Q: A Redis cache server is placed in front of a slow database. During a major marketing campaign launch, the cache suddenly purges its keys, and the database instantly crashes under the thundering herd of traffic. What is this phenomenon called, and how do you mitigate it?

This is called a Cache Stampede (or Thundering Herd). It happens when a highly popular cached key expires. Instantly, thousands of concurrent requests hit the cache, find a miss, and *all* of them simultaneously bypass the cache and query the database to regenerate the exact same data, crashing the database.

• Adding Jitter: Instead of setting a strict 1-hour expiration on keys, add random jitter (e.g., 60 minutes + random 0 to 5 minutes) so keys expire staggeringly, preventing massive simultaneous purges.
• Mutex Locks: When a cache miss occurs, the first thread acquires a distributed lock (e.g., Redis SETNX) for that specific key. It proceeds to query the database. The other 999 concurrent threads fail to get the lock, and sleep/poll for 50ms until the first thread populates the cache for them.

Mitigation strategies: