HP P4300 SAN performance

prodriveit
prodriveit used Ask the Experts™
on
We are setting up a new HP P4300 SAN environment to use as shared storage for VMware ESXi 4.1. We have four storage nodes and two dedicated HP 2910al-24G ISCSI switches. The nodes are connected to the switches using a 1Gb NIC (also set as the SAN/IQ port) and two bonded 10Gb NICs. The hypervisors are connected using two 1Gb NICs, which are set up in for round-robin MPIO.

When running an IOMeter performance test, I see the performance of a drive on the SAN compared to a drive on local storage to be around a third to a sixth. For instance, the 512 byte 100% read, 0% random test shows 0.97MB/s on the SAN and 5.97MB/s on local storage. The 64K 100% read, 0% random test shows 36.30MB/s on the SAN and 222.60MB/s on local storage.

I have enabled jumbo frames end-to-end, and flow control on the SAN nodes and the switches. Is this an acceptable level of performance, or is there something else I can try to improve it?
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
Andrew Hancock (VMware vExpert / EE Fellow)VMware and Virtualization Consultant
Fellow 2018
Expert of the Year 2017

Commented:

Author

Commented:
I enabled jumbo frames after prior testing, and the performance actually seemed to decrease a bit (at least, on the 512 byte test).

Each node is configured as RAID5, and all four nodes are in a Network RAID10 configuration.
Andrew Hancock (VMware vExpert / EE Fellow)VMware and Virtualization Consultant
Fellow 2018
Expert of the Year 2017

Commented:
Yes, it's not a given with Jumbo Frames, it will increase performance.
Acronis in Gartner 2019 MQ for datacenter backup

It is an honor to be featured in Gartner 2019 Magic Quadrant for Datacenter Backup and Recovery Solutions. Gartner’s MQ sets a high standard and earning a place on their grid is a great affirmation that Acronis is delivering on our mission to protect all data, apps, and systems.

I found a configuration issue on the switch - two of the 10Gb trunks on each switch were untagged on the incorrect VLAN. I changed this and saw in improvement in some tests.

For example, the 512 byte read 100% read test now shows about 2MB/s on the SAN. The 32K 100% read test shows the same speed as before (36MB/s), though some tests show a much closer parity with local storage (particularly with the larger transfer sizes).

Would anyone be able to give me an idea of the sort of results that I should be seeing from IOMeter on a SAN of this size?
Top Expert 2014

Commented:
There are some benchmark results in the quickspecs - http://h18000.www1.hp.com/products/quickspecs/13254_div/13254_div.html

Author

Commented:
Solved the problem.

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial