Stop the loop

The brief

Modelled on Facebook's 2010 outage, scaled down. 450 requests a second ride on a warm cache, and the database sees only a trickle. From second 20 to 30 the sim purges (flushes) the whole cache again and again, standing in for every client throwing away a value it thought was invalid. Unprotected, a database sized for the warm cache is flooded, falls over and never comes back. Keep the database standing so the cache can refill and the site recovers on its own.

What it checks

1. 1Choose a mitigation: request coalescing, jittered TTLs or early refresh — or a brake (rate limit, circuit breaker, load shedding) in front of the database.

Brush up first

• Cache stampede

More practice

1. Keep it aliveOpen8 min
2. Launch dayFree with an account20 min