We have a physical vCenter server that is running out of space and no way of remedying that situation, so we are thinking of either rebuilding it or migrating from P2V.
What are the pros and cons and has anyone done this before?
What issues did you run into?
Which did you do and why?
The other idea we played with was possibly just running a VCSA, but we would have to build it from scratch, correct?
What version of vCenter are you currently running on the physical server? Is the database external or on the same machine?
We're running 5.1 and will probably upgrade to 5.5 after the migration or rebuild on 5.5. DB is on the same machine.
There is no migrating to VCSA if your DB is on the same machine from what I recall.
OK, so you guys are due for an upgrade anyway. If there is a risk adverseness (is that a word?) to going straight to 6.0 U1, I would say you are still setting yourself up for the future push to 6, by going with the 5.5 VCSA. Make sure you document your domain and SSO Site Name, that way when you do go to 6, you can use the CMSSO-UTIL command to repoint your old VCSA server at new ones should you decide to go with external PSC's in the future. Basically, try to be as forward thinking as possible having the advantage of knowing the architecture behind 6, even though you are going to 5.5.
My advice, in short, is to go with a new deployment, use the appliance, and get away from the Windows w/ SQL setup. It is far simpler (albeit not perfect) way to go. It is also the better direction going forward.
Just the typical wait 6+ months before going to the newest release is what not going to 6.0 boils down to. Not my call on that, but 5.5 has been around long enough to where we are comfortable with upgrading to it. I would like to go to VCSA just to avoid having to eventually move to it.
If possible, I would try to talk the powers that be into going with VCSA 6.0 U1, but holding off on the hosts for a few more months, and then going straight to 6. I know how difficult that can be though. Either way, VCSA, in my opinion.
I used to be a big advocate of the physical vCenter and DB server. I felt that keep it separate from the virtual environment brought an added layer of resiliency to the network. I still do in some regards but now feel it is very dependant on the type of environment you manage. I used to manage a virtual environment that had an over-provisioned SAN, regular P1 outages, network failures. A physical vCenter really helped.
Now I manage dozens of vCenters and keeping them all physical isn't practical. The infrastructure is much more stable and redundant that running them all virtual is definitely the smarter move.
I'm a big fan of the VCSA. No Windows licenses, no SQL license. Soooo much easier to manage.
originaluko .... There are a couple of nice compromises that can be made to keep the management environment separate from the very infrastructure it is managing. I understand this mindset, as I share it with you. I use blade servers, but run all of my VMware management on standalone servers. vCenter, vROps, vRA, etc. It's all on those two nodes, which have independent network connections and an independent SAN connections (you could even get a little VNXe or NetApp) to keep the storage totally separate (or use hyperconverged for management I suppose). Point being, you can meet your desire to keep them isolated from the rest of the environment while still running them as virtual.
In addition, if you have moved to vSphere 6, you can have external PSC's, with or without a load balancer in front of them, along with vCenter in Enhanced Linked Mode, to provide yet another layer of resiliency. To me, the extra hardware outside of the blade chassis is an easy sell, because eitehr way the VMware products are going to consume hardware. Blades tend to be more costly, so why not run those products on standalone hardware?
Just a thought.