Replacing SAN Switches

[outbacker]

We are about to replace our existing 9216 san switches that our esx farm connects to with some new 4 gb model. The HP xp10000 array will also be connecting to these new switches. Because the xp chip ports are moving, the path of the storage target will change as well. My question is, what happens on the ESX 3.5 side when the target changes? What will happen to my vm disks, will they disappear, will a rescan bring them back? Will they show up as new storage? I'm not sure where to start. I imagine anyone assisting would need more questions answered, but thanks for any help.

[beckhamk]

i would also like to know the answer to this

[malaysiavm]

I think the information you put here is not clear enough. If you are replacing a SAN switches, and your HBA connection is provide with redundancy and failover path, I believe most of this cases, you may not experience interruption if you have round robin configure. If the SAN zoning is gone or change, it might require down time of your ESX. From storage level, as long as the data is remain in the LUN and not been wiped off, you can reconfigure it and you should able to get your VM or data back. Just some idea for your reference here. I will suggest yuo should further test this out.

Malaysia VMware Communities -

[williambishop]

What he's asking is whether or not he will have to resignature, will have an outage, or will have to make any changes to address this switch change.

Are you staying in a cisco network? Are you using static fcid's or dynamic?

If you're changing out the switches, and these are edge switches, I imagine you will have no issues, but to double check, go to cisco's forums....I can tell you that I have changed out switches, but they were edge, not core, with no issues. My 9509's are still intact. I'm about to do another round myself when I move from my 9216i's to the new 9136's at the edge.

[outbacker]

The 9216's are core now, we will introduce new 9143's as edge and plug the XP ports into the new 9143's. The esx hosts will remain in the 9216's. I would think this would change the target for the esx hosts if the XP ports were changed to a new edge switch. Does that answer the question, I'm relaying from our SAN admin, not my specialty. Thanks. Oh, and they are dynamic fcid's.

[williambishop]

Core and dynamic, yeah, you're probably looking at downtime. Roll on over to the cisco forums and you should be able to get some in depth help on the subject(that's where I got mine).

[outbacker]

So if i take away one path from the 9216 (there's two) to the esx host, the host will failover to the one that is left. If we move one storage processor on the xp over to the new switch (again, theres two) rezone the luns and represent to esx, the target shouldn't change, just the underlying addressing will correct? Does esx just care about the target, and if it's presented with the same lun id, won't esx just think it's the same lun? or would you have to resignature the disks?