomy and reduce tight coupling.

Pareto Principle

The Pareto principle states that for many outcomes, roughly 80% of consequences come from 20% of causes. This principle shows up in numerous technical and business problems such as 20% of code has the 80% of errors; customers use 20% of functionality 80% of the time; 80% of optimization improvements comes from 20% of the effort, etc. It can also be used to identify hotspots or critical paths when scaling, as some microservices or teams may receive disproportionate demands. Though, scaling computing resources is relatively easy but scaling a team beyond an organization boundary is hard. You will have to apply other management tools such as prioritization, planning, metrics, automation and better communication to manage critical work.

Metcalfeâ€™s Law

The Metcalfeâ€™s law states that if there are N users of a telecommunications network, the value of the network is N². It’s also referred as Network effects and applies to social networking sites.

Number of possible pair connections = N * (N – 1) / 2

Reed’s Law expanded this law and observed that the utility of large networks can scale exponentially with the size of the network.

Number of possible subgroups of a network = 2^N – N – 1

This law explains the popularity of social networking services via viral communication. These laws can be applied to model information flow between teams or message exchange between services to avoid peer to peer communication with extremely large group of people or a set of nodes. A common alternative is to use a gossip protocol or designate a partition leader for each group that communicates with other leaders and then disseminate information to the group internally.

Dunbar Number

The Dunbar’s number is a suggested cognitive limit to the number of people with whom one can maintain stable social relationships. It has a commonly used value of 150 and can be used to limit direct communication connections within an organization.

Wirth’s Law and Parkinson’s Law

The Wirth’s Law is named after Niklaus Wirth who observed that the software is getting slower more rapidly than hardware is becoming faster. Over the last few decades, processors have become exponentially faster as a Moor’s Law but often that gain allows software developers to develop more complex software that consumes all gains of the speed. Another factor is that it allows software developers to use languages and tools that may not generate more efficient code so the code becomes bloated. There is a similar law in software development called Parkinson’s law that work expands to fill the time available for it. Though, you also have to watch for Hofstadter’s Law that states that “it always takes longer than you expect, even when you take into account Hofstadter’s Law”; and Brook’s Law, which states that “adding manpower to a late software project makes it later.”

The Wirth’s Law, named after Niklaus Wirth, posits that software tends to become slower at a rate that outpaces the speed at which hardware becomes faster. This observation reflects a trend where, despite significant advancements in processor speeds as predicted by Moor’s Law , software complexity increases correspondingly. Developers often leverage these hardware improvements to create more intricate and feature-rich software, which can negate the hardware gains. Additionally, the use of programming languages and tools that do not prioritize efficiency can lead to bloated code.

In the realm of software development, there are similar principles, such as Parkinson’s law, which suggests that work expands to fill the time allotted for its completion. This implies that given more time, software projects may become more complex or extended than initially necessary. Moreover, Hofstadter’s Law offers a cautionary perspective, stating, “It always takes longer than you expect, even when you take into account Hofstadter’s Law.” This highlights the often-unexpected delays in software development timelines. Brook’s Law further adds to these insights with the adage, “Adding manpower to a late software project makes it later.” These laws collectively emphasize that the demand upon a resource tends to expand to match the supply of the resource but adding resources later also poses challenges due to complexity in software development and project management.

Principle of Priority Inversion

In modern operating systems, the concept of priority inversion arises when a high-priority process needs resources or data from a low-priority process, but the low-priority process never gets a chance to execute due to its lower priority. This creates a deadlock or inefficiency where the high-priority process is blocked indefinitely. To avoid this, schedulers in modern operating systems adjust the priority of the lower-priority process to ensure it can complete its task and release the necessary resources, allowing the high-priority process to continue.

This same principle applies to organizational dynamics when scaling teams and projects. Imagine a high-priority initiative that requires collaboration from another team whose priorities do not align. Without proper coordination, the team working on the high-priority initiative may never get the support they need, leading to delays or blockages. Just as in operating systems, where a priority adjustment is needed to keep processes running smoothly, organizations must also ensure alignment across teams by managing a global list of priorities. A solution is to maintain a global prioritized list of projects that is visible to all teams. This ensures that the most critical initiatives are recognized and appropriately supported by every team, regardless of their individual workloads. This centralized prioritization ensures that teams working on essential projects can quickly receive the help or resources they need, avoiding bottlenecks or deadlock-like situations where progress stalls because of misaligned priorities.

