Pillar Axiom: How to calculate queue depth for SAN hosts that access the Pillar Axiom

Asset ID:	1-71-1968676.1
Update Date:	2017-01-08
Keywords:

Solution Type Technical Instruction Sure

Solution 1968676.1 : Pillar Axiom: How to calculate queue depth for SAN hosts that access the Pillar Axiom

Applies to:

Pillar Axiom 600 Storage System - Version All Versions and later
Pillar Axiom 500 Storage System - Version All Versions and later
Information in this document applies to any platform.

Goal

The purpose of the document is to provide a general guideline on how to calculate queue depth settings to be applied to SAN hosts that access Pillar Axiom.

Solution

The Axiom has a total of 4 Fibre Channel/iSCSI HBA ports per Slammer, 2 Fibre Channel/iSCSI HBA ports per Slammer Control Unit.

The HBA ports are QLogic HBA's, the SCSI Command queue is also known as the Execution Throttle/Queue Depth for the HBA adapter. For Fibre Channel, each HBA port can handle 2048 outstanding I/O requests per port at any one time. For iSCSI, each HBA port can handle 512 outstanding I/O requests per port at any one time.

The recommended qdepth settings to be applied on the hosts are determined by 2 factors:

Total number of LUNs exposed to the target ports
Array target port queue depth

Below is the formula that can be used to determine queue depth per Slammer CU:

Queue Depth = Target Queue Depth / number of LUNs

Example:
A single Slammer Axiom containing Fibre Channel HBA ports, with 10 LUNs assigned to Slammer CU0.

The queue depth would be 2048/10 = 204.8

If there are 10 SAN hosts in the environment that access all 10 LUNs on Slammer CU0, then the queue depth for each host would be 204.8/10 = 20.48

This example does not factor in High Availability. If you want to factor in HA, for example if a Slammer Control Unit fails, then you can use the same formula but include all LUNs that are assigned to both Slammer CU0 and Slammer CU1 and the SAN hosts have access to both Slammer CU0 and CU1.

The same formula can be applied if your Axiom is using iSCSI. In iSCSI each HBA port can handle 512 outstanding I/O requests per port at any one time. For more information on iSCSI configuration and best practices, refer to <Document 1497895.1> Pillar Axiom: iSCSI best practices.

Please make sure to follow the HBA vendor or iSCSI software initiator configuration to set execution throttle or queue depth to the appropriate settings.

Attachments

This solution has no attachment