fio を Homebrew, apk, chocolatey, apt, dnf, Nix, pacman, zypper, scoop, winget でインストール

プロジェクトリンク

• ホームページ
• 上流ドキュメント
• パッケージマネージャページ

コマンドとエイリアス

• fio
• fio-btrace2fio
• fio-dedupe
• fio-genzipf
• fio-histo-log-pctiles.py
• fio-verify-state
• fio2gnuplot
• fio_generate_plots
• fio_jsonplus_clat2csv
• fiologparser.py
• fiologparser_hist.py
• genfio

プロジェクトの歴史

The fio manual includes its own origin story: Jens Axboe wrote fio because he was tired of writing special-purpose test programs every time he needed to test a particular workload for performance analysis or bug reproduction. The manual dates that authorship note to September 5, 2006.

That motivation shaped fio's design. Instead of being a single benchmark with one workload model, fio became a flexible engine for simulating many combinations of threads, processes, I/O engines, block sizes, queue depths, files, devices, and verification settings.

The canonical source is documented as Jens Axboe's kernel.org Git repository, with GitHub listed as an official mirror. That kernel-adjacent hosting history fits fio's role: it is a tool used by storage developers, kernel engineers, filesystem maintainers, database operators, and performance testers who need reproducible I/O workloads.

採用の歴史

fio's own documentation records distribution adoption across Debian starting with Squeeze, Ubuntu starting with 10.04 LTS, Fedora starting with Fedora 9 and EPEL 4, plus Mandriva, Arch Linux, Solaris via OpenCSW, BSD package repositories, and Windows installers beginning with fio 3.31.

That adoption pattern is unusually broad for a benchmark: fio is not just a lab tool, but a package-manager-standard executable that support teams and performance engineers expect to find on servers, workstations, CI images, and storage appliances.

使われ方

The normal fio workflow is to write a job file that matches the I/O load to simulate, then run fio with that job file. Users can also pass options directly on the command line, and multiple job files can be serialized from one invocation.

fio is used to measure throughput, latency, IOPS, queue-depth behavior, async I/O, filesystem behavior, device saturation, and failure or verification cases. Because it can reproduce tailored workloads without writing new C test programs, it is a common bridge between bug reports and repeatable performance experiments.

パッケージ好きにとっての重要性

fio is package-nerd significant because it is the de facto portable storage torture tool. When a package, filesystem, block device, cloud volume, or database stack claims performance behavior, fio is often the small executable used to make that claim reproducible.

It is also a reminder that good benchmarking tools are languages. fio job files encode workload intent, and packagers care because shipping fio gives users a shared grammar for comparing machines, kernels, filesystems, and storage backends.

タイムライン

• 2006: Jens Axboe authors fio to avoid repeatedly writing one-off I/O workload test programs.
• 2008: fio documentation records Fedora 9 and EPEL 4 as starting points for Fedora/EPEL packages.
• 2010: fio documentation records Ubuntu 10.04 LTS as the starting point for Ubuntu packages.
• 2011: fio documentation records Debian Squeeze as the starting point for Debian packages.
• 2022: fio 3.31 begins providing Windows installers for tagged releases.
• 2026: fio documentation identifies 3.41-era development and continued maintenance.

Related projects

• Linux kernel block-layer and I/O scheduler development are closely related to fio's origin and ongoing use.
• iozone, bonnie++, dd, stress-ng, and filesystem-specific benchmarks are neighboring tools, but fio is usually chosen when workload shape matters.
• Kernel.org infrastructure and the fio mailing list provide the project's canonical development context.

ソース

• https://fio.readthedocs.io/en/master/fio_doc.html
• https://raw.githubusercontent.com/axboe/fio/master/README.rst
• https://git.kernel.org/pub/scm/linux/kernel/git/axboe/fio
• https://github.com/axboe/fio