how do you design for services that degrade gracefully

we been thinkin alot about what happens when one of our critical microservices has an outage or really bad perf issues. beyond basic retries, what patterns or strategies do you use to make sure the overall user experience stays as functional as possible? any specific libraries or frameworks you find helpful for implementin things like circuit breakers, bulkheads, or fallback mechanisms? curious to hear real-world examples of how this has played out in your systems