Windows 2008 - Full Fat or Diet?

Out of curiosity (cats give it a bad name in my opinion) I recently chose to do a little experimentation. No bunsen burners and weird coloured liquids though, I wanted to do a light weight comparison of Windows Server 2008 R2 vs Windows Server Core 2008 R2 in my home lab. Let it never be said that I don’t know how to have fun!

It was a rather simple comparison with a fairly simple aim: I wanted to know if there was a difference in how many resources were consumed on an ESXi host (specifically one of my low power HP Microservers). Certainly not rocket science.

I started off by building brand new VMs for each in vSphere 5. Aside from the OS itself, both were identically configured with:

  • 1 vCPU
  • 2Gb Ram
  • 1 x 40Gb, thin provisioned VMDK hosted on an NFS datastore

I installed VMtools into both, enabled RDP in both and gave them IP addresses. That was it. No updates, no added features etc.

I powered them up on separate, otherwise empty hosts within a few seconds of each other and then sat back to watch.

After each had about 5 minutes to settle down, the above snapshot from vCenter showed:

  1. 2008 R2 Server Core base install uses nearly 4Gb less disk space
  2. 2008 R2 Server Core is uses less host memory (about 81Mb at the time of the screenshot)
  3. 2008 R2 Server Core uses fractionally more CPU (13Mhz)

Of course this was without any applications or additional services running and so the results may be different when the two are used in anger but vanilla installs seemed like the simplest way to compare them.

Server Core seems to use less physical memory and have a smaller disk footprint. Those, in my home lab, are probably amongst the biggest constraints that I have to deal with so in future I'm going to try and make more use of Server Core VMs. As long as the differences in how they're managed doesn’t inconvenience me too much that is.