Load Balancing (Round Robin, Least Connection)

Load balancing algorithms distribute tasks across multiple servers to optimize resource utilization and prevent any single server from becoming overwhelmed. Common strategies include round-robin (distributing tasks evenly across servers) and least connection (directing new tasks to the server with the fewest active connections).

Load balancing can be applied to distribute work among teams or individuals. For instance, round-robin can ensure that tasks are equally assigned to team members, while the least-connection principle can help assign tasks to those with the lightest workload, ensuring no one is overloaded. This leads to more efficient task management, better resource allocation, and balanced work distribution.

MapReduce

MapReduce splits a large task into smaller sub-tasks (map step) that can be processed in parallel, then aggregates the results (reduce step) to provide the final output. In a large project, teams or individuals can be assigned sub-tasks that they can work on independently. Once all the sub-tasks are complete, the results can be aggregated to deliver the final outcome. This fosters parallelism, reduces bottlenecks, and allows for scalable team collaboration, especially for large or complex projects.

Deadlock Prevention (Banker’s Algorithm)

The Banker’s Algorithm is used to prevent deadlocks by allocating resources in such a way that there is always a safe sequence of executing processes, avoiding circular wait conditions. In managing interdependent teams or tasks, it’s important to avoid deadlocks where teams wait on each other indefinitely. By proactively ensuring that resources (e.g., knowledge, tools, approvals) are available before committing teams to work, project managers can prevent deadlock scenarios. Prioritizing resource allocation and anticipating dependencies can ensure steady progress without one team stalling another.

Consensus Algorithms (Paxos, Raft)

Consensus algorithms ensure that distributed systems agree on a single data value or decision, despite potential failures. Paxos and Raft are used to maintain consistency across distributed nodes. In projects involving multiple stakeholders or teams, reaching consensus on decisions can be challenging, especially with different priorities and viewpoints. Consensus-building techniques, inspired by these algorithms, could involve ensuring that key stakeholders agree before any significant action is taken, much like how Paxos ensures agreement across distributed systems. This avoids misalignment and fosters collaboration and trust across teams.

Rate Limiting

Rate limiting controls the number of requests or operations that can be performed in a given timeframe to prevent overloading a system. This concept applies to managing expectations, particularly in teams with multiple incoming requests. Rate limiting can be applied to protect teams from being overwhelmed by too many requests at once. By limiting how many tasks or requests a team can handle at a time, project managers can ensure a sustainable work pace and prevent burnout, much like how rate limiting helps protect system stability.

Summary

Above laws offer strategies for optimizing both technical systems and team dynamics. Amdahl’s Law and the Universal Scalability Law highlight the challenges of parallelizing work, emphasizing the need to manage coordination and communication overhead as bottlenecks when scaling teams or systems. Brook’s and Metcalfe’s Laws reveal the exponential growth of communication paths, suggesting that effective team scaling requires managing these paths to avoid coordination paralysis. Little’s Law and Kingman’s Formula suggest limiting work in progress and preventing 100% resource utilization to ensure reliable throughput, while Conway’s Law underscores the alignment between team structures and system architecture. Teams and their responsibilities should mirror modular architectures, fostering autonomy and reducing cross-team dependencies.

The Pareto Principle can guide teams to make small but impactful changes in architecture or processes that yield significant productivity improvements. Wirth’s Law and Parkinson’s Law serve as reminders to prevent work bloat and unnecessary complexity by setting clear deadlines and objectives. Dunbar’s Number highlights the human cognitive limit in maintaining external relationships, suggesting that team dependencies should be kept minimal to maintain effective collaboration. The consensus algorithms used in distributed systems can be applied to decision-making and collaboration, ensuring alignment among teams. Error correction algorithms are useful for feedback loops, helping teams iteratively improve. Similarly, techniques like load balancing strategies can optimize task distribution and workload management across teams.

Before applying these laws, it is essential to have clear goals, metrics, and KPIs to measure baselines and improvements. Prematurely implementing these scalability strategies can exacerbate issues rather than resolve them. The focus should be on global optimization of the entire organization or system, rather than focusing on local optimizations that don’t align with broader goals.

