DevOps · interactive
SLIs, SLOs and error budgets
No service is 100% up. An SLO picks a target, like 99.9% of minutes healthy over 30 days, and whatever is left over is the error budget you're allowed to spend. The same four incidents, against three targets.
Budget left, day by daybudget 43.2 min · used 52.0 min
day 1day 30
- Day 4: Bad deploy, rolled back12 min
- Day 11: Database failover6 min
- Day 17: Expired TLS certificate25 min
- Day 23: Cache stampede9 min
Budget gone on day 23: freeze risky releases
99.9% over 30 days allows 43.2 min of downtime. Measured availability was 99.880%.
SLO target
1 2 Each extra nine costs a lot more and leaves ten times less room. Pick what users actually need.
99.99% is 4.3 minutes a month.
3 4
In production
Tools and technologies
SLOs are built into most observability platforms.
Not tied to one language — use these names when you search docs or design a system.