In my environment I have some high end storage arrays which have expensive disks and I also have a number of older EVA SANS which are cheaper but still perform well.
I'd like to only use my high end storage for tier 1 virtual machines and everything else will sit on the EVA's as tier 2.
Both SANS are connected to the same brocade directors and can have their storage zoned to my ESX hosts easily enough but I wanted to make sure I'm not missing something obvious in terms of issues I could face further down the line.
My naming convention will show if a LUN is presented from tier 1 or tier 2, so that's not a problem.
Do I need to care about LUN numbering at all?
Thanks in advance
for me on the safer side and better management i would have unique LUN ID for each LUN. easier for me to trace later too ..
LUN ID on host side are not so important (only with RDM).
So no problem.
But to avoid confusion (at user level ) I suggest to start presenting a single storage to all host, then go to second, and so on...
Andre
Andre,
I'm not sure I follow what you're suggesting there.
How do you mean single storage to each host and then second?
You have multiple storage, as you have written.
So start with the first EVA, make a rescan to all hosts.
Then go to second EVA, make a rescan to all hosts.
... An so on
With with "trick" you will have same LUN ID on each host.
Andre
Oh ok.
I known it's best practice to present the same LUN ID to each host which makes sense.
I could say storage array A uses LUNS from 1 to 150 and storage arrayB uses 151 to 255
The only trouble with this approach is you half the number of pottential LUNS you could allocate.
I wouldn't worry about LUN ID's. Since the LUNs are presented from different targets they are unique anyway.
Btw. regarding
The only trouble with this approach is you half the number of pottential LUNS you could allocate.
The limit on FC LUNs per host is 256.
André