VMware Cloud Community
idingsdale
Contributor
Contributor

Performance of P2V'd HP DL380

A little while ago, I P2V'd some of our DL380 G4 servers onto other 380 G4s. They are Dual 3.6Ghz boxes with 8Gb RAM.

After converting them (mostly cold conversions), we've seen a large and steady decline in performance. This was so bad, I set about doing a clean installation of the machines from scratch on ESX natively. It was so bad it was nearly unusable (eg shared printers would take 30 seconds to open the properties of). After a rebuild, the performance shot through the roof, so I'm fairly sure there's a problem with the conversion process, or some driver left behind etc. Oddly, no performance metrics within the guest or within VC showed any bottlenecks or problems.

Has anyone else done P2V conversions on these, or similar boxes and had terrible performance within a VM? If so, did you manage to find the source of the problem without reinstalling from scratch? I've disabled a lot of the HP management services, but I dont want to go uninstalling things without knowing the result.

I've done a rebuild on all but one of the servers (stuff like Exchange and the file server were straight forward), but our SQL server will be a real problem as there are lots of bespoke apps that will need external companies to be paid to reinstall.

Any advice would be hugely appreciated

Tags (3)
Reply
0 Kudos
13 Replies
christianZ
Champion
Champion

External or internal storage there?

Reply
0 Kudos
idingsdale
Contributor
Contributor

All internal. 6 x 146gb 15k disks. 2 in a mirror for system volumes (this is where esx is installed) and 4 in a RAID5 for data

Reply
0 Kudos
christianZ
Champion
Champion

For testing could you clone one VM - how fast is that.

What tool have you used for p2v?

Reply
0 Kudos
idingsdale
Contributor
Contributor

Do you mean how fast is the clone process, or clone to another machine for testing?

It was the VMware converter boot CD that was used

Reply
0 Kudos
christianZ
Champion
Champion

I mean cloning on the same machine.

Were the cloned server hp machines; have you cleared them from all the hp software/driver things?

Reply
0 Kudos
idingsdale
Contributor
Contributor

They have all been cloned from Dl380 G4s to DL380 G4s, so identical hardware.

I disabled the HP management services, but I dont want to just start removing things unless I know it will fix the problem. Was hoping someone might have seen this before and know where the problem lies.

Reply
0 Kudos
christianZ
Champion
Champion

Definitely saw this issue - but it you not sure then you can make a snapshot (by vm offline) from that vm before starting it.

Reply
0 Kudos
christianZ
Champion
Champion

I guess the Vmware tools are installed there...

Reply
0 Kudos
idingsdale
Contributor
Contributor

So you reckon removing anything HP related will sort it?

Yes, VM tools are installed and up to date..

Reply
0 Kudos
idingsdale
Contributor
Contributor

Scratch that, looks like someone's already had the idea of removing anything HP related. The only thing left is the agent for our HP UPS....

Reply
0 Kudos
christianZ
Champion
Champion

...then check the cloning speed now

Reply
0 Kudos
christianZ
Champion
Champion

...and generally I would prefer to make a one array with all the six disks and then make one small logical drive (e.g. 20 GB) for Esx as R1 and a second or third logical drive as R5 for your vmfs volumes.

Which raid controller have your there - w/ BBU, write caching on?

Reply
0 Kudos
idingsdale
Contributor
Contributor

It's a smart array 6i, I believe all the caches are enabled etc.

The host itself is performing fine other guests are no problem, its just this one that has been P2V'd..

Reply
0 Kudos