Comments (0)

March 24, 2022

Architecture Patterns and Practices for Sustainable Software Delivery Pipelines

Filed under: Project Management,Technology — Tags: agile, architecture, complexity, patterns, technical debt — admin @ 10:31 pm

Abstract

Software is eating the world and today’s businesses demand shipping software features at a higher velocity to enable learning at a greater pace without compromising the quality. However, each new feature increases viscosity of existing code, which can add more complexity and technical debt so the time to market for new features becomes longer. Managing a sustainable pace for software delivery requires continuous improvements to the software development architecture and practices.

Software Architecture

The Software Architecture defines guiding principles and structure of the software systems. It also includes quality attribute such as performance, sustainability, security, scalability, and resiliency. The software architecture is then continuously updated through iterative software development process and feedback cycle from the the actual use in production environment. The software architecture decays if itâ€™s ignored that results in a higher complexity and technical debt. In order to reduce technical debt, you can build a backlog of technical and architecture related changes so that you can prioritize along with the product development. In order to maintain consistent architecture throughout your organization, you can document the architecture principles to define high-level guidelines for best practices, documentation templates, review process and guidance for the architecture decisions.

Quality Attributes

Following are major quality attributes of the software architecture:

Availability â€” It defines percentage of time, the system is available, e.g. available-for-use-time / total-time. It is generally referred in percentiles such as P99.99, which indicates a down time of 52 minutes in a year. It can also be calculated in terms of as mean-time between failure (MTBF) and mean-time to recover (MTRR) using MTBF/(MTBF+MTRR). The availability will depend not only on the service that you are providing but also its dependent services, e.g. P-service * P-dep-service-1 * P-dep-service-2. You can improve availability with redundant services, which can be calculated as Max-availability - (100 - Service-availability) ** Redundancy-factor. In order to further improve availability, you can detect faults and use redundancy and state synchronization for fault recovery. The system should also handle exceptions gracefully so that it doesn’t crash or goes into a bad state.
Capacity â€” Capacity defines how the system scales by adding hardware resources.
Extensibility â€” Extensibility defines how the system meets future business requirements without significantly changing existing design and code.
Fault Tolerance â€” Fault tolerance prevents a single point of failure and allows the system to continue operating even when parts of the system fail.
Maintainability â€” Higher quality code allows building robust software with higher stability and availability. This improves software delivery due to modular and loosely coupled design.
Performance â€” It is defined in terms of latency of an operation under normal or peak load. A performance may degrade with consumptions of resources, which affects throughput and scalability of the system. You can measure user’s response-time, throughput and utilization of computational resources by stress testing the system. A number of tactics can be used to improve performance such as prioritization, reducing overhead, rate-limiting, asynchronicity, caching, etc. Performance testing can be integrated with continuous delivery process that use load and stress testing to measure performance metrics and resource utilization.
Resilience â€” Resilience accepts the fact that faults and failure will occur so instead system components resist them by retrying, restarting, limiting error propagation or other measures. A failure is when a system deviates from its expected behavior as a result of accidental fault, misconfigurations, transient network issues or programming error. Two metrics related to resilience are mean-time between failure (MTBF) and mean-time to recover (MTTR), however resilient systems pay more attention to recovery or a shorter MTTR for fast recovery.
Recovery â€” Recovery looks at system recover in relation with availability and resilience. Two metrics related to recovery are recovery point objective (RPO) and recovery time objective (RTO), where RPO determines data that can be lost in case of failure and RTO defines wait time for the system recovery.
Reliability â€” Reliability looks at the probability of failure or failure rate.
Reproducibility â€” Reproducibility uses version control for code, infrastructure, configuration so that you can track and audit changes easily.
Reusability â€” It encourages code reuse to improve reliability, productivity and cost savings from the duplicated effort.
Scalability â€” It defines ability of the system to handle increase in workload without performance degradation. It can be expressed in terms of vertical or horizontal scalability, where horizontal reduces impact of isolated failure and improves workload availability. Cloud computing offers elastic and auto-scaling features for adding additional hardware when higher request rate is detected by the load balancer.
Security â€” Security primarily looks at confidentiality, integrity, availability (CIA) and is critical in building distributed systems. Building secure systems depends on security practices such as a strong identity management, defense in depth, zero trust networks, auditing, ad protecting while data in motion or at rest. You can adopt DevSecOps that shifts security left to earlier in software development lifecycle with processes such as Security by Design (SbD), STRIDE (Spoofing, Tampering, Repudiation, Disclosure, Denial of Service, Elevation of privilege), PASTA (Process for Attack Simulation and Threat Analysis), VAST (Visual, Agile and Simple Threat), CAPEC (Common Attack pattern Enumeration and Classification), and OCTAGE (Operationally Critical Threat, and Vulnerability Evaluation).
Testability â€” It encourages building systems in a such way it’s easier to test them.
Usability â€” It defines user experience of user interface and information architecture.

