I am not sure if this information will help here but I am running 2 different
pools of desktops. The first is a basic desktop with JUST Microsoft
Office 2003 on them. The hard drive size is 5 GB. The ESX server
they are on is at ESX 3.5 Build 153875. The second is part of a Cluster
made with a Fujitsu BX 600 (2 blades). The desktops here are 8GB drive
size and have a few more applications on them. The network connecting all
of the servers are GB except for the Fujitsu stuff is part of<span style="border: 1pt none windowtext; padding: 0in"> </span>ether channel environment but all network ports
are in a GB port. My question is why my clients would be seeing slow
performance. This is mostly noticed at outside locations that have a t1
or dual t1 back to the main location that has the vm’s hosted. What are
some areas to look at as potential problem?</span>
Any and all help would be great.
For users on slow WAN links their performance will degrade as the link gets saturated.
All my links are point to point T1's and smoe are dual. The responce time for other applicaitons and processes are good. Its just this.
Here are some things to look at:
Are the users seeing slow performance overall or with a particular app they are using in the VM?
What is your link utilization on these T1's? You may need to look at implementing a QOS for RDP traffic.
It could be a variety of factors. Take a methodical approach and eliminate:
1. View Config
2. ESX config
3. Desktop VM config
4. Application
5. network
Also, make sure your master image vmdk is aligned with the underlying storage. I have seen performance and app problems when the vmdk is not aligned.
Follow VM best practices for virtual desktops OS config.
Hope this helps.
How many users are on the far end of the WAN links? How many are VDI? What traffic besides RDP travels through that pipe?
Depending on the operations RDP could be generating more WAN traffic than the apps would if installed on physical boxes at the remote site. Also, congestion is going to have a more significant impact on the users experience when its the entire desktop that is lagging. I would suggest mjsvirt's suggestion about implementing QOS.
Though, at this day in age I dont think 1.5-3mbits is all that fast.
We experienced performance problems which were caused by Internet Explorer proxy settings.
A full report can be found here: http://www.vmguru.nl/wordpress/2009/06/vmware-view-performance-issue/