← All patternsSystem architecture · Architecture pattern
Microservices
Choose independently deployable business boundaries and explicit service-owned state.
The problem
Different teams need to evolve inventory and checkout independently, but shared tables and long call chains couple releases and failures.
When it fits
Business boundaries and team ownership justify the network, consistency and operational cost of independent services.
The whole architecture
Architecture diagram
A gateway routes to business services with owned data, asynchronous integration and shared operational capabilities.
Blue arrows are API/data calls; purple arrows are events. Operational bars apply across the architecture.
Scroll to pan. Select a component for its role.
API / data call
Event / job flow
Select a component, or focus it and press Enter, to read its role.
The gateway and messaging are architectural choices, rather than requirements for every microservice system.
Operational capabilities are cross-cutting and do not forward business requests.
The practice canvas pools app capacity; this diagram does not add service-owned scoring roles.
Walk through the design
Follow one operation
Select a step to highlight its participants. This conceptual walkthrough is separate from challenge scoring.
Each service calls its own data store. The service-to-service edge represents a contract; SQL nodes in the practice canvas remain shards, not automatic service ownership.
Step 1 of 3
Accept a checkout command
Checkout owns the command and its local state; inventory owns availability decisions.
Trade-offs to make explicit
Independent deployment requires compatible contracts and operational ownership, rather than simply drawing more app boxes.
Service-owned state makes local invariants clear while cross-service operations need a coordination and recovery policy. A modular monolith can remain a simpler choice.
Failure modes to plan for
A slow dependency can exhaust a caller’s resources without timeouts and bounded concurrency.
Writing another service’s tables bypasses its invariants. A chain of successful local writes is not one cross-service ACID transaction.
Before you build
Explain your decision
Which boundary deserves independent deployment, and what would fail if its owner became unavailable?
Use Plan & estimate inside practice to record your assumptions and trade-offs.
These links open the original challenges. Existing round unlocks, completions and points apply.
One provider stalled; the others still deliver
Related practice in dependency isolation. The canvas pools app capacity and does not score per-service ownership, app-to-app traffic or independent deployments.
Original PractiseGrid explanations and conceptual diagrams, with primary references for further study. Dedicated scored exercises can be added as coverage grows.