Distributed Monolith Detection And Remediation
Use When
- Multiple services require lockstep releases
- Synchronous call chains frequently fail under load
Workflow
- Audit coupling hotspots: synchronous depth, shared schemas, release dependencies.
- Identify anti-patterns (shared DB, chatty RPC, implicit contracts).
- Define target decoupling strategy by dependency class.
- Introduce contract versioning and async boundaries where needed.
- Track dependency reduction and release independence metrics.
Required Checks
- Lockstep deployment paths are identified and prioritized
- Shared database write paths have migration plan
- New boundaries reduce runtime blast radius
Decision Framework
- Measure coupling by release independence, synchronous call depth, and shared mutable state—not folder layout.
- Prioritize breaking synchronous chains before splitting services; async boundaries reduce lockstep deploy pressure.
- Extract along bounded contexts with anti-corruption layers; avoid shared-database writes across services.
- Track decoupling metrics release-over-release (deploy coupling index, blast-radius score).
Common Rationalizations And Rebuttals
- "We already have microservices." -> Shared DBs and chatty sync RPC still form a distributed monolith.
- "Big-bang decomposition is faster." -> Big-bang cutovers fail on hidden dependencies; extract incrementally with strangler patterns.
- "Team structure can follow later." -> Conway's law will re-couple services; align ownership with extraction slices.
Evidence Pack
- Coupling heatmap: sync call graph depth, shared schema writes, co-deploy frequency
- Target boundary map with anti-corruption layer design per extraction slice
- Contract stabilization plan with versioning policy for extracted services
- Before/after deploy independence metrics with baseline and target
Exit Criteria
- Architectural coupling trends downward release-over-release
- Teams can ship high-priority services independently