Circuit breaker

This pattern is used to prevent cascading failure in your distributed system. If a service is failing, it can impact in other services causing bigger problems. The idea behind this patter is to automatically detect when a service is failing. You can either stop sending requests to the service, have a default fallback or return an error.

After a certain period of time, circuit breaker will allow a limited number of request to see if the service is recovered. If it is, all traffic is allowed again. If it is not, circuit breaker will be enabled again and the process is repeated.

Metrics are important to monitor the circuit breaker. Tools like prometheus or grafana can help you to identify anomalies in your system.