What are the patterns that ensure reliability in microservices?

To ensure the reliability of microservices, several patterns are used:

• Circuit Breaker — prevents repeated calls to a non-responsive service to avoid cascading failures. If the service does not respond, the Circuit Breaker opens, and requests are not sent to it, instead an error or fallback response is returned immediately.

• Retry — in case of temporary errors, requests are retried with a delay and a limit on the number of attempts.

• Timeout — limits the waiting time for a response from the service to avoid resource blocking.

• Bulkhead — isolates resources for different services or threads so that a failure in one does not affect others.

• Fallback — provides an alternative response or behavior if the main service is unavailable.

• Idempotency — ensures that repeated calls to an operation do not change the system state additionally.

• Health Checks and Circuit Monitoring — regular checks of service status and automatic switching to backup options.

• Event Sourcing and CQRS — for resilient data storage and processing, aiding recovery after failures.

These patterns are often combined and implemented using libraries (e.g., Hystrix, Resilience4j) and infrastructural tools to enhance microservice fault tolerance.