Capacity and cost planning

Measure before changing defaults. The starter’s baseline targets roughly 100k registered users, not 100k concurrent users.

Capacity worksheet

Copy this table into the deployment record and replace every placeholder with observed peak and projected values.

Signal Current peak 12-month projection Limit/SLO Headroom Action trigger
edge requests/s         sustained >70% for 15m
p95 latency         error-budget burn
concurrent WebSockets         >70% room/process cap
PostgreSQL connections         >70% pool/server max
database size + monthly growth         restore exceeds RTO
Redis memory         >70% maxmemory
stream ingress and max lag         oldest job exceeds SLO
worker jobs/s         lag grows for 10m
outbound email/webhooks         provider quota >70%
backup size and restore duration         RPO/RTO missed

Load-test the expected traffic mix using scripts/k6/performance.js; do not extrapolate a single endpoint benchmark to the whole platform.

Monthly cost worksheet

Provider prices vary, so the repository records quantities and formulas rather than stale currency claims.

Environment Compute Database/storage Redis Egress/providers Observability/backup Formula
dev developer machine hours local volumes local test mail only local retention hours × internal workstation rate
lab service/app container hours lab GB-month lab GB-hour test traffic short retention sum(resource quantity × provider unit price)
UAT prod-shaped instance hours UAT GB + backup UAT node acceptance traffic production-like retention measured monthly quantities × current price sheet
demo scheduled instance hours demo GB + backup demo node stakeholder traffic bounded retention UAT formula × active-time ratio
prod redundant service hours primary GB + WAL + off-host backup HA capacity real mail/webhooks/egress logs, metrics, traces, uptime quantity × contracted unit price + 20% contingency

Record currency, region, tax, discounts, commitment term, pricing date, owner, and source URL next to the completed estimate. Review quarterly and before any topology, retention, or traffic change.