We help IT Professionals succeed at work.

citrix, desktops

pramod1 asked
I want to gain understanding on citrix with regard to citrix provisioning method: PVS

Is there any good article which shows below related to PVS

1) merge vdisks. vdisk update
2) vdisk versioning
3) creating new vdisk version ( deleting the bad patch )
3) how roll back takes place on account of corruption on desktops
Watch Question

Senior Citrix Engineer
There are tons of blogs, support articles and documentation on all of these subjects.  A few minutes on google will bring up tons of great stuff.  Any of the big names in the Citrix space will be helpful.. (Daniel Feller, Doug Brown, Jarian Gibson, Helge Klein, Francois Xavier-Cat, Carl Stahlhood etc.).  

2. Versioning - you select your disk, right click (or menu click depending on the version), select new version.  That's really the guts of it.  In the background, a new .avhd(x) disk is created in read/write mode.  You assign that new version to a machine, make your modifications and seal it up.  It looks scarier than that, but it is not.  You can maintain several versions (each version extends the chain).  However, when you get above 5 disks in the chain, the performance hits become much more noticeable.

1. Merge disks - It's been a while since I've done, this, but you select the top level disk you want, right click, select merge, and the .avhd(x)'s are merged back into the main disk.  As far as I remember, the disk must not be in use for the merge to take place.  The more disks in the chain, the longer it takes - plan for it.  (It can be as little as 30 minutes to several hours).
1.a. Update disks - this is really separate... but after the modifications are made, and the disk is sealed up, you promote the new version to production, assign it to the correct machines, and when they reboot, they will reboot on the new image. You also have the option to force them to reboot immediately (typically only done for an emergency patch).

3. New Version - this is the same as #2.  it is not different.

4. Rollback - this is the crowning jewel of PVS.  Your images are read-only.  So, in the situation where you have multiple versions, you select the previous version (or whichever one you want in the chain) and promote it to production.  Assign it to the devices, reboot them and they are back up.  It is literally that easy.

The other disk management method is just the manual copies.  When you want a new version, you copy the files (the pvp & vhd(x) files) and the copy is your new version.  When you want to roll back, you just assign the previous copy & reboot.  If your updates go bad, you just delete the files, recopy & try again.  It uses a lot more disk space, since every image is a new copy of the disks, but it is simpler.

A few important notes about using PVS:
  1. When the disk is in use, it is read only - you can copy it off at your leisure... When it is in private (read/write) mode, you can't copy it if it is in use.
  2. Always use thin provisioning - your disk is read-only when in use, so it will never expand while in use. So, your 100GB disk may really only be 30GB on disk
  3. Put your production disks on a share (preferably NFS), and connect to that share. If you do this, and put as much RAM as you can in your PVS servers, and it will perform faster.  As your disk images get used, the PVS servers will cache the data in memory.  As the image gets cached, it will go to disk less & less.  (a typical Win7 image will use about 700-800Mb of the disk and that will be in RAM).  If you go this route, then the storage can be slower drives and the performance really won't matter except during update times.