In distributed systems, failures are expected, not rare. Hardware fails, networks partition, software has bugs, and dependencies become unavailable. A single slow dependency, like a payment service timing out during peak traffic, can leave orders stuck mid-checkout and turn a small problem into a user-facing outage.

What separates a resilient system from a fragile one is how it contains failure: it keeps the blast radius small, preserves correctness, and protects the most important user flows.