Configuring CLARiiON LUNs for VMware Servers

Can somebody tell me in one page what are best practices for configuring CX4 LUNs for VMware.

Our current practice is this:

1. Tier 1 RAID Groups are FC 15K R1/0. LUNs are 500 GB, presented to ESX Servers and used as VMFS.

2. Tier 2 RAID Groups are FC 15K R5. LUNs are 500 GB, presented to ESX Servers and used as VMFS.

3. Tier 3 RAID Groups are SATAII 7200RPM R5. LUNs are 500 GB, presented to ESX Servers and used as VMFS.

All R5 CLARiiON LUNs are Striped MetaLUNs on two "adjacent" DAEs on different busses.

We don't use RDMs.

ESX Admins assign "logical volumes" to Virtual Machines as they see fit. We (in SAN Team) don't even know what's on LUNs.

I have heard of people seperating their OS, Swap and Application volumes on different physical LUNs, but we don't do that. We just give 500 GB LUNs to ESX Admins, and they "configure" the OS, Swap and Application "logical volumes" as they see fit.

Responses(1)

Allen Ward

2.1K Posts

1

April 6th, 2010 11:00

I can share what we do, but I don't think it qualifies as a "best practice" anywhere but in our environment :-)

All ESX LUNs are a uniform size: 268GB
RAID Groups for ESX LUNs are RAID5(8+1) using 15k drives
Only 2 ESX LUNs provisioned from each RAID group, with any extra space allocated to low IO applications (file servers, web servers, etc.)
- We originally used 73GB drives for this so the two LUNs consumed the entire RG. Now we can't get 73GB drives so we had to modify our "standard" a bit
We try to allocate the two LUNs from each RG to different ESX pods, but that is not a hard and fast rule
Once we allocate the LUN to the server the ESX Support admins use it however they want (but no RDM). They are responsible for trying to balance the activity on their own LUNs

This has worked well for us for several years now and has even translated well into a modified (but very similar) version of the strategy for ESX on our DMX3 arrays.

View All

No Events found!