Drive queue depth

For performance reasons, you might want to change the queue depth of the disk command. The disk queue depth limits the maximum number of commands that the Linux software can issue concurrently to that disk at any time.

Increasing the disk queue depth might improve the disk performance by increasing disk throughput (or I/O) but might also increase latency (response delay). Decreasing the disk queue depth might improve disk response time but decrease the overall throughput. The queue depth is viewed and changed on each individual disk. When changing the disk queue depth, the command elements and data transfer window on the parent adapter might also need to be changed.

Viewing the drive queue depth

To view the current queue depth on any disk (JBOD or RAID), use the iprconfig -c query-qdepth sda command from the iprconfig utility command line. Alternatively, you can use iprconifg graphical interface (iprconfig > Work with disk configuration).

The queue_depth attribute contains the current setting. The default value for the disk queue depth is determined by the adapter family.

Table 1. Drive queue depth for different adapter families
Queue depth	PCI-X and PCIe adapter family	PCIe2 adapter family
Default JBOD disk queue depth	16	16
Default RAID disk queue depth	4 times the number of disks in the RAID array	16 times the number of disks in the RAID array

Example

To list the current queue_depth attribute value for the sda disk, type the following command:

iprconfig -c query-qdepth sda

The system displays a message similar to the following: 16.

Changing the drive queue depth

You can change the drive queue depth from the command line by running the following command:

iprconfig -c set-qdepth sda 64

Last updated: Wed, May 24, 2017