liveBook · Manning

Chapter 4: Design for evolution: architecture that survives change

What you will learn in chapter 4 of Software Architecture Decision-Making

The chapter introduction explains why architecture must remain effective as a system’s context changes, rather than merely preserve its initial design.

After this chapter you will be able to

What it builds on

Section by section

Introduction

The chapter introduction explains why architecture must remain effective as a system’s context changes, rather than merely preserve its initial design.

Key terms: good-enough architecture, assumptions, context, big ball of mud, spaghetti architecture, architectural awareness

4.1 Systems never sit still

This section shows why systems must be able to respond to ongoing, sometimes unexpected change rather than being treated as finished.

Key terms: evolution, boxed products, change pressures, likelihood, impact, current readiness, owner, Hyrum’s Law, observable behavior, formal contract

4.2 Evolution as a first-class architectural quality

The section defines evolvability as a necessary architectural quality and gives five practical ways to judge whether a system can change safely.

Key terms: evolvability, good-enough architecture, modularity, boundaries, coupling, cohesion, replaceability, stable contracts, ownership, consumers

4.3 Feedback loops and sensing change

This section shows how lightweight feedback loops reveal architectural rigidity early enough for teams to decide deliberately whether to defer, adapt, or replace a design.

Key terms: change signals, architectural rigidity, feedback, dependency updates, automated tests, observability, rollback plans, feedback loops, sensing mechanisms, architectural drift

4.4 Designing with future change in mind

This section explains how to prepare for plausible future changes without overbuilding and how to time architectural decisions at the last responsible moment.

Key terms: plausible change, speculative change, seam, stable contract, deferred risk, last responsible moment, cost of waiting, cost of reversal, uncertainty, delivery risk

4.4.1 Avoiding premature commitment

This subsection explains how to avoid hard-to-reverse architecture decisions before they are necessary, while recognizing when waiting creates delivery risk.

Key terms: last responsible moment, premature commitment, vendor lock-in, real delivery risk, smaller decision, seam, adapter, feature flag, versioned contract

4.4.2 Designing seams

This section explains how deliberately placed seams contain change and give an architecture room to evolve without adding unnecessary indirection.

Key terms: seams, stable interfaces, unnecessary indirection, façade, adapter, third-party API, blast radius, messaging, queue, timing and lifecycle

4.4.3 The Strangler Fig pattern

The section explains how the Strangler Fig pattern replaces a legacy system incrementally through seams, reducing the risk of a single large rewrite.

Key terms: Strangler Fig pattern, legacy system, seams, gradual traffic shift, risk control, big-bang replacement, routing points, integration boundaries, database ownership boundaries

4.4.4 Architecture decision records (ADR)

This subsection explains how lightweight architecture decision records preserve the reasoning behind decisions and help future teams revisit them when circumstances change.

Key terms: Architecture Decision Records, ADRs, context, alternatives, trade-offs, review triggers, revisit conditions, modular monolith, lightweight record, architecture documentation

4.5 Keeping architecture soft, not rigid

This section explains how to recognize architectural softness through ordinary changes and why an evolvable system should permit deliberate replacement when needed.

Key terms: architectural softness, stable boundary, tight coupling, cognitive load, evolutionary architecture, premature rigidity, hidden coupling, controlled risk, big-bang rescue

4.5.1 Modular monolith vs. distributed monolith

This section explains why distributing a system too early can create rigidity, and when a modular monolith is a better choice than microservices.

Key terms: microservices, distributed monolith, independent team ownership, responsibility boundaries, modular monolith, clear boundaries, distributed architecture

4.5.2 Avoiding accidental complexity

The section distinguishes unavoidable problem complexity from design-created complexity and explains why architects should challenge complexity that makes change harder without solving a real need.

Key terms: essential complexity, accidental complexity, context boundaries, evolution paralysis, architectural softness

4.5.3 Team maturity matters

This section explains why architectural complexity must be introduced at a pace the team can operate and evolve, with stakeholder communication and team development supporting that path.

Key terms: team maturity, operational and cognitive demands, operational fragility, sequencing, milestones, capability growth, delivery risk, training, mentoring, pairing

4.5.4 Backward compatibility and versioning

This section explains how backward-compatible contracts and deliberate versioning let producers and consumers change at different speeds without forcing coordinated releases.

Key terms: backward compatibility, producers, consumers, contract, schema validation, contract tests, versioned test suites, deprecation policies, versioned APIs, metadata-enveloped messages

4.5.5 Event-driven vs. request-driven evolvability

This section explains when request-driven or event-driven design better supports change, and why loose coupling depends on operational clarity as well as structure.

Key terms: request-driven design, OrderPlaced event, producer, consumers, seam, topology, ownership, contracts, failure modes, operational implications

4.6 Cloud-specific evolution traps

This section explains how cloud services can speed delivery while creating costs, operational burdens, and constraints on future change if adopted without deliberate evaluation.

Key terms: managed services, architectural decisions, provisioning choices, cost amplification, organizational readiness, managed-service sprawl, Infrastructure-as-Code drift, serverless binding lock-in, reproducibility

4.7 Designing for incremental replaceability

The section explains how to prepare an architecture for controlled, incremental replacement and how to review where that flexibility is worth preserving.

Key terms: contracts, boundaries, design seams, selective replaceability, compatibility, incremental replaceability, Strangler Fig pattern, redirection, substitution, incremental retirement

4.8 Evolution as collaboration

Evolutionary architecture depends on team conditions and everyday habits that make change safe to discuss, learn from, and carry out.

Key terms: evolvability, psychological safety, shared understanding, candid feedback, assumptions, trade-offs, evidence, incidents, ADRs, decision context

4.9 The architect’s role in evolutionary systems

This section explains how architects use experience and feedback to guide adaptable systems, while leadership shapes whether those systems have the conditions to evolve.

Key terms: evolving risk landscape, boundaries, long-term coherence, adaptability, experience, feedback, pattern recognition, plausible futures, leadership, ownership

4.10 Summary

Concepts explained in this chapter

← Chapter 3: Good enough architecture vs. perfect architecture

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.