Understanding Cache Penetration, Cache Breakdown, and Cache Avalanche in Redis

During a weekend interview the author asked a candidate about three common Redis cache problems—cache penetration, cache breakdown, and cache avalanche—and then explained each concept in detail.

Cache Penetration

Cache penetration occurs when a request queries a key that is absent both in the cache and the database, causing every request to bypass the cache and hit the database directly. The solution is to cache null values for nonexistent keys with an appropriate expiration time, and to use a Bloom filter to pre‑filter obviously invalid keys.

Cache Breakdown

Cache breakdown (or cache stampede) happens when a hot key expires and a large number of concurrent requests simultaneously query the database, overwhelming it. Mitigation strategies include asynchronous periodic refresh of hot keys before they expire and employing a mutex lock so that only the first request loads the data while others wait.

Cache Avalanche

Cache avalanche refers to a scenario where many keys expire at the same time or the cache service crashes, causing a sudden surge of database traffic that can lead to overload or failure. Preventive measures involve assigning different expiration times to different keys, regularly refreshing them, and deploying the cache in a clustered architecture to avoid single‑point failures.

Additional techniques such as using Bloom filters for request validation, setting appropriate TTLs, and scaling out with multiple cache nodes are recommended to maintain system stability.