Glossary

Headroom (latency)

Latency headroom is the gap between your measured end-to-end latency and your SLA or SLO ceiling. A service running at 85 ms against a 100 ms SLO has 15 ms of headroom — 15%. Headroom is not wasted capacity; it is a buffer that absorbs unpredictable spikes from GC pauses, connection pool exhaustion, noisy neighbours in shared infrastructure, and traffic bursts.

Formula: Rule of thumb: maintain at least 20% headroom at p99 under peak load; below 10%, spikes often breach the SLO.

Last updated: March 2026