VMWare backup SAN Transport on HP Lefthand

[ALiving]

Hello

We are looking at configuring SAN Transport for backing up our VM's. I have read the following document and had some questions:

http://www.symantec.com/business/support/index?page=content&id=TECH155831&key=56792&actp=LIST

It says the LUNs should be presented to the BE server with read/write ability, but the Lefthand can mount LUNS read only. Can I mount the LUN read only instead? Will the backup still work even though it can't write to the LUN, I'm a little uncomfortable having Windows attached to our VMFS volumes, even after turning off Automount.

I have backup exec 2010 version 13.

[teiva-boy]

You go into diskpart, you type in: automount disable

and then type: automount scrub

Then you make the LUN read/wirte in the SAN mgmt interface.

The BE server can see the LUN, and you never give the volume a drive letter or disk signature, nothing.  Just let it be an unknown disk within the disk manager.  

That is the practice, literally tens of thousands of customers not just BE customers, but NBU, Commvault, ArcServe, IBM TSM, and more all do this.  

[ALiving]

I understand the process, I have read the KB article. What I want to know, is if the LUN can be presented read only to the BE server without having any affect on backups.

If all I am doing is backing up from that LUN, then as far as I can see, no need to allow the BE server to write to that LUN, either on purpose or by accident. This adds that little bit extra layer of safety to the VMFS volume.

On the other hand the VMWare API's might write to the LUNs, but no documentation I can find verifies this.

[ZeRoC00L · Accepted solution]

Yes, read only access is sufficient for backups (restore directly of a VMDK will not work).

I have it running (on a HP P2000 G3 FC SAN).

[Colin_Weaver]

I take it you never intend restoring using SAN?

[ALiving]

You do realise read only is only a toggle, right? a very simple change that can be made when required to restore a VM using SAN Transport, which will be documented in our restore checklist. Other than that, we can use NBD without modifying anything. For the other 99% of the time when we're not restoring VM's, the disks are protected from accidental writes.

I can't believe this isn't a more often asked question? Am I wrong in assuming that most people would have reservations in presenting a live non native file system to a Windows machine in read write mode? The chance of human error is huge, espescially in a team of more than 1 person.

[ZeRoC00L]

Correct, I have once seen that the VMFS volume was destroyed because the disks were mounted to the backup server.

I realy prefer read only, restores can be done via NBD transport method, as they only occur incidently.

[teiva-boy]

Until VMware updates their best practices, or changes the vStorage API's.  Folks are not going to make their own guesses on reinventing the software.

Try it yourself if it's such a simple toggle and see what breaks (if any).  Personally with good documentation and change controls, the documented current/preferred/best practice works for tens of thousands of customers today.