Caches

A cache is a storage layer that holds copies of data from a slower primary store to reduce access time. They exploit temporal and spatial locality to serve many requests quickly rather than fetching from the slower source.

In hardware, CPU caches are small, fast memory near the processor, typically organized into levels such as L1, L2, and L3. L1 is fastest and smallest, often split into instruction and data caches. Caches use properties such as inclusivity or exclusivity, write-back or write-through policies, and set associativity with cache lines. In multi-core systems, cache coherence protocols (such as MESI) maintain consistency across caches.

A cache hit occurs when requested data is found in the cache; a miss requires fetching from the next slow tier. Misses are classified as compulsory, capacity, or conflict. Performance is summarized by the hit rate and miss penalties, and depends on the replacement policy (LRU, LFU, FIFO) and cache size.

Software caches store results of expensive operations or web resources to improve performance. Examples include memoization, database query caches, and web caches such as browser caches, proxies, and content delivery networks. HTTP caching uses headers like Cache-Control, Expires, ETag, and Last-Modified to control freshness and validation.

Caching requires managing data validity and coherence. Invalidation, expiration, and cache warming are common concerns, as are security and privacy considerations when caching sensitive data. Properly designed caches can dramatically reduce latency and conserve bandwidth while maintaining acceptable data freshness.