VMware Cloud Community
richrussell
Contributor
Contributor
‎08-02-2007 02:56 AM

New HP MSA1000 SAN - Active/Active or Active/Passive?

I'm in the process of setting up a brand new MSA1000 and VI3 infrastructure. I have four ESX 3.0.1 servers, each with two HBAs leading to two separate 4/10 FC switches connected one each to two controllers in an MSA1000. The SAN kit isn't even in the rack yet, so there's no danger of corrupting data (only one ESX server is currently running too, using internal SCSI)

However, I'm not sure whether I should leave the firmware on the controllers as the Active/Passive v5.20, or install the Active/Active v7.00 version. The thing that I'm confused about is that as there's only one FC connection per controller, if one of the switches fails, will the MSA1000 fail over to the other controller in Active/Passive mode? Or do I need it to be in Active/Active to ensure there's a path to the LUNs if a switch fails?

I'm fairly new to SANs - so this might be a very foolish question! As the v7.00 firmware only came out in February, none of HP or VMware's documentation really mentions running MSA1000's in Active/Active.

Thanks

Richard

0 Kudos
3 Replies
arnoldvp
Contributor
Contributor
‎08-02-2007 03:35 AM

We have an MSA1000 running with 5 ESX 3.0.1 servers, each with 2 HBA's.

The firmware is version 4.48, which is active/passive.

It's running fine for a year now.

According to the HP website, active/active is not yet supported for ESX.

0 Kudos
BUGCHK
Commander
Commander
‎08-02-2007 04:47 AM

Richard,

in active/passive mode, ALL[/b] logical disks are maintained by one controller.

If one host causes a failover, the controllers swap their roles (active/standby) and ALL[/b] logical disks will fail over. ALL other[/b] hosts must notice this and just switch to the other controller.

In active/active mode, an individual logical disk is maintained by one controller. The other controller can manage a different logical disk.

If a controller receives an I/O request for a logical disk it does not manage, it does no longer return a 'NOT READY' status - it forwards the request to the managing controller for doing the actual disk I/O.

0 Kudos
richrussell
Contributor
Contributor
‎08-02-2007 05:28 AM

Thanks, those answers do help me.

Also I found that the following thread answers my questions completely:

http://www.vmware.com/community/message.jspa?messageID=458996

Many thanks.

Richard.

0 Kudos