Go to the first, previous, next, last section, table of contents.
In old versions of e2compr, compression would only happen when the file in question had been closed by all processes.
However, compression now also happens whenever a cluster boundary is reached during writing.
Decompression occurs when needed, i.e. every time the kernel wants to
read a compressed cluster from disk into memory. The cluster remains
compressed on the disk. The kernel does cache the file data that it
reads from disk, just like any other file (see the `cached' heading
of the output of free).
(At the time of writing, this caching is suboptimal in that all files are given equal weight, whether they're compressed or uncompressed, and whether they're on a fast hard drive or on a slow floppy or CD-ROM. I understand that things like weighted caching and even caching to swapspace are planned for the 2.3 kernel series.)
Go to the first, previous, next, last section, table of contents.