Microservices Governance: Establishing Standards and Best Practices for Microservices Development

Roman Glushach
12 min readOct 11, 2023
Microservices Principles

Microservices are a popular software architecture that aims to deliver large, complex applications as a collection of small, independent, and loosely coupled services. Each service is responsible for a single piece of domain logic and communicates with other services using language-agnostic APIs. Microservices enable faster development, easier maintenance, and greater scalability than traditional monolithic applications.

Benefits

  • Autonomy: Each service can be developed, tested, deployed, and updated independently by a small team of developers
  • Loose coupling: Services are loosely coupled and communicate through well-defined interfaces
  • Reuse: Services can be reused across different applications or domains, increasing the efficiency and consistency of software development
  • Fault tolerance: Services are designed to handle failures gracefully and recover quickly
  • Composability: Services can be composed together to create new functionality or applications
  • Discoverability: Services can be easily discovered and registered using service discovery mechanisms, such as DNS or service registries

Challenges

  • Complexity: Microservices increase the complexity of the system as a whole, as there are more moving parts and interactions to manage
  • Operational overhead: Microservices require more operational resources and skills to deploy and run
  • Cultural shift: Microservices require a cultural shift in the way software is developed and delivered

Principles

Single concern

It should do one thing and do it well. This principle helps to achieve high cohesion and low coupling among services. Cohesion refers to how well the elements of a service are related to each other and contribute to a single purpose. Coupling refers to how much a service depends on or affects other services.

A high-cohesive and low-coupled service is easier to understand, maintain, test, and reuse. It also reduces the risk of introducing errors or inconsistencies when making changes or adding new features.

To identify the single concern of a microservice, you can use techniques such as domain-driven design (DDD), which helps you model your system based on the business domain and its concepts. DDD also helps you define clear boundaries between different domains or subdomains and assign them to different services.

High Cohesion and Low Coupling

Cohesion refers to how closely related the functions of a service are. A service should have high cohesion, meaning that it should do one thing and do it well. A service should not have multiple responsibilities or depend on other services for its functionality.

Low coupling refers to how much a service knows about or depends on other services. A service should have low coupling, meaning that it should not have direct or indirect dependencies on other services. A service should only communicate with other services through well-defined APIs and avoid sharing data or state.

High cohesion and low coupling help us achieve several benefits:

  • Easier development and testing: Each service can be developed and tested independently, without requiring the whole system to be up and running
  • Faster deployment and scaling: Each service can be deployed and scaled independently, without affecting other services or requiring coordination
  • Improved fault isolation and resilience: Each service can handle its own failures, without impacting other services or causing cascading failures

Discrete Boundaries

A service boundary is the scope of the functionality and data that a service owns and exposes. A service boundary should be aligned with a business domain or a subdomain, following the Domain-Driven Design (DDD) approach. DDD is a way of modeling complex systems based on the business context and the language used by the domain experts.

Defining discrete boundaries for each service help avoid cross-functional dependencies and ensure that each service has a clear purpose and responsibility. It also help enforce data consistency and integrity within each service, by applying the principle of data encapsulation. Data encapsulation means that each service should own and manage its own data, without exposing it to other services or allowing them to modify it directly. Instead, other services should use the APIs provided by the service to access or manipulate its data.

Discrete boundaries also help achieve better performance and scalability, by allowing each service to use the most suitable technology stack for its needs. For example, one service might use a relational database for storing structured data, while another service might use a NoSQL database for storing unstructured data. This way, each service can optimize its data storage and processing according to its requirements.

Keep it Simple, Stupid

The “Keep it Simple, Stupid” principle emphasizes the importance of keeping microservices simple and focused on their core functionality. This principle acknowledges that complexity can creep into services, and that it is important to actively work to avoid unnecessary complexity.

The benefits of keeping microservices simple include:

  • Improved understandability: Simple services are easier to understand, making it easier for developers to work with them
  • Reduced maintenance costs: Simple services are easier to maintain, as there is less code to maintain and fewer potential issues to address
  • Improved scalability: Simple services are easier to scale, as they have fewer dependencies and are easier to distribute

Techniques for keeping microservices simple include:

  • Focusing on the core functionality of the service and avoiding unnecessary features
  • Using simple design patterns, such as the Single Responsibility Principle
  • Avoiding complex dependencies and ensuring that services are loosely coupled

Organized Around Business Capabilities

Microservices architecture should be organized around the business capabilities that each service provides. This means that each service should be designed to perform a specific business function, and should be responsible for all the components and functionality needed to perform that function.

Event-Driven

A common way of implementing communication between microservices is using an event-driven approach. Event-driven means that instead of making synchronous requests or calls to other services, a service publishes events to notify other services about changes in its state or data. Other services can subscribe to these events and react accordingly, without blocking or waiting for a response.

