Graceful Degradation and Fallbacks

1. The Question the Mechanisms Cannot Answer

Everything in this module so far controls how our system treats itself. When to stop waiting. When to stop retrying. When to stop calling a sick dependency at all.

Now suppose every one of those mechanisms works perfectly. The recommendation service goes down.

The timeout fires in 800 ms. The breaker opens, meaning we stop sending calls to the broken service for a while. Not one thread is wasted.

And the user gets... what?

If the answer is an error page, all that work just delivered the failure faster