2011-01-25 11:25 AM
The Netapp and VMWare vSphere Storage Best Practices guide has as the recommended layout (page 70, link below) to locate the vmkernel swap file on a separate datastore from each VM. I have done this over a year ago.
This same guide on page 72 has an optional layout where the Windows pagefiles are also located on a separate datastore. How many of you create a Netapp volume, add this volume to each of your ESX hosts, add a disk to each of your Windows VMs that use this volume for storage, and then relocate your Windows pagefiles to this disk?
2011-01-25 11:19 PM
We do. :-) Although we are looking at relocating the vSwap file to the default location, because it doesn't work for us. Problem is that you have to place all vSwap's to the same datastore within a cluster, since you can't specify multiple datastores.
But placing the Windows Pagefiles on separate datastores works fine for us.
2012-10-30 01:12 PM
Do you happen to use SRM 5.1 and if so, do you replicate the windows pagefile swap datastores? And if you do not replicate them, how do you set it up so the drive letter works correctly on the protected site and also when they fail back to the primary site?
2012-10-31 02:24 AM
Sorry, no, we don't use SRM at all, but I understand your dilemma. If we were to use SRM, we probably would replicate them infrequently, maybe even once.
2013-10-03 06:19 AM
I've recently configured SRM 5.1 for a client and I did break out the page files into their own datastore (as well as the vswp files). In short, here's a link that describes the process: http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2009324
Here's how I copied the page file template .vmdks to the recovery site: http://www.vhersey.com/2013/05/copy-files-between-esxi-hosts-using-scp/
Honestly, it's turned out to be quite a task. Yes, it saves bandwidth and time not replicating transient data, but at a price. The initial time to set up, especially for multiple VMs with different amounts of memory, is not trivial. Then, of course, the client should change their operational procedures for deploying replicated VMs.
As far as drive letters, when the VM comes up at the recovery site, it will look for the same drive letter. The recovery site page files are cloned from the same template the the protected site VM is using.