Event-driven communication has several advantages over synchronous communication:

  • Decoupling: Services don’t need to know about each other’s existence or location. They only need to know about the events they are interested in
  • Scalability: Services can handle events asynchronously, without affecting their performance or availability
  • Resilience: Services can handle failures gracefully, by using techniques such as retries, timeouts, circuit breakers, or dead letter queues
  • Extensibility: Services can easily add new functionality or features by subscribing to new events or publishing new events

However, event-driven communication also introduces some challenges:

  • Complexity: Services need to handle partial failures, eventual consistency, duplicate events, ordering issues
  • Testing: Services need to test their behavior in different scenarios involving multiple events and sources
  • Monitoring: Services need to monitor their event streams and ensure that they are processed correctly and timely

To overcome these challenges, we need to use appropriate tools and frameworks that support event-driven communication, such as message brokers (Apache Kafka), event buses (Azure Service Bus), event sourcing (EventStore), or stream processing (Apache Spark).

Fault Tolerance

Microservices are inherently distributed systems, which means that they are prone to failures. Failures can occur due to various reasons, such as network issues, hardware failures, software bugs, malicious attacks, etc. Therefore, we need to design our microservices to be fault-tolerant, meaning that they can handle failures gracefully and continue to provide service as much as possible.

Some of the techniques that we can use to achieve fault tolerance are:

  • Retry: If a service fails to communicate with another service or an external system, it can retry the operation a few times, with a delay or a backoff strategy, before giving up
  • Timeout: If a service doesn’t receive a response from another service or an external system within a specified time limit, it can abort the operation and return an error or a default value
  • Circuit Breaker: If a service detects that another service or an external system is failing repeatedly or is overloaded, it can stop sending requests to it and redirect them to a fallback service or a cache, until the original service recovers
  • Bulkhead: If a service has multiple dependencies or tasks, it can isolate them into separate threads or processes, so that if one of them fails or consumes too much resources, it does not affect the others
  • Fail Fast: If a service detects that it cannot perform its function or fulfill its SLA (service level agreement), it can fail fast and return an error or a degraded response, instead of wasting resources or causing delays

Composability

Composability is the ability to combine multiple services to create new functionality or features. Composability is essential for microservices, as it enables us to reuse existing services and avoid duplication of code and logic. Composability also allows us to deliver value faster and more frequently, by adding new services or modifying existing ones without affecting the whole system.

To achieve composability, we need to follow some guidelines:

  • Design for reuse: We should design our services with reuse in mind, by providing clear and consistent APIs, documentation, and examples. Follow common standards and conventions for naming, versioning, and security
  • Design for interoperability: We should design our services to be interoperable with other services and systems, by using common protocols, formats, and data models. Should also avoid vendor lock-in and use open source technologies whenever possible
  • Design for extensibility: We should design our services to be extensible, by allowing other services to customize or enhance their behavior or functionality. Should also expose hooks or events that other services can use to integrate with our services

Discoverability

Discoverability is the ability to find and access the services that are available in the system. Discoverability is crucial for microservices, as it enables us to locate and consume the services that we need, without hard-coding their addresses or configurations. Discoverability also facilitates dynamic scaling and load balancing of services, by allowing us to add or remove instances of services without affecting other services.

To achieve discoverability, we need to use some mechanisms:

  • Service Registry: A service registry is a central repository that stores information about the services in the system, such as their names, locations, statuses, metadata, etc. A service registry can be implemented using tools such as Consul, Eureka, or ZooKeeper
  • Service Discovery: Service discovery is a process that allows a service to query the service registry and obtain the information about other services that it needs. Service discovery can be implemented using tools such as Ribbon, Feign, or Envoy
  • Service Mesh: A service mesh is a layer of infrastructure that manages the communication between services. A service mesh can provide features such as service discovery, load balancing, routing, security, monitoring, etc. A service mesh can be implemented using tools such as Istio, Linkerd, or Consul Connect

Decentralized Data Management

In a microservices architecture, each service is responsible for its own data management. This means that each service can use its own database or data storage technology, and can manage its own data independently of other services.

Black Box Requirements

Each service in a microservices architecture should have a well-defined interface that specifies its functionality and behavior. This interface should be treated as a black box, meaning that the implementation details of the service are hidden from other services. This allows for changes to be made to individual services without affecting the entire system.

Platform Independence

Microservices architecture should be platform-independent, meaning that each service can be developed and deployed on a different platform or technology stack. This allows for the use of different programming languages, frameworks, and databases for each service, and enables the system to take advantage of the best tools and technologies for each specific business capability.

