Skip to content
Latchkey

What Is a Latency Budget?

A latency budget is a target ceiling on how long an end-to-end operation may take, allocated across the components and stages that make it up. If a request must complete in 300ms, each hop, database, service, and network, gets a slice of that budget. Exceeding any slice threatens the overall objective.

Why it matters

Without a budget, each component optimizes in isolation and the end-to-end experience drifts past acceptable limits. Allocating latency explicitly makes trade-offs visible and ties component-level performance back to a user-facing SLO. The same idea applies to CI: a pipeline can carry a budget for time-to-feedback.

Related concepts

  • Allocated across stages of an operation
  • Ties component perf to a user-facing SLO
  • Often expressed against a p95 or p99 target

Related guides

Tired of flaky CI? Latchkey auto-heals failed jobs and retries them for you. Start free →