Building Microservices

Overview

Building Microservices (2021, Second Edition) by Sam Newman is the comprehensive guide to designing, building, and operating systems using the microservice architectural style. Newman covers the entire lifecycle from decomposition strategies to production operations.

The second edition has been substantially updated to reflect the maturing microservice landscape, with new chapters on Kubernetes, serverless, observability, and the reconciliation of microservices with monoliths.

---

Key Takeaways

1. Microservices are not a free lunch. They introduce complexity in testing, deployment, networking, and data management.

2. Domain-driven design is the foundation. Service boundaries should align with business domain boundaries, not technical layers.

3. Choose integration technology carefully. REST, gRPC, and messaging each have different trade-offs for coupling, performance, and evolvability.

4. Data decentralization is essential. The database-per-service pattern prevents tight coupling through shared schemas.

5. Observability is non-negotiable. Without centralized logging, metrics, and distributed tracing, microservices are unmanageable.

6. Testing microservices requires strategy. Move from a testing pyramid to a testing trophy — emphasize contract and integration tests over unit tests.

7. Organizational alignment matters. Conway's Law means your architecture will mirror your communication structures.

8. Start with a monolith. Extract microservices as you discover natural boundaries rather than predicting them upfront.

9. Kubernetes changes everything. Container orchestration simplifies deployment but adds operational complexity.

10. Idempotency is your friend. Design operations to be safe to retry, and design for partial failure.

---

Who Should Read

| Reader Type | Why | |---|---| | Software architects | Framework for decomposition and integration decisions | | Senior backend engineers | Practical patterns for building distributed systems | | Technical leads | Insight into organizational and technical trade-offs | | CTOs and VPs of Engineering | Understanding the costs and benefits before committing | | Platform engineers | Guidance on infrastructure for microservice deployments |

---

Who Should Skip

• Junior developers with less than 2 years of experience
• Teams of fewer than 5 developers — the complexity outweighs benefits
• Anyone building a simple CRUD application with a single database

---

Related Books

| Book | Author | Connection | |---|---|---| | Designing Data-Intensive Applications | Martin Kleppmann | Distributed systems foundations for microservices | | The Twelve-Factor App | Adam Wiggins | Best practices for cloud-native applications | | Monolith to Microservices | Sam Newman | Migration patterns for existing systems | | Accelerate | Nicole Forsgren | Metrics for measuring delivery performance | | Site Reliability Engineering | Beyer et al. | Operating reliable distributed systems |

---

Final Verdict

The definitive practical guide to microservices. Newman balances enthusiasm with caution, clearly laying out when microservices help and when they hurt. The second edition is significantly better than the first.

Rating: 8.5/10 — Essential reading for anyone considering or already using microservices.

The Microservice Mindset

Microservices are independently deployable services modeled around business domains. Each service owns its data, runs in its own process, and communicates via well-defined APIs.

graph TD
    subgraph Monolith["Monolithic Architecture"]
        UI["User Interface"]
        BL["Business Logic"]
        DB["Database"]
        UI --> BL --> DB
    end

    subgraph Microservices["Microservice Architecture"]
        S1["Service A<br/>(Orders)"]
        S2["Service B<br/>(Payments)"]
        S3["Service C<br/>(Shipping)"]
        S4["Service D<br/>(Notifications)"]
        GW["API Gateway"]
        GW --> S1
        GW --> S2
        GW --> S3
        GW --> S4
    end

---

Service Decomposition

Newman recommends using bounded contexts from Domain-Driven Design to identify service boundaries.

| Decomposition Strategy | Approach | Best For | |----------------------|----------|----------| | Business capability | Map to business functions | Stable domains | | Subdomain | Follow DDD bounded contexts | Complex domains | | Volatility | Separate by change frequency | Rapidly evolving features |

---

Inter-Service Communication

flowchart LR
    subgraph Synchronous["Synchronous"]
        R["REST / gRPC"]
        R -->|"Request/Response"| S["Service"]
    end

    subgraph Asynchronous["Asynchronous"]
        M["Message Broker"]
        P["Producer"] --> M
        M --> C["Consumer 1"]
        M --> C2["Consumer 2"]
    end