Scaling

Microservices architecture should be designed to scale horizontally, meaning that individual services can be scaled up or down independently of other services. This allows for the system to scale to meet changing demands and to handle increased traffic or load.

Use Machine Learning and AI

Machine learning and AI are becoming increasingly important in modern software development, and microservices architecture is no exception. Microservices architecture emphasizes the importance of using machine learning and AI to improve the functionality and efficiency of services.

The benefits of using machine learning and AI include:

  • Improved predictive capabilities: Machine learning and AI can be used to make predictions about user behavior, system performance, and other important metrics
  • Improved automation: Machine learning and AI can be used to automate decision-making, reducing the need for human intervention and improving the efficiency of services
  • Improved anomaly detection: Machine learning and AI can be used to detect anomalies in system behavior, improving the ability to identify and respond to issues

Techniques for using machine learning and AI include:

  • Using machine learning frameworks like TensorFlow or PyTorch to build predictive models
  • Using AI algorithms like decision trees or clustering to automate decision-making

Continuous Delivery

Microservices architecture should be designed to support continuous delivery, meaning that changes to individual services can be developed, tested, and deployed independently of other services. This allows for faster time-to-market and enables the system to respond quickly to changing business needs.

Monitoring and Logging

Microservices architecture requires a different approach to monitoring and logging. Each service should be monitored and logged independently, and there should be a centralized monitoring and logging system that can monitor and log all services.

Security

Microservices architecture should be designed with security in mind. Each service should be secured independently, and there should be a centralized security system that can manage security for all services.

API-First Development

API-first development is an essential principle in microservices architecture. It emphasizes the importance of designing and implementing APIs before developing the services that implement them. This approach ensures that the APIs are well-defined, stable, and meet the needs of the services that will consume them.

The benefits of API-first development include:

  • Better API design: By designing the API before implementing the services, you can ensure that the API is well-structured, easy to use, and meets the needs of the services that will consume it
  • Reduced coupling: API-first development helps reduce coupling between services, as the API serves as a contract that defines the interface between services
  • Improved service quality: By focusing on the API first, you can ensure that each service is designed to meet the needs of the API, resulting in higher-quality services

Techniques for API-first development include:

  • Designing APIs using API definition languages like OpenAPI, GraphQL, or gRPC
  • Using API mocking tools to simulate the behavior of services that have not yet been implemented
  • Creating API documentation that clearly defines the API’s endpoints, parameters, and response formats

Design for Failure

Principle acknowledges that failures will occur and emphasizes the importance of designing services that can handle these failures gracefully.

The benefits of designing for failure include:

  • Improved reliability: By designing services that can handle failures, you can increase the overall reliability of the system
  • Reduced downtime: Designing for failure helps reduce downtime, as services can continue to operate even when one or more services fail
  • Better fault tolerance: Designing for failure enables services to tolerate faults and continue operating, even when failures occur

Techniques for designing for failure include:

  • Implementing redundancy: Use redundancy to ensure that there are backup instances of services that can take over when a failure occurs
  • Using failover mechanisms: Implement failover mechanisms that automatically switch to a backup instance when a failure is detected
  • Implementing circuit breakers: Use circuit breakers to detect when a service is not responding and prevent further requests from being sent to it

Embrace Imperative Programming

Imperative programming is a programming paradigm that focuses on describing how a program should execute, rather than just what it should do. Microservices architecture emphasizes the importance of embracing imperative programming, as it enables services to be designed that are easier to understand, test, and maintain.

The benefits of imperative programming include:

  • Improved readability: Imperative programming makes code easier to read, as it describes the steps that the program should take to achieve a specific outcome
  • Better testability: Imperative programming makes it easier to write tests for services, as the focus is on the steps that the service should take, rather than the outcome
  • Improved maintainability: Imperative programming makes it easier to maintain services, as the focus is on the steps that the service should take, rather than the outcome

Techniques for imperative programming include:

  • Focusing on the steps that the service should take, rather than the outcome
  • Using descriptive variable names and function names to make the code easier to read

Governance

Microservices architecture should be governed by a set of policies and procedures that ensure consistency and quality. This includes policies for design, development, testing, deployment, and maintenance, as well as procedures for monitoring and logging, security, and scalability.

Continuous Improvement

Microservices architecture should be designed with continuous improvement in mind. This means that each service should be designed to be flexible and adaptable, and should be regularly reviewed and improved to meet changing business needs.

Conclusion

Microservices are a powerful way to design software applications that are modular, scalable, resilient, and adaptable. However, they also require careful planning, design, implementation, and management to ensure their success.

--

--

Roman Glushach

Senior Software Architect & Engineer Manager at Freelance