Migrating a large amount of data via P2V

[bshubinsky]

We're planning on migrating a few very large servers via P2V into our new VMware environment. One of them is our file/print server. It's roughly 300-400GB of data. What was the process you went through when migrating the data over? Did you stop certain services to keep people from changing data while it was doing the P2V? Did you remove the system entirely?

Any thoughts or suggestions would be appreciated.

[BenConrad]

I did a bunch of 800GB & 1TB SQL servers late last year. I found that the P2V tool takes way too long when dealing with this amount of data. In addition, one of the huge P2V's completed but had a few databases that were corrupt.

This is what I do with big P2Vs:

1) Shut down all services that provide access to the data on your disks, for a file server you may want to either set all shares to read-only, disable the shares or shut off the 'server' service on the box. I'm not sure if 'Server' is required for the VSS snapshot Converter uses.

2) Create a helper VM and attach it to the destination .VMDK or RDM

3) Robocopy sourcedir destdir . /E /SEC from the source machine over to the helper

4) Start your P2V and don't select the disk(s) from step 3 above

5) Configure your P2V using your post configuration tasks

6) Re-run the Robocopy to make sure no files have changed, should all be skipped.

7) On the helper VM, remove the .VMDK or RDM

😎 On the P2V'ed VM mount the .VMDK or RDM

9) You are in business!!!

Now, if your file server already has volumes on the SAN you can skip most of this stuff and just unmount and remount to the P2V'ed VM.

Ben

[rogard]

I would probabaly build another file server.

Whenever I am doing P2V conversions, I will always build a separate box, move the data across then change the "pointer" to point to the new box.

That way if anything goes wron you can simple change the pointers back.

In your case the "pointer" is a mapped network drive? So you could move them over department by department using group policy to change the mappings as you go.

Or use DFS etc.