How to use the command 'fio' (with examples)

Fio is a flexible I/O tester tool that allows users to spawn multiple threads or processes to perform I/O actions. It is commonly used to benchmark storage and filesystems.

Use case 1: Test random reads

Code:

fio --filename=path/to/file --direct=1 --rw=randread --bs=4k --ioengine=libaio --iodepth=256 --runtime=120 --numjobs=4 --time_based --group_reporting --name=job_name --eta-newline=1 --readonly

Motivation:

• This use case is useful for measuring the performance of a storage system in terms of random read operations. It helps to determine the throughput and latency of the storage device when handling random read workloads.

Explanation:

• --filename=path/to/file: Specifies the file or device to perform the I/O operations on.
• --direct=1: Enables direct I/O, bypassing the operating system cache.
• --rw=randread: Sets the I/O pattern to random reads.
• --bs=4k: Defines the block size for each I/O operation as 4 kilobytes.
• --ioengine=libaio: Sets the I/O engine to libaio, which is a Linux asynchronous I/O access library.
• --iodepth=256: Sets the depth of I/O queue to 256, allowing multiple I/O operations to be issued in parallel.
• --runtime=120: Sets the duration of the test to 120 seconds.
• --numjobs=4: Specifies the number of jobs (threads or processes) to spawn for the test.
• --time_based: Runs the test for a specific duration instead of a fixed number of I/O operations.
• --group_reporting: Aggregates the results of multiple jobs into a single report.
• --name=job_name: Specifies a name for the test job.
• --eta-newline=1: Displays the estimated time of arrival (ETA) for the completion of the test on a new line.
• --readonly: Performs read-only operations, ensuring the data is not modified during the test.

Example output:

...
Read IOPS: xxx
Read Bandwidth: xxx MB/s
Average Latency: xxx ms
...

Use case 2: Test sequential reads

Code:

fio --filename=path/to/file --direct=1 --rw=read --bs=4k --ioengine=libaio --iodepth=256 --runtime=120 --numjobs=4 --time_based --group_reporting --name=job_name --eta-newline=1 --readonly

Motivation:

• This use case is useful for measuring the performance of a storage system in terms of sequential read operations. It helps to determine the throughput and latency of the storage device when handling sequential read workloads.

Explanation:

• --rw=read: Sets the I/O pattern to sequential reads.
• All other arguments have the same meaning and functionality as explained in Use case 1.

Example output:

...
Read IOPS: xxx
Read Bandwidth: xxx MB/s
Average Latency: xxx ms
...

Use case 3: Test random read/write

Code:

fio --filename=path/to/file --direct=1 --rw=randrw --bs=4k --ioengine=libaio --iodepth=256 --runtime=120 --numjobs=4 --time_based --group_reporting --name=job_name --eta-newline=1

Motivation:

• This use case is useful for measuring the performance of a storage system in terms of random read/write operations. It helps to determine the overall I/O capabilities of the storage device when handling mixed workloads.

Explanation:

• --rw=randrw: Sets the I/O pattern to random read/write operations.
• All other arguments have the same meaning and functionality as explained in Use case 1.

Example output:

...
Read IOPS: xxx
Read Bandwidth: xxx MB/s
Write IOPS: xxx
Write Bandwidth: xxx MB/s
Average Latency: xxx ms
...

Use case 4: Test with parameters from a job file

Code:

fio path/to/job_file

Motivation:

• This use case is useful when you have defined the test parameters in a job file, which contains all the necessary arguments for fio. By providing the path to the job file, you can easily execute the test with the specified parameters without having to manually specify each argument on the command line.

Explanation:

• path/to/job_file: Specifies the path to the job file containing the test parameters.

Use case 5: Convert a specific job file to command-line options

Code:

fio --showcmd path/to/job_file

Motivation:

• This use case is useful when you want to convert a specific job file into its equivalent command-line options. It helps to understand the actual arguments that would be used when running the test with the specified job file.

Explanation:

• --showcmd: Instructs fio to display the command-line options based on the specified job file.
• path/to/job_file: Specifies the path to the job file for which the command-line options should be displayed.

Example output:

fio --filename=/path/to/file --direct=1 --rw=randread --bs=4k --ioengine=libaio --iodepth=256 --runtime=120 --numjobs=4 --time_based --group_reporting --name=job_name --eta-newline=1 --readonly

Conclusion:

By using the fio command with different options and parameters, you can effectively benchmark the performance of storage systems and evaluate their ability to handle various I/O workloads. Whether it’s testing random or sequential reads, random read/write operations, or using predefined job files, fio offers flexibility to simulate real-world scenarios and gather valuable performance metrics.