Distributed monolith
Services that look separate but have to be changed, tested and deployed together, so you pay for microservices and still have a monolith.
The usual signs: a feature needs changes in four services released in a fixed order, services read each other’s database tables, one service down takes the rest with it, and the test environment only works with every service at the same version. The network adds latency and failure modes, and the coupling stays.
It often comes from splitting a big ball of mud along technical lines (a “user service”, a “notification service”) instead of along bounded contexts, or from sharing a database. The cure is fewer, larger services with their own data (database per service) and contracts checked in CI with consumer-driven contracts. Sometimes the honest fix is to merge back into a modular monolith.
Related terms
- Microservices: An architecture where a system is built from small, separately deployed services, each owning its data and talking to the others over the network.
- Modular monolith: One deployable application split inside into modules with clear boundaries, each owning its data and using the others only through defined interfaces.
- Big ball of mud: A system with no visible structure: everything depends on everything, so every change is risky and nobody can explain where a rule lives.
- Database per service: Each service owns its data and no other service reads its tables directly; the others go through its API or its events.
- Consumer-driven contracts (contract testing): Each consumer of an API writes down what it actually uses, and the provider’s build checks every change against those expectations before release.
