Monolith to Microservices: Modularizing Fineract 1.x
summary
Proposes splitting a ~300000 LOC monolith into published jars across 12 top-level areas (portfolio has 29 modules) to enable targeted deployments and easier upstream contributions. Shows concrete optimizations: swap to Hikari connection pool, use Tomcat NIO2, run Zulu JVM with low-latency GC, tune JVM params, fix costly SQL and offload sync command processing to Apache Kafka with ~99% backwards compatibility. Describes running separate instances for REST, background jobs and reporting, using read-replicas, sharding and PostgreSQL+EclipseLink experiments.
AI-written summary · about
chapters (7)
- 0:00Session introduction
- 1:38Performance considerations
- 5:25Optimization techniques
- 11:00Architecture challenges
- 13:31Modularization proposal
- 27:18Scaling and microservices
- 36:52Community roadmap
related talks
23mMicroservices within a Monolith
59mMajestic Modular Monoliths
30mMonolith-ifying perfectly good microservices
57mModularizing the Monolith
12mTen Things I Hate About Microservices
52mMonoliths to Modular: Navigating the Transition
50mImplement the monolith - the one well modularized
50mMajestic Modular Monoliths