Let's say we are building a shopping application. What happens if the recommendation microservice is down? How does the application react? Does it completely crash? Or will it let the customer shop? These kinds of situations happen more often as we adopt the microservices architecture.

As we make the services small, the chance that a service goes down increases. How the application reacts to these situations becomes an important question. In the previous example, a fault-tolerant application would show some default recommendations, while letting the customer shop.

As we move into microservices architecture, applications should be more fault-tolerant. Applications should be able to provide toned-down behavior when services are down.