Increasing the virtual SCSI queue depth might provide performance improvements for some virtual configurations. Understand the factors involved in determining a change to the virtual SCSI queue depth value.
The virtual SCSI queue depth value determines how many requests the disk head driver will queue to the virtual SCSI client driver at any one time. For AIX® and Linux® client logical partitions, you can change this value from the default value of 3 to any value from 1 to 256. You modify this value using the chdev command. For IBM® i client logical partitions, the queue depth value is 32 and cannot be changed.
Increasing this value might improve the throughput of the disk in specific configurations. However, several factors must be taken into consideration. These factors include the value of the queue-depth attribute for all of the physical storage devices on the Virtual I/O Server being used as a virtual target device by the disk instance on the client logical partition, and the maximum transfer size for the virtual SCSI client adapter instance that is the parent device for the disk instance.
For AIX and Linux client logical partitions, the maximum transfer size for virtual SCSI client adapters is set by the Virtual I/O Server, which determines the value based on the resources available on the server and the maximum transfer size set for the physical storage devices on that server. Other factors include the queue depth and maximum transfer size of other devices involved in mirrored-volume-group or Multipath I/O (MPIO) configurations. Increasing the queue depth for some devices might reduce the resources available for other devices on that same shared adapter and decrease the throughput for those devices. For IBM i client logical partitions, the queue depth value is 32 and cannot be changed.
chdev -1 hdiskN -a "queue_depth=value"hdiskN represents the name of a physical volume and value is the value you assign between 1 and 256.
lsattr -El hdiskN