Move 2008 DC to a new Server.

Dude, i have given you advice not to P2V a domain controller. Other servers sure go for it but your DCs are a bad idea. If you want to disregard this advice then go ahead.

If you want to understand the issues go ahead and do it, just make sure you do it in a test environment.

I dropped in this advice to prevent you making a mistake or someone else from following your post and making that mistake, im not prepared to have a long discussions on all the pro's and con's.
 
And I advised based on significant experience convertig a large number of servers of every type including domain controllers from nt4 up. We'll just have to disagree on the approach
 
P2V is not recomended for production servers? That's 1 of the funniest things I've read on here in ages.

If your getting that sort of failure rate I'd suggest you get yourself a better process as I've done 100's over the last couple of years with a sub 5% failure rate on 1st try and that is usually caused by boot cd issues rather than the P2V process.

P2V is the lazy way of doing it - great for certain things, legacy apps, testing etc but for a DC is completely stupid and other servers I'd always investigate other avenues first.

Certain things work OK, and to some extent certain source servers have better results than others. For example, HP G4 servers as a source seem to yeild very variable results, especially in performance terms. Terminal/citrix servers are another one that it is a VERY bad idea to P2V
 
P2V is the lazy way of doing it - great for certain things, legacy apps, testing etc but for a DC is completely stupid and other servers I'd always investigate other avenues first.

Certain things work OK, and to some extent certain source servers have better results than others. For example, HP G4 servers as a source seem to yeild very variable results, especially in performance terms. Terminal/citrix servers are another one that it is a VERY bad idea to P2V

I agree with TS/Citrix boxes, primarily as a VM'd TS/Citrix server will have a different usage and scaling profile than that of a physical piece of tin.

But I have never experienced issues with performance based on legacy drivers or other wise. I've done HP G3, G4, G5, Dell, IBM just about everything I can think of hardware wise.

I disagree with you just get over it, P2V'ing is an standard, accepted and tested way of converting servers, DC's or otherwise. It forms a significant part of the Virtualilsation deployments and processes that I design and which are implemented for customers. In some circumstances there maybe be no need to P2V but in the vast majority it is quicker, less disruptive and more cost effective to do it.
 
We'll agree to disagree, but I really see no benefit in P2Ving it in this case - building a new one will involve a lot less downtime and is one of the simplest things in the world to do, any sysadmin should be able to do it.

The problems I've had (namely with servers such as Exchange and SQL boxes) when P2Ving have mostly been performance related. They were all done cold and all HP software removed afterwards, but just didnt perform as I'd expect them to. Never got to the bottom of why, no performance metrics in VMWare or the guest showed any problems but they were genuinely a lot slower. I'd love to have been able to get some benchmarks but unfortunately getting the problem sorted was the priority.

File server was a peice of cake - just built a new box and attached the virtual disk from the P2V'd box to it. Browsing file shares and opening files instantly became a lot quicker. Logon times (roaming profiles) pretty much halved.

Exchange server was just as simple. On the P2V'd image, it got to the stage where it couldnt complete an incremental crawl of the index ovenight (starting at 7pm, store size was circa 30gb). Built a new one, scheduled the move of the mailboxes overnight and performance went through the roof there.

Oddly, a Progress server that was on a Dell 4600 box performed absolutely flawlessly.

I'm not saying the P2V process doesnt have it's place, just that there are other options that may be quicker, more suitable and less risky. For example a clean build of an Exchange server allows you to easily test it without affecting your users, then move them over when you know it's working properly.
 
Back
Top Bottom