Reliability and Availability
Measure uptime with SLIs, SLOs and nines, then survive failures with retries, backoff, idempotency keys and circuit breakers.
What reliable means
Reliability is the system doing the right thing over time. Availability is the share of time it is up and answering. Everything fails eventually: disks, networks, deploys, other teams' services. Design assumes failure and limits the damage.
Three terms that people mix up:
| Term | Meaning | Example |
|---|---|---|
| SLI (Service Level Indicator) | A measured number | 99.95% of requests returned non-5xx in the last 30 days |
| SLO (Service Level Objective) | The internal target for an SLI | 99.9% success over 30 days |
| SLA (Service Level Agreement) | A contract with penalties if missed | Refund credits if monthly uptime falls below 99.9% |
The SLA is looser than the SLO so you have room to react before you owe money.
1 of 8