$linuxjunkies
>

error budget

also: error rate budget, availability budget, failure budget

A defined tolerance for failures or errors in a system, often expressed as a percentage of uptime or acceptable downtime, used in Site Reliability Engineering (SRE) to balance reliability with feature velocity.

An error budget is a metric that quantifies how much downtime or how many errors a service can tolerate while still meeting its Service Level Objective (SLO). For example, if a service commits to 99.9% uptime annually, the error budget is the remaining 0.1%, which translates to approximately 8.76 hours of acceptable downtime per year.

Once an error budget is defined, teams can spend it strategically. High-risk deployments, infrastructure changes, or feature experiments can proceed knowing they have a buffer before violating the SLO. When the budget is depleted (errors exceed the threshold), teams must shift focus to stability and reliability fixes rather than new features.

This framework encourages engineering teams to balance innovation with stability. Instead of treating all errors as equally catastrophic, error budgets acknowledge that some failures are acceptable and can be learning opportunities, while still enforcing accountability to users and business goals.

Related terms