1 · The big picture¶
One repo, three moving parts. The whole project exists to teach how a stateless HTTP service talks to a stateful dependency — and what breaks when you scale or failover.
| Component | Code | Job |
|---|---|---|
| Frontend | frontend/ |
React app served by nginx; proxies /api to the backend so the browser sees one origin |
| Backend | backend/ |
Express service — the only thing that touches Redis |
| Redis | redis:7-alpine |
Holds every key; in-memory, no volume, single replica |
| Docs | docs/ |
This site, plus embedded Swagger UI that calls the API directly from your browser |
The one mental model¶
Remember this
Backend replicas are cattle — kill one and nothing is lost. Redis is a pet — kill it and all data is gone. There is no volume, no replication, no persistence anywhere in this project.
/healthand/readyexist so an orchestrator can tell "process alive" apart from "Redis reachable" — covered in step 5./api/inforeports which region / cluster / pod answered, which is what makes multi-region demos visible — covered in step 8.
Not production
No auth, no TLS, unrestricted CORS, errors leak internals, and the Redis URL is logged at startup. Trusted networks and synthetic data only.
Next: 2 · Run it locally