How to measure disk I/O performance in Linux

Use dd for quick testing, fio for deep analysis, and iostat for real-time monitoring. dd can evaluate the sequential read and write speed, such as dd if=/dev/zero of=testfile bs=1G count=1 oflag=direct test write; fio supports multiple I/O modes, such as fio --name=randwrite --rw=randwrite --bs=4k --direct=1 --iodepth=16 simulate random write IOPS; iostat -x 1 checks %util, await, rkB/s and other indicators in real time to judge disk bottlenecks. Combining the three, it can comprehensively evaluate Linux disk performance.

To measure disk I/O performance in Linux, you can use a combination of built-in tools and simple benchmarks that evaluate read, write, and random access speeds. The most effective methods involve dd , fio , and iostat , each serving different purposes depending on whether you want quick tests or detailed analysis.

Using dd for basic disk speed testing

dd is a simple command-line tool available on almost every Linux system. It's useful for a quick assessment of sequential read/write performance.

To test write speed:

This creates a 1GB file using direct I/O (bypassing cache), giving a realistic write speed. After the test, remove the file:

To test read speed, use:

Note: Results are rough and can be affected by filesystem settings and caching. Use oflag=direct and iflag=direct to minimize cache impact.

Using fio for advanced I/O benchmarking

fio (Flexible I/O Tester) is a powerful tool for simulating various I/O workloads—sequential, random, read/write mix, different block sizes, and queue depths.

Install fio (on Debian/Ubuntu):

Example: Test random write IOPS with 4K blocks:

Key options:

• --rw=randwrite : Random writes (use randread, read, write, etc.)
• --bs=4k : Block size (simulate database or small-file workloads)
• --direct=1 : Bypass page cache
• --iodepth=16 : Simulate concurrent I/O operations
• --ioengine=libaio : Use asynchronous I/O for better performance simulation

fio outputs detailed stats like bandwidth, IOPS, and latency—ideal for tuning storage systems.

Monitoring real-time I/O with iostat

iostat (part of sysstat package) shows live disk utilization and throughput, helping identify bottlenecks during system operation.

Install and run:

This prints extended statistics every second. Key metrics include:

• %util : Device utilization percentage. Consistently near 100% indicating satisfaction.
• await : Average wait time (ms) for I/O requests.
• svctm : Average service time (deprecated but still shown).
• rkB/s & wkB/s : Read and write throughput in KB per second.

Use this during application load to see how disks behave under real conditions.

For accurate results, always unmount the target filesystem or use a dedicated test file, and avoid running benchmarks on production systems during peak hours. Warm up the system with a preliminary run when using fio.

Basically, start with dd for a quick check, use fio for deep analysis, and rely on iostat for ongoing monitoring. Each tool gives insight into different aspects of disk performance.

The above is the detailed content of How to measure disk I/O performance in Linux. For more information, please follow other related articles on the PHP Chinese website!