reliability-strategy-builder

A circuit breaker tracks failure counts and transitions between closed, open, and half-open states. When failures exceed a threshold such as 5, it opens and rejects requests until a reset timeout, typically 60 seconds, allows a trial request.

Wrap the operation in a loop that catches errors and waits baseDelay times 2 to the power of the attempt number before retrying. Cap the maximum retries, for example 3, and rethrow the error once exhausted.

Retries handle transient failures by re-attempting immediately with backoff, while circuit breakers stop all requests to a persistently failing service to prevent cascading failures. They are commonly combined, with retries inside a circuit breaker.

Define SLOs as measurable targets such as 99.9% availability over a 30-day window using ratio or percentile measurements. The error budget is 100% minus the SLO, so 99.9% availability allows about 43.2 minutes of downtime per month.

Use bulkheads when you need to isolate workloads so one failing or overloaded operation cannot exhaust shared resources. Separate thread pools or semaphores for critical, standard, and background tasks prevent resource contention.

Fallbacks return degraded or stale data, such as cached values or minimal placeholder objects, so they trade accuracy for availability. They should be a last resort after retries and must be monitored to avoid masking persistent primary failures.