MSA Storage

Windows 2008 cluster with MSA 2012 fc - performance issue

Chris S. S.
Occasional Contributor

Windows 2008 cluster with MSA 2012 fc - performance issue

I have set up a Windows 2008 cluster with two HP 360 G5s and an HP MSA 2012 fc. It works, I have managed to run the cluster validation wizard without any warnings and to form the cluster.

However, when I test the IO performance using SQL IO Sim, it is hopelessly slow, so something must be wrong (I get a Running average IO duration around 50 ms). I also get a number of warnings about IO requests which are "outstanding for more than 15 sec".

Apart from the SQL IO Sim results I don't really have any indications of anything wrong. However I get a bunch of informational messages in the Windows Evnet log from source 'ssproxy' with event id 513. They say thigs like:

"BuildAndSendPacket: CAPI_COMMAND_U_GET_CONFIG_SEQ_NUMBER BAD result No decode(-1) 13 seconds CAPI timeout"
"BuildAndSendPacket: CAPI_COMMAND_U_GET_DRIVE_ENCLOSURE BAD result No decode(-1) 74 seconds CAPI timeout"

What could be wrong?

I have upgraded the MSA with the most recent firmware and installed the CAPI Proxy and VDS and VSS providers on the servers.

Any help or advice would be deeply appreciated.



P.S. This thread has been moved from General to Storage Area Networks (SAN) (Small and Medium Business). - Hp Forum Moderator

New Member

Re: Windows 2008 cluster with MSA 2012 fc - performance issue

Hello Chris,

Did you connect the MSA directly to your servers, or through the use of an HP san switch? In our case we needed to manually put the correct speed settings on the HP San switch ports. This correction made an improvement of about +100MB write speed per second.

Make sure you install the MPIO feature that is available under "Features" in Windows 2008 if you have 2 HBA's in your server, after that make sure you configure the settings on the SAN disk in Windows, by going to disk management and right click on the disk. You will have an extra tab called MPIO. You can configure windows loadbalancing and redundancy from there.

Let me know if you need any more support.