Architecture Patterns

Following is a list of architecture patterns that help building a high quality software:

Asynchronicity

Synchronous services are difficult to scale and recover from failures because they require low-latency and can easily overwhelm the services. Messaging-based asynchronous communication based on point-to-point or publish/subscribe are more suitable for handling faults or high load. This improves resilience because service components can restart in case of failure while messages remain in the queue.

Admission Control

The admission control adds a authentication, authorization or validation check in front event queue so that service can handle the load and prevent overload when demand exceeds the server capacity.

Back Pressure

When a producer is generating workload faster than the server can process, it can result in long request queues. Back pressure signals clients that servers are overloaded and clients need to slow down. However, rogue clients may ignore these signals so servers often employ other tactics such as admission control, load shedding, rate limiting or throttling requests.

Big fleet in front of small fleet

You should look at all transitive dependencies when scaling a service with a large fleet of hosts so that you don’t drive a large network traffic that needs to invoke dependent services with a smaller fleet. You can use use load testing to find the bottlenecks and update SLAs for the dependent services so that they are aware of network load from your APIs.

Blast Radius

A blast radius defines impact of failure on overall system when an error occurs. In order to limit the blast radius, the system should eliminate a single point of failure, rolling deploy changes using canaries and stop cascading failures using circuit breakers, retry and timeout.

Bulkheads

Bulkheads isolate faults from one component to another, e.g. you may use different thread pool for different workloads or use multiple regions/availability-zones to isolate failures in a specific datacenter.

Caching

Caching can be implemented at a several layers to improve performance such as database-cache, application-cache, proxy/edge cache, pre-compute cache and client-side cache.

Circuit Breakers

The circuit breaker is defined as a state machine with three states: normal, checking and tripped. It can be used to detect persistent failures in a dependent service and trip its state to disable invocation of the service temporarily with some default behavior. It can be later changed to the checking state for detecting success, which changes its state to normal after a successful invocation of the dependent service.

CQRS / Event Sourcing

Command and Query Responsibility Segregation (CQRS) separates read and update operations in the database. It’s often implemented using event-sourcing that records changes in an append-only store for maintaining consistency and audit trails.

Default Values

Default values provide a simple way to provide limited or degraded behavior in case of failure in dependent configuration or control service.

Disaster Recovery

Disaster recovery (DR) enables business continuity in the event of large-scale failure of data centers. Based on cost, availability and RTO/RPO constraints, you can deploy services to multiple regions for hot site; only replicate data from one region to another region while keeping servers as standby for warm site; or use backup/restore for cold site. It is essential to periodically test and verify these DR procedures and processes.

Distributed Saga

Maintaining data consistency in a distributed system where data is stored in multiple databases can be hard and using 2-phase-commit may incur high complexity and performance. You can use distributed Saga for implementing long-running transactions. It maintains state of the transaction and applies compensating transactions in case of a failure.

Failing Fast

You can fail fast if the workload cannot serve the request due to unavailability of resources or dependent services. In some cases, you can queue requests, however it’s best to keep those queues short so that you are not spending resources to serve stale requests.

Function as a Service

Function as a service (FaaS) offers serverless computing to simplify managing physical resources. Cloud vendors offer APIs for AWS Lambda, Google Cloud Functions and Azure Functions to build serverless applications for scalable workloads. These functions can be easily scaled to handle load spikes, however you have to be careful scaling these functions so that any services that they depend on can support the workload. Each function should be designed with a single responsibility, idempotency and shared nothing principles that can be executed concurrently. The serverless applications generally use event-based architecture for triggering functions and as the serverless functions are more granular, they incur more communication overhead. In addition, chaining functions within the code can result in tightly coupled applications, instead use a state machine or a workflow to orchestrate the communication flow. There is also an open source support for FaaS based serverless computing such as OpenFaas and OpenWhisk on top of Kubernetes or OpenShift, which prevents locking into a specific cloud provider.

Graceful Degradation

Instead of failing a request when dependent components are unhealthy, a service may use circuit-breaker pattern to return a predefined or default response.

Health Checks

Health checks runs a dummy or synthetic transaction that performs the action without affecting real data to verify the system component and its dependencies.

Idempotency

Idempotent services completes API request only exactly once so resending same request due to retries has no side effect. Idempotent APIs typically uses a client-generated identifier or token and the idempotent service returns same response if duplicate request is received.

Layered Architecture

The layered architecture separates software into different concerns such as:

Presentation Layer
Business Logic Layer
Service Layer
Domain Model Layer
Data Access Layer

Load Balancer

Load balancer allows distributing traffic among groups of resources so that a single resource is not overloaded. These load balancers also monitors health of servers and you can setup a load balancer for each group of resources to ensure that requests are not routed to unhealthy or unavailable resources.

Load Shedding

Load shedding allows rejection the work at the edge when server side exceeds its capacity, e.g. a server may return HTTP 429 error to signal clients that they can retry at a slower rate.

Loosely coupled dependencies

Using queuing systems, streaming systems, and workflows isolate behavior of dependent components and increases resiliency with asynchronous communication.

MicroServices

Microservices evolved from service oriented architecture (SOA) and support both point to point protocols such as REST/gRPC and asynchronous protocols based on messaging/event bus. You can apply bounded-context of domain-driven design (DDD) to design loosely coupled services.

Model-View Controller

It decouples user interface from the data model and application functionality so that each component can be independently tested. Other variations of this pattern include modelâ€“viewâ€“presenter (MVP) and modelâ€“viewâ€“viewmodel (MVVM).

NoSQL

NoSQL database technology provide support for high availability and variable/write-heavy workloads that can be easily scaled with additional hardware. NoSQL optimizes CAP and PACELC tradeoffs of consistency, availability, partition tolerance and latency, A number of cloud vendors provide managed NoSQL database solutions, however they can create latency issues if services accessing these databases are not colocated.

No Single Point of Failure

In order to eliminate single-points of failures for providing high availability and failover, you can deploy redundant services to multiple regions and availability zones.

Ports and Adapters

Ports and Adapters (Hexagon) separates interface (ports) from implementation (adapters). The business logic is encapsulated in the Hexagon that is invoked by the implementation (adapters) when actors operate on capabilities offered by the interface (port).

Rate Limiting and Throttling

Rate-limiting defines the rate at which clients can access the services based on the license policy. The throttling can be used to restrict access as a result of unexpected increase in demand. For example, the server can return HTTP 429 to notify clients that they can backoff or retry at a slower rate.

Retries with Backoff and Jitter

A remote operation can be retried if it fails due to transient failure or a server overload, however each retry should use a capped exponential backoff so that retries don’t cause additional load on the server. In a layered architecture, retry should be performed at a single point to minimize multifold retries. Retries can use circuit-breakers and rate-limiting to throttle requests. In some cases, requests may timeout for clients but succeed on the server side so the APIs must be designed with idempotency so that they are safe to retry. In order to avoid retries at the same time, a small random jitter can be added with retries.

Rollbacks

The software should be designed with rollbacks in mind so that all code, database schema and configurations can be easily rolled back. A production environment might be running multiple versions of same service so care must be taken to design the APIs that are both backward and forward compatibles.

Stateless and Shared nothing

Shared nothing architecture helps building stateless and loosely decoupled services that can be easily horizontally scaled for providing high availability. This architecture allows recovering from isolated failures and support auto-scaling by shrinking or expanding resources based on the traffic patterns.

Startup dependencies

Upon start of services, they may need to connect to certain configuration or bootstrap services so care must be taken to avoid thundering herd problems that can overwhelm those dependent services in the event of a wide region outage.

Timeouts

Timeouts help building resilient systems by throttling invocation of external services and preventing the thundering herd problem. Timeouts can also be used when retrying a failed operation after a transient failure or a server overload. A timeout can also add a small jitter to randomly spread the load on the server. Jitter can also be applied to timers of scheduled jobs or delayed work.

Watchdogs and Alerts

A watchdogs monitors a system component for a specific action such as latency, traffic, errors, saturation and SLOs. It then send an alert based on the monitoring configuration that triggers an email, on-call paging or an escalation.

Virtualization and Containers

Virtualization allows abstracting computing resources using virtual machines or containers so that you don’t depend on physical implementation. A virtual machine is a complete operating system on top of hypervisors whereas container is an isolated, lightweight environment for running applications. Virtualization allows building immutable infrastructure that are specially design to meet application requirements and can be easily deployed on a variety of hardware resources.

