GZIP

A lossless single-file compression format and utility created for the GNU Project, using the DEFLATE algorithm, that compresses exactly one file or data stream per .gz output — never multiple files as a bundle.

Extension

.gz

Full name

GNU zip

Introduced

1992

Vendor

GNU Project (Jean-loup Gailly and Mark Adler)

What is inside the container

Unlike ZIP or TAR, gzip compresses a single input stream; it carries no concept of multiple files or folders, which is why compressing a whole directory under Unix conventionally means running tar first to bundle the files, then gzip to compress that single resulting stream (producing .tar.gz). Internally it uses the same DEFLATE algorithm as ZIP, wrapped in its own minimal header/trailer format specified in RFC 1952, and it is entirely lossless.

How to open it

Linux/Unix/macOS The built-in gzip/gunzip command-line tools ship by default on virtually every Unix-like system

Windows 7-Zip or WinRAR both extract .gz files through their GUI

Programmatic access Every mainstream language's standard library includes gzip support (Python's gzip module, Node's zlib, Java's GZIPInputStream) for reading/writing streams in code

DAM/asset pipeline Web servers and DAM delivery layers commonly apply gzip on-the-fly (HTTP Content-Encoding: gzip) to compress text-based API responses and metadata payloads in transit, separate from archival storage

For archives

gzip is a safe, stable choice wherever a DAM needs single-stream compression — most commonly as the second stage of a TAR.GZ backup, or for on-the-fly HTTP compression of metadata and text payloads — because its specification (RFC 1952) is open, patent-free, and near-universally implemented. It's not a substitute for a true multi-file archive format, so a DAM should reach for TAR.GZ or ZIP whenever bundling more than one file is the actual goal.

Jean-loup Gailly wrote gzip for the GNU Project, releasing version 0.1 on October 31, 1992, specifically to give Unix users a patent-free replacement for the compress utility, whose LZW algorithm was encumbered by Unisys and IBM patents that wouldn’t expire until the early 2000s. Gailly designed gzip’s DEFLATE algorithm together with Mark Adler, combining LZ77 dictionary matching with Huffman coding — the same core technique Phil Katz had used for ZIP a few years earlier — and the format was later documented independently as RFC 1952, keeping it free of the licensing questions that dogged its predecessor.

Because gzip compresses exactly one stream and has no concept of bundling multiple files, it almost never appears alone in a DAM workflow; it’s the second half of the ubiquitous TAR.GZ pattern (tar to bundle, gzip to shrink) used for backups and large asset transfers between Linux systems, and it’s also the compression a DAM’s own web server applies transparently to shrink JSON metadata and API responses in transit. Its open specification and universal support across every server-side language make it close to a non-decision for a DAM’s infrastructure layer — the only real question is whether gzip’s ratio is good enough or whether a newer algorithm like Brotli or Zstandard is worth the added complexity for a given pipeline.

Frequently asked

Can a .gz file contain more than one file?

No — gzip compresses exactly one input stream; bundling multiple files first requires a separate tool like tar, which is why multi-file gzip compression appears as the combined .tar.gz format rather than plain .gz.

Why was gzip created when compress already existed on Unix?

The compress utility's LZW algorithm was covered by Unisys and IBM patents that limited free redistribution, so the GNU Project created gzip in 1992 as a patent-free replacement using the DEFLATE algorithm instead.

Is gzip the same compression as ZIP?

They use the same underlying DEFLATE algorithm, but different container formats — a gzip file compresses one stream with a minimal RFC 1952 header, while a ZIP file wraps possibly many separately-compressed entries with an index (central directory).

Where does gzip typically show up in a DAM's infrastructure without anyone archiving anything?

In HTTP transport — web servers and APIs commonly apply gzip Content-Encoding on the fly to shrink JSON metadata, HTML, or text responses in transit, entirely separate from any asset archival use of the format.

Is gzip a lossy or lossless compression method?

Lossless — every byte of the original data is recovered exactly on decompression, which is required for it to be usable on binary assets and metadata alike, not just text.

Why is TAR.GZ so common for backing up a DAM's asset library?

Because plain TAR bundles files but doesn't shrink them, and plain gzip shrinks a stream but can't bundle multiple files — piping tar's output through gzip gets both bundling and compression in one step, which is the standard Unix pattern.

Does every operating system support gzip natively?

Linux and macOS ship gzip/gunzip by default; Windows has no built-in GUI support, though 7-Zip, WinRAR, and Windows' bundled tar.exe can all handle .gz files.

Who designed the DEFLATE algorithm gzip uses?

Jean-loup Gailly and Mark Adler designed gzip's implementation of DEFLATE for the GNU Project; the same core algorithm (LZ77 plus Huffman coding) had originally been developed by Phil Katz for PKZIP.

Sources

gzip was released by the GNU Project, with version 0.1 available on October 31, 1992, as a patent-free replacement for the LZW-based compress utility. checked 2026-08-07 — Wikipedia — Gzip
Jean-loup Gailly designed the gzip file format, which was later specified independently in RFC 1952, distinct from the DEFLATE algorithm itself (RFC 1951). checked 2026-08-07 — GNU Gzip Manual