site-reliability

To calculate error budgets for distributed systems, you define a Service Level Objective (SLO) for availability and subtract the achieved uptime from the target. This establishes a measurable policy for prioritizing reliability work and managing outages.

Dependency arithmetic calculates true system availability by multiplying the reliability percentages of synchronous dependencies. This reveals hidden reliability risks where multiple dependent services degrade the overall system availability below acceptable SLO targets.

You should implement circuit breakers and retry budgets when managing load during partial outages to prevent cascading failures. These strategies ensure graceful degradation and maintain service availability by shedding excess load.

The best way to define SLOs for capacity planning is analyzing your architecture and dependency chain to set measurable availability targets. This translates abstract reliability goals into concrete engineering constraints for load management.

Yes, SLO error budgeting directly supports incident prevention by providing a quantitative threshold for risk tolerance. When error budgets are exhausted, it triggers policies to halt feature deployments and prioritize reliability improvements.