Architecture Practices

Following are best practices for sustainable software delivery:

Automation

Automation builds pipelines for continuous integration, continuous testing and continuous delivery to improve speed and agility of the software delivery. Any kind of operation procedures for deployment and monitoring can be stored in version control and then automatically applied with CI/CD procedures. In addition, automated procedures can be defined to track failures based on key performance indicators and trigger recovery or repair for the erroneous components.

Automated Testing

Automated testing allows building software with a suite of unit, integration, functional, load and security tests that verify the behavior and ensures that it can meet production demand. These automated tests will run as part of CI/CD pipelines and will stop deployment if any of the tests fail. In order to run end-to-end and load tests, the deployment scripts will create a new environment and setup a tests data. These tests may replicate synthetic transactions based on production traffic and benchmark the performance metrics.

Capacity Planning

Using load testing, monitoring production traffic patterns and demand with workload utilization help forecast the resources needed for future growth. This can be further strengthened with a capacity model that calculates unit-price of resources and growth forecast so that you can automate addition or removal of resources based on demand.

Cloud Computing

Adopting cloud computing simplifies resource provisioning and its elasticity allows organizations to grow or shrink those resources based on the demand. You can also add automation to optimize utilization of the resources and reduce costs when allocating more resources.

Continuous Delivery

Continuous delivery automates production deployment of small and frequent changes by developers. Continuous delivery relies on continuous integration that runs automated tests and automated deployment without any manual interventions. During a software development process, a developer picks a feature, works on changes and then commits changes to the source control after peer code-review. The automated build system will run a pipeline to create a container image based on the commit and then deploy it to a test or QA environment. The test environment will run automated unit, integration and regression tests using a test data in the database. The code is then promoted to the main branch and the automated build system tags and build the image on the head commit of main-branch, which that is pushed to the container registry. The pre-prod environment pulls the image, restarts the pre-prod container and runs more comprehensive tests with a larger set of test data in the database including performance tests. You may need multiple stages of pre-prod deployment such as alpha, beta and gamma environments, where each environment may require deployment to a unique datacenter. After successful testing, the production systems are updated with the new image using rolling updates, blue/green deployments or canary deployments to minimize disruption to end users. The monitoring system watches for error rates at each stage of the deployment and automatically rollbacks changes if a problem occurs.

Deploy over Multiple Zones and Regions

In order to provide high availability, compliance and reduced latency, you can deploy to multiple availability zones and regions. Global load balancers can be used to route traffic based on geographic proximity to the closest region. This also helps implementing business continuity as applications can easily failover to another region with minimal data.

Service Mesh

In order to easily build distributed systems, a number of platforms based on service-mesh pattern have emerged to abstract a common set of problems such as network communication, security, observability, etc:

Dapr – Distributed Application Runtime

The Distributed Application Runtime (Dapr) provides a variety of communication protocols, encryption, observability and secret management for building secured and resilient distributed services.

Envoy

Envoy is a service proxy for building cloud native application with builtin support for networking protocols and observability.

Istio service mesh

Istio is built on top of Kubernetes and Envoy to build service mesh with builtin support for networking, traffic management, observability and security. A service mesh also addresses features such as A/B testing, canary deployments, rate limiting, access control, encryption, and end-to-end authentication.

Linkerd

Linkerd is a service mesh for Kubernetes and consists of control-plane and data-plane with builtin support for networking, observability and security. The control-plane allows controlling services and data-plane acts as a sidecar container that handles network traffic and communicate with the control-plane for configuration.

WebAssembly

The WebAssembly is a stack-based virtual machine that can run at the edge or in cloud. A number of WebAssembly platforms have adopted Actor model to build a platform for writing distributed applications such as wasmCloud and Lunatic.

Documentation

The architecture document defines goals and constraints of the software system and provides various perspectives such as use-cases, logical, data, processes, and physical deployment. It also includes non-functional or quality attributes such as performance, growth, scalability, etc. You can document these aspects using standards such as 4+1, C4, and ERD as well as document the broader enterprise architecture using methodologies like TOGAF, Zachman, and EA.

Incident management

Incident management defines process of root-cause analysis and actions that organization can take when an incident occurs affecting production environment. It defines best practices such as clear ownership, reducing time to detect/mitigate, blameless postmortems and prevention measures. The organization can then implement preventing measures and share lessons learned from all operational events and failures across teams. You can also use pre-mortem to identify potential areas that can be improved or mitigated. Another way to simulate potential problems is using chaos engineering or setting up game days to test the workloads for various scenarios and outage.

