Evidence tension

When does architectural decomposition improve a system, and when does it become the constraint?

Architecture transitions are often discussed as maturity upgrades. The evidence instead suggests that decomposition can solve real constraints while introducing new technical and organisational obligations.

Evidence confidence: Moderate

What the evidence shows

Decomposition can solve demonstrated architecture constraints
Peer-reviewed decision-support research argues that major architecture transition should be based on evidence about the actual system and the constraints that migration is intended to solve.
EVD-0057
Transition introduces new operational and data risks
A real industrial decomposition case found that incremental microservice adoption, data transition and the new distributed architecture created different categories of incidents while modularisation of the remaining monolith also delivered important benefits.
EVD-0061

Do not represent the evidence as showing that monoliths or microservices are generally superior.

How the evidence compares

What can be compared

The sources support the idea that architecture transition should be evaluated as a trade-off against current constraints and target operating requirements.

What cannot be directly compared

The evidence does not provide a universal numerical ranking of monoliths, modular monoliths and microservices.

Reasons for the difference

System coupling
Workload
Data model
Team structure
Operational maturity
Service boundaries
Transition strategy
Business continuity requirements

Lornets interpretation

Architecture is not a maturity ladder. The relevant question is whether changing a particular boundary creates enough benefit to justify the transition cost and the new obligations it creates.

Remaining uncertainty

Architecture research is highly contextual and much industrial evidence remains case-based. Findings from one migration should inform hypotheses and decisions rather than become universal rules.

Evidence needed

More comparative longitudinal studies following architecture transitions across similar organisations, including cost, reliability, delivery and maintainability outcomes.