Capacity and reliability
This page lists practical scaling and reliability considerations.
Scaling dimensions
| Component | Scales with |
|---|---|
| API | request rate, payload size, auth verification cost |
| workers | webhook volume, background jobs, aggregation workload |
| Postgres | write rate (events, commits, ledger), read rate (catalog, reports) |
Reliability practices
- Use bounded retries and stable idempotency keys.
- Monitor rate limiting and treat sustained 429 as a capacity signal.
- Keep Postgres connection pooling and limits tuned to avoid saturation.
- Prefer incremental rollout and canaries for application changes.
Next
- Monitoring: Monitoring and alerting