Infrastructure as Code

Infrastructure as code uses declarative language to define development, test and production environment, which is managed by the source code management software. These provisioning and configuration logic can be used by CI/CD pipelines to automatically deploy and test environments. Following is a list of frameworks for building infrastructure from code:

Azure Resource Manager

Azure cloud offer Azure Resource Manager (ARM) templates based on JSON format to declaratively define the infrastructure that you intend to deploy.

AWS Cloud Development Kit

The Cloud Development Kit (CDK) supports high-level programming languages to construct cloud resources on Amazon Web Services so that you can easily build cloud applications.

Hashicorp Terraform

Terraform uses HCL based configurations to describe computing resources that can be deployed to multiple cloud providers.

Monitoring

Monitoring measures key performance indicators (KPI) and service-level objectives (SLO) that are defined at the infrastructure, applications, services and end-to-end levels. These include both business and technical metrics such as number of errors, hot spots, call graphs, which are visible to the entire team for monitoring trends and reacting quickly to failures.

Multi-tenancy

If your system is consumed by a different groups or tenants of users, you will need to design your system and services so that it isolates data and computing resources for secure and reliable fashion. Each layer of the system can be designed to treat tenant context as a first-class construct, which is tied to the user identity. You can capture usage metrics per tenant to identify bottlenecks, estimate cost and analyze the resource utilization for capacity planning and growth projections. The operational dashboards can also use these metrics to construct tenant-based operational views and proactively respond to unexpected load.

Security Review

In order to minimize the security risk, the development teams can adopt shift-left on security and DevSecOps practices to closely collaborate with the InfoSec team and integrate security review into every phase of the software development lifecycle.

Version Control Systems

Version control systems such as Git or Mercurial help track code changes, configurations and scripts over time. You can adopt workflows such as gitflow or trunk-based development for check-in process. Other common practices include smaller commits, testing code and running static analysis or linters/profiling tools before checkin.

Summary

The software complexity is a major reason for missed deadlines and slow/buggy software. This complexity can be essential complexity within the business domain but it’s often result of accidental complexity as a result of technical debt, poor architecture and development practices. Another source of incidental complexity comes from distributed computing where you need handle security, rate-limiting, observability, etc. that needs to be applied consistently across the distributed systems. For example, virtualization helps building immutable infrastructures and adopting infrastructure as a code; functions as a service simplifies building micro-services; and distributed platforms such as Istio, Linkerd remove a lot of cruft such as security, observability, traffic management and communication protocols when building distributed systems. The goal of a good architecture is to simplify building, testing, deploying and operating a software. You need to continually improve the systems architecture and its practices to build sustainable software delivery pipelines that can meet both current and future demands of users.

Comments (0)

March 24, 2021

Task Scheduling Algorithms

Filed under: Computing,Technology — admin @ 8:16 pm

A task is defined as a unit of work and a task scheduler is responsible for selecting best task for execution based on various criteria. The criteria for selecting best scheduling policy includes:

Maximize resource utilization such as CPU or Memory but not exceeding a certain limit such as 80%.
Maximize throughput such as tasks per seconds.
Minimize turnaround or wall time from task submission to the completion time.
Minimize waiting time in ready queue before executing the task.

Above criteria assumes that pending tasks will be stored in a bounded ready-queue before execution so that no new tasks will be stored when the queue reaches the maximum capacity. In this context, the task is more coarse grained and executes to the completion as compared to CPU scheduling that may use preemptive or cooperative scheduling with context switching to dispatch different processes for execution.

Following is a list of common scheduling algorithms:

First-Come First-Serve (FCFS)

This algorithm simply uses FIFO queue to process the tasks in the order they are queued. On the downside, a long-running task can block other tasks and yield very large average wait times in the queue.

Shortest Job-First (SJF)

This algorithm picks smallest job that needs to be done and then next smallest job. It results in best performance, however it may not be possible to predict runtime of a job before the execution. You may need to pass hints about the job runtime or use historical data to predict the job runtime with exponential average such as:

NewEstimatedRuntime = alpha * PreviousActualRuntime + (1.0 - alpha) * PreviousEstimatedRuntime

where alpha is between 0.0 and 1.0 and represents a weighting factor for the relative importance of recent data compared to older data, e.g. alpha with 0 value ignores previous actual time and 1.0 ignores past history of estimates.

