Compression

Network performance depends on how far data travels, how many exchanges are required, and how many bytes cross the path. Compression targets the third factor.

A compressor represents repeated or predictable byte patterns with a smaller sequence of bytes. The receiver reverses that representation to recover the original content. HTTP commonly uses the lossless content codings gzip, br for Brotli, and zstd for Zstandard.

Smaller responses can finish sooner, consume less bandwidth, occupy fewer packets, and reduce origin or CDN egress. Compression also consumes CPU, memory, and time at both endpoints. It can delay the first byte, expand small or already compressed content, complicate caching, and expose sensitive information through response sizes.

An effective policy selects a coding only when it reduces end-to-end cost for the resource and client.