paginacache

paginacache, commonly referred to as the page cache in Unix-like operating systems, is the memory resident cache used to speed up file input and output by keeping recently accessed disk pages in RAM. When a program reads data from a file, the kernel first checks the page cache; if the data is present, it is returned from memory, avoiding a disk seek. If not, the kernel reads from storage and stores the page in the cache for subsequent accesses.

The page cache stores both file data and filesystem metadata, such as inodes and directory blocks. It is managed by the virtual memory subsystem and the filesystem layer, using mechanisms such as readahead to anticipate sequential access and writeback to flush dirty pages asynchronously. Modified (dirty) pages are written back to disk according to the system’s policies, improving write throughput while masking latency. Clean pages may be evicted to satisfy memory pressure.

Administration and tuning: under memory pressure, the kernel may reclaim page cache pages. Administrators can influence behavior with tools to drop cached pages or with kernel parameters like vm.vfs_cache_pressure. Monitoring is typically done via procfs or system metrics, observing cache size, hit rates, and reclaim activity.

Limitations and considerations: while page cache generally improves read performance, it consumes RAM and can interact with devices or workloads that bypass or bypass the cache (for example, direct I/O). Workloads with random access to large datasets or those requiring deterministic latency may need configuration or alternative I/O paths to achieve predictable performance.

See also: page cache, disk caching, Linux kernel, virtual memory.