Unsolved
This post is more than 5 years old
1 Rookie
•
56 Posts
0
24347
May 22nd, 2013 10:00
One group, with 2 members.. best practice?
Hi, We have one EqualLogic group with two members (2 x 6510E). The SAN is primary used as VMWare datastores for ESXi hosts. Currently I have two storage pools (one for each member); a RAID10 and a RAID6. I'm trying to figure out what best practice is - to leave it as is with two separate pools, or to "glue them together" such that one pool spans both members, with one RAID level. Seems from a performance standpoint, with MPIO, there are benefits to having both members in the same pool. We do not "need" to have multiple RAID levels per-say, nor do we "need" to have multiple pools, but we are trying to decide what to do. If we do this, if we lose an entire array we're hosed, but I'm assuming the level of redundancy built in to these is sufficient enough so that should not be a concern?
Thanks
Bryan
bealdrid
1 Rookie
•
56 Posts
0
May 22nd, 2013 10:00
Hi Don, yes I would convert the RAID10 to a RAID6. We do have the MEM module installed on all our ESXi hosts. I have SAN HQ pulled up, is there anything in particular I can look at to help make the decision? Connections count is very low right now.
Dev Mgr
6 Operator
•
9.3K Posts
•
3 Points
0
May 22nd, 2013 11:00
Running that much IO from virtual machines on raid 6 may not be the best move for performance.
Like Don suggested, check your SANHQ info. One thing to look for is to see what your read/write ratio is (if you're over 25-30% write its probably not a bad idea to keep 1 unit raid 10 when merging the pools).
Raid 10 yields typically the best performance, but the least space. Raid 6 yields the most space, but write performance is quite a bit lower than raid 10 (or even 50). Read performance on raid 6 is good, but mostly for sequential IO (not random like multiple VMs).
Raid 50 is in between the 2, but depending on your drive size and speed (rpm), this may not be the best choice (rebuild times can run into a couple of days with 1+TB 7200rpm drives, during which you're in a critical state (i.e. lose 1 more drive and you lose (all) data)).