Sharing Raw disk among VM's without a cluster.

[MarioVerschuur]

I have a question about using Physical RAW LUN's in VM's. I've created a RAW disk mapping which I use in VM1. I've created the same mapping to the same LUN in an other VM called VM2. When I shutdown VM1 and boot VM2, I can use the datadisk in VM2. When I shutdown VM2 and boot VM1 the same datadisk is available in VM1.

This seems to be working. The question is... is it good practice? For testing environments okay. But we like to have a high availability SQL Server, without having to use a MS-w2k3-cluster. So the VM1 and VM2 are actually nicely vmotionable, where cluster-members aren't and having one server and one backup might be good enough.

Does anyone have any experience with this, good or bad?

Thanks in advance for any answers to this question.

Gr. Mario.

[Texiwill]

Hello,

LifeKeeper HA solution uses exactly what you discuss but may still have the same limitations as MSCS. To use MSCS in VI3 you should have the boot disks on local storage (No vMotion support) and use RDM/Raw devices.

I have used both methods you discuss and it all boils down to what uptime you require and the tools you desire to use. VI3 supports both.

I think the shared RAW LUN you are discussing has less locking impact on ESX and less limitations as in where the boot drive may live. But you loose some immediacy in failover depending on how MSCS is implemented. It may take just as long for VM2 to boot than it does for MSCS to switch services. You do need something that will automatically start VM2 if VM1 is down (not just unreachable).

Best regards,

Edward

Message was edited by:

Texiwill

[MarioVerschuur]

The systemdisks are on a alternate VMFS. The aspect of having a automatic failover (MCSC) vs Manual failover (shared RAW LUN) is what brings a slightly lower availability. With some scripting in VMWare we should be able to detect VM1 going down and starting VM2 up.

We have had some issues in the pas where MS clustering brought us more trouble than availability. Failing over did cost a large amount of time and the failover initiated without interference from a system administrator. Users were having trouble accessing there files/printers for a few minutes. In that time a VM is booted and up and running.

What did occur to me is that when you add the RAW LUN to VM1, format it and then add it to VM2, you can't seem to do it on the same ESX Server. You have to add the disk via another ESX, but once added to both VM's it's no problem to have them both on the same ESX, as long as you keep one VM powered down.

Greetings,

Mario

PS thanks for your swift answer.

[Nithiyan]

How to map same raw lun to two VM machine? we are using SAN.

What are the steps need to do.

asist on this

Regards

Anand