Priority Scheduling

This algorithm picks highest priority job that needs to be done and then next highest priority job. It may result in starvation of low-priority tasks and they may wait indefinitely. You can fix this drawback by supporting aging where priority of old tasks is slowly bumped.

Earliest Deadline

This algorithm gives highest priority to the task with the earliest deadline, which is then used to pick the next task to execute. This scheduler improves resource utilization based on the estimated runtime and data requirements.

Speculative Scheduler

The speculative scheduler detects slow running task and starts another instance of the task as a backup to minimize the response time. This generally works for short jobs with homogeneous resources but it doesn’t guarantee complete reliability.

Multilevel Queues

This algorithm allows categorizing tasks and then dispatching the task to the queue based on the category, e.g. you may define distinct queues for small/medium/large tasks based on runtime or definer queues for interactive/background/system/batch tasks. In some cases, a task may need to be jumped from one queue to another based on runtime characteristics of the task and you can use aging and priority inversion to promote low-priority tasks.

Resource aware scheduler

This scheduler tracks provisioned resources and their utilization required by the tasks. The scheduler may simply allocate resources they need when scheduling the tasks or use admission-control algorithm to prevent certain tasks to start until the required resources are available.

Matchmaking scheduler

This scheduler uses affinity based scheduling so that it routes the tasks to workers or nodes where the data resides locally and provide greater data locality.

Delay scheduler

This scheduler waits until the data required by the task is available on the node or worker where task will run to improve data locality. However, in order to prevent starvation, it allows scheduling tasks to another worker or node.

Capacity scheduler

This scheduler shares system resources among tasks where tasks specify the required resources such as memory and CPU. The scheduler tracks resource utilization and allocates resources specified by the tasks.

Fair scheduler

This scheduler ensures that common resources such as CPU, memory, disk and I/O are shared fairly among different tasks.

References

Comments (0)

Older Posts »

Shahzad Bhatti Welcome to my ramblings and rants!

January 1, 2024

Lessons from “Tidy First?”

1. Guard Clauses

2. Dead Code

3. Normalize Symmetries

4. New Interface, Old Implementation

5. Reading Order

6. Cohesion Order

7. Move Declaration and Initialization Together

8. Explaining Variables

9. Explaining Constants

10. Explicit Parameters

11. Chunk Statement

12. Extract Helper

13. One Pile

14. Explaining Comments

15. Delete Redundant Comments

16. Separate Tidying

17. Chaining

18. Batch Size

19. Rhythm

20. Getting Untangled

21. First, After, Later, Never

22. Beneficially Relating Elements

23. Structure and Behavior

24. Economics: Time Value and Optionality

25. A Dollar Today > A Dollar Tomorrow

26. Options

27. Options Versus Cash Flow

28. Reversible Structure Changes

29. Coupling

30. Constantine’s Equivalence

31. Coupling Versus Decoupling

32. Cohesion

33. Conclusion

August 26, 2023

Modern Software Development Process

1. Goals

2. Elements of Software Development Process

3. Roles in Modern Processes

4. Phases of Modern SDLC

4.1 Inception Phase

4.1.1 Objectives

4.1.2 Practices

4.1.3 Roles and Responsibilities:

4.2. Planning Phase

4.2.1 Objectives

4.2.2 Practices:

4.2.3 Roles and Responsibilities:

4.2.4 Key Deliverables:

4.3. Design Phase

4.3.1 Objectives:

4.3.2 Practices:

4.3.3 Roles and Responsibilities:

4.3.4 Key Deliverables:

4.4 Development Phase

4.4.1 Objectives:

4.4.2 Practices:

4.4.3 Roles and Responsibilities:

4.4.4 Key Deliverables:

4.4.5 Best Practices for Development:

4.5. Testing Phase

4.5.1 Objectives:

4.5.2 Practices:

4.5.3 Types of Testing and Responsibilities:

4.5.4 Key Deliverables:

4.5.5 Best Practices for Testing:

4.6. Deployment Phase

4.6.1 Objectives:

4.6.2 Key Components and Practices:

4.6.3 Roles and Responsibilities:

4.6.4 Key Deliverables:

4.6.5 Best Practices:

4.7 Maintenance Phase

4.7.1 Objectives:

4.7.2 Key Components and Practices:

4.7.3 Roles and Responsibilities:

4.7.4 Key Deliverables:

4.7.5 Best Practices for Maintenance: