I've done some testing on VMware vs. XenDesktop and for our situation VMware just couldn't compete on the broker side. (Hypervisor side is another story). Our major sticking point was RDP. We have a long distance high latency link and RDP is just not usable over that link for day to day work. We attempted to test locally using a latency generator and I couldn't stand to use it for more than about 10 minutes.

Locally without the high latency I much preferred the VMware solution but RDP vs. ICA was a deal breaker for us. I think that may be one of the biggest benchmark concerns in many cases. We weren't really concerned as much with how many desktops can we fit per processor, etc...

Point taken, it was just that with 'standard' implementations the broker was the piece that we could change back and forth to test without customizing too much. Basically we have a significant VMware investment and we weren't interested in going another way with the hypervisor portion. The broker was sort of our generic term for everything beyond the virtualization infrastructure (probably wrong choice of terms but easiest for everyone to understand).

You are correct on the other comments, and we've been patiently awaiting the (software/virtualized) Teradici version to take a look at. as well. We are looking to go with a single solution and not solution A for local users and solution B for remote/home users.

Are you willing to share more about what is important to you regarding WAN performance?

Absolutely...

You felt PortICA out performed RDP with higher latency. What level of latencies are you looking to be able to deliver across?

Our testing is based on round trip times of 280 ms.

Is that only delivering a typical desktop experiance with office apps and some web browsing or does that also included themeing and visual affects turned on? Or does that also include a more advanced desktop experiance content such as Multimedia, Flash etc..

For the most part pretty typical. I tried a 'standard' build for us and then also tried it using the recommendations for VDI implementations like some effects off, themes off, plain backgrounds, etc. That didn't make a huge difference really for what we did. My testing was not extremely scientific, but I did the same tasks on each solution and timed them. They weren't all specific tests to the applications we are using, but more general using some web, office, etc... This did include a few multimedia and flash tests.

Also for each desktop experiance what price are you willing to pay regarding bandwidth needs per user is that important?

I'd like to get a remote user to have a local experience at no added cost . Ok, back to reality I think for the most part we are trying to squeeze what we can from current bandwidth without having to bump up the size of that pipe. One of our main goals is to do this without added hardware in the area of accelerators or solutions that involve back end hardware other than the ESX infrastructure (like the Teradici hardware solution or Sunray type scenario.