What fio is
fio is a free, open-source tool for simulating storage I/O workloads to assess performance and reproduce bugs. You define jobs in files, with the option to describe multiple threads, processes, and files. Jobs can issue sequential, random, or mixed reads and writes through buffered, direct, or raw I/O. Settings include block and I/O size, engine, queue depth, target files or devices, and job count. Documented engines include Linux io_uring and libaio, POSIX asynchronous I/O, and synchronous read/write. Separate engines support Ceph RADOS and RBD; the HTTP engine sends GET and PUT requests to WebDAV or S3 endpoints. Results are available as normal text, terse output, JSON, JSON+, and bandwidth or latency logs. Terse output is suited to scripts, tables, or graphs, and verification options can check written data after a run, including repeated burn-in tests. fio supports Linux, Solaris, AIX, HP-UX, macOS, NetBSD, OpenBSD, Windows, FreeBSD, and DragonFly. The project uses the GPL-2.0 license.
Who it suits
fio suits storage administrators and developers who need configurable I/O workloads for performance checks or bug reproduction. Its job files, script-friendly output, and automation support also suit repeatable testing workflows.
Good to know, and what to watch
- Free, GPL-2.0-licensed software.
- Configurable sequential, random, and mixed I/O.
- Job files can describe multiple threads and processes.
- Exports JSON, terse output, and bandwidth/latency logs.
- Verification options can check written data.
- HTTP direct I/O is limited to depth 1.
- Bug reports go to the mailing list or GitHub, not direct email.
Plans
Every plan and pricefio
Free plan
The verdict
fio offers detailed workload controls, multiple output formats, and data verification for storage I/O testing. Its documented HTTP direct-I/O depth limit is worth noting when planning those workloads.
What the maker publishes
- CLI or automation
- Yes source
- Comparison database
- No source
- Custom test profiles
- Yes source
- Data verification
- Verification options can check written data after a workload, including repeated burn-in tests. source
- Distribution
- The repository identifies GPL-2.0 as the project license. source
- Founded
- 2006 source
- Free plan
- Yes source
- I/O engines
- Documented engines include Linux io_uring and libaio, POSIX asynchronous I/O, and synchronous read/write engines. source
- I/O patterns
- Jobs can perform sequential, random, or mixed reads and writes, using buffered, direct, or raw I/O. source
- Notable limit
- The documented HTTP engine supports direct I/O only at I/O depth 1, with scaling through multiple jobs. source
- Operating systems
- Linux,Solaris,AIX,HP-UX,macOS,NetBSD,OpenBSD,Windows,FreeBSD,DragonFly source
- Output
- Results can be written in normal, terse, JSON, or JSON+ formats, with terse output intended for scripts, tables, or graphs. source
- Platforms
- The project lists Windows installers for tagged releases and packages for Linux distributions, Solaris, and BSDs; its release notes also mention macOS. source
- Purpose
- fio simulates configurable I/O workloads for performance testing and reproducing bugs. source
- Result export formats
- normal text,terse (semicolon-delimited),JSON,JSON+,bandwidth/latency logs source
- Storage integrations
- The HTTP engine supports GET and PUT requests to WebDAV or S3 endpoints, and separate engines provide Ceph RADOS and RBD access. source
- Support
- The maintainers direct bug reports to the fio mailing list or GitHub rather than to them by direct email. source
- Tuning
- Workload parameters include block size, I/O size, I/O engine, queue depth, target files or devices, and job count. source
- Workload areas
- Storage I/O: read,write,trim,mixed workloads; bandwidth,IOPS,and latency source
- Workload setup
- Workloads are defined in job files that can describe multiple threads, processes, and files. source
Checked against the maker's pages on 4 Oct 2026.