What is cache stampede, and how can it be prevented?

Question

Khushi Singh · Answer

Multi-user requests create a cache stampede when the cache contains missing or expired data at the same time. The simultaneous multiple requests reach the backend database or external data source causing extreme load and possible system downtime alongside performance reduction. Such cache stampedes frequently occur when numerous users simultaneously require the same data that is frequently accessed.

The following strategies should be used to stop cache stampede:

Distributed lock systems use mutex implementation to guarantee one request accesses and refreshes the cache when a miss occurs allowing other requests to wait for updated data.
Stale data will be made available for users as a background operation maintains the cache updates in an asynchronous manner. The database remains protected against multiple simultaneous requests which otherwise could cause an overload condition.
A request collapsing mechanism collects identical requests from various users before a single backend data retrieval occurs so the response data is shared among them.
Probabilistic Early Expiration makes the cache refreshes itself early by having various requests refresh the cache at different times which reduces the chance of multiple cache expirations at once.
Multiple levels of caching can spread cache expiration impact through a system that includes in-memory content along with CDN resources and database query storage.
Regular jobs scheduled to refresh cache content before expiry ensure that latest data will be available in the cache when new requests arrive.

These implementation techniques enable businesses to deliver ongoing high application performance and protect databases from overload which results in uninterrupted user experiences under traffic surges.

forum

What is cache stampede, and how can it be prevented?

Can you answer this question?

1 Answers