---

Data Management

The database-per-service pattern is central to microservices:

graph TD
    subgraph Service_Data["Data per Service"]
        S1["Order Service"] --> DB1["Order DB"]
        S2["Payment Service"] --> DB2["Payment DB"]
        S3["Shipping Service"] --> DB3["Shipping DB"]
    end

    subgraph Patterns["Data Sharing Patterns"]
        P1["Saga Pattern<br/>(eventual consistency)"]
        P2["CQRS<br/>(read vs write models)"]
        P3["Event Sourcing<br/>(event store)"]
    end

---

Observability

| Pillar | Tool Examples | What It Answers | |--------|--------------|-----------------| | Logging | ELK Stack, Loki | What happened? | | Metrics | Prometheus, Datadog | When did it happen? | | Tracing | Jaeger, Zipkin | Where did it happen? |

---

Testing Strategy

Newman advocates moving from the testing pyramid to a testing trophy:

1. Unit tests — fast, isolated, high coverage
2. Contract tests — verify service interactions
3. Integration tests — test against real dependencies
4. End-to-end tests — critical paths only

---

Deployment and Containers

graph LR
    subgraph Pipeline["CI/CD Pipeline"]
        C["Commit"] --> B["Build"]
        B --> T["Test"]
        T --> P["Package (Docker)"]
        P --> D["Deploy"]
    end

    subgraph Orchestration["Orchestration"]
        D --> K["Kubernetes"]
        K --> N1["Pod 1"]
        K --> N2["Pod 2"]
        K --> N3["Pod 3"]
    end

---

Reading Guide

| Chapter | Topic | Est. Time | Priority | |---------|-------|-----------|----------| | 1-2 | Microservice fundamentals | 2h | Essential | | 3-4 | Decomposition and DDD | 2.5h | Essential | | 5-6 | Communication patterns | 3h | Essential | | 7-8 | Data and transactions | 2.5h | Essential | | 9-10 | Testing and deployment | 3h | Important | | 11-13 | Monitoring and security | 2h | Important | | 14-16 | Organizational and migration | 2h | Important |

Strengths

• Comprehensive coverage. Covers the entire microservice lifecycle from decomposition to production operations.
• Pragmatic approach. Newman acknowledges when microservices are not the right choice, unlike many evangelists.
• Second edition improvements. Significantly updated with Kubernetes, serverless, and observability content.
• Real-world examples. Draws on experience from ThoughtWorks projects across many organizations.
• Organizational perspective. Includes Conway's Law, team structures, and migration strategies.
• Clear writing. Technical concepts are explained with clarity.

---

Weaknesses

• Length can be daunting. At 612 pages, the book is comprehensive but could be tighter.
• Some sections are shallow. Topics like service mesh, event sourcing, and CQRS deserve deeper treatment.
• Kubernetes focus may date. The emphasis on Kubernetes as the deployment platform may not age well.
• Light on code. The book is architectural — few concrete code examples.

---

Criticism

The "Microservice Hype" Critique

Some critics argue that the book overstates the benefits of microservices without adequately emphasizing the costs. Organizations without significant scale often take on unnecessary complexity.

The "Too ThoughtWorks" Critique

The book draws heavily on ThoughtWorks' experiences, which some critics argue is not representative of most engineering organizations.

The "Second Edition Still Not Enough" Critique

While substantially updated, some topics like WebAssembly, eBPF, and edge computing are not covered.

---

Comparison with Similar Books

| Book | vs. Building Microservices | |------|---------------------------| | Monolith to Microservices (Newman) | Migration-specific companion | | Designing Data-Intensive Applications (Kleppmann) | Data and distributed systems depth | | Team Topologies (Skelton & Pais) | Organizational design focus | | Accelerate (Forsgren) | Metrics for DevOps effectiveness |

---

Scientific Grounding

| Concept | Source | |---------|--------| | Conway's Law | Melvin Conway (1968) | | DDD Bounded Contexts | Eric Evans (2003) | | CAP Theorem | Eric Brewer (2000) | | Twelve-Factor App | Adam Wiggins (2011) |

---

Final Assessment

| Dimension | Rating | Notes | |-----------|--------|-------| | Depth | 8/10 | Broad but not always deep | | Breadth | 9/10 | Covers all major topics | | Readability | 8/10 | Clear and practical | | Practical Utility | 8/10 | Actionable guidance | | Lasting Value | 7/10 | Some parts will age | | Overall | 8.0/10 | Essential microservices reference |