Error Budgets Explained

An error budget is the amount of unreliability a service is allowed within its SLO. If the SLO is 99.9% of requests succeeding over 30 days, the budget is the remaining 0.1%.

Why it matters

Error budgets turn reliability from a debate into a shared, measurable constraint. Product and engineering agree in advance how much risk is acceptable, and the data decides when to slow down.

Using the budget

• Budget healthy: ship features and run experiments at normal pace
• Budget burning fast: alert on burn rate, not single errors
• Budget exhausted: prioritize reliability work and slow risky changes until it recovers

Burn-rate alerting

Alerting on how quickly the budget is being consumed, over both short and long windows, catches serious incidents quickly while avoiding pages for brief, harmless blips.

Make the policy explicit

Write down what happens at each budget state and who decides. A policy agreed before an incident is far easier to follow than one negotiated during it.