Dualing iPads

VMware View vs Citrix XenDesktop: Dualing iPads 

So the bequest of @virtualtal I have been working on comparing the Citrix Receiver running XenDesktop 5.5 to the VMware View Client running View 5.0.  In the previous test I did of this I actually compared Citrix XenApp 6.x to VMware View 5.0.   It wasn’t my intention to compare these very different products, I just happened to be onsite with a Customer and Citrix was leading with XenApp and VMware was leading with View and the customer was nice enough to let me record a video using my iPhone.

Citrix XenApp vs Vmware View

The video was crap, I’ll be the first person to say that, I just wanted to capture the moment.  In the spirit of openness and fairness, Citrix let me know that they released a new Citrix Receiver (this is old news by now) and asked me to shoot a new video comparing the two.  I promised to do so and although I’m late to the party, I just got down shooting the video.  This post is about my experience shooting the video. 


Lab Facts:
Both VMs are running on the same back end storage and the same ESXi host.  The VMs are identical from a virtual hardware perspective.  They are not identical from a “whats installed on them perspective”.  That’s because I am doing this video as a favor, and I didn’t have time to build two new VMs.  So the Citrix VM is a clean VM, the VMware VM is just the VM I demo often and is full of junk.  (I don’t feel it will have any impact on the results.)
Here we go:

Big Buck Bunny (I’ll refer to this as BBB)
  • Frame Rate:
    • The one thing I notice right away is that Citrix initially looks better than VMware.  In the first 30 seconds of the BBB video Citrix is much smoother.
      • Now the point of my site is to share both products so I’ll explain this difference.  PCoIP as a protocol is dynamic and wants to use as little bandwidth as possible, so initially Citrix looks better because PCoIP has to ramp up to the required bandwidth.
    • Throughout the BBB video I’m pretty sure Citrix is running at a higher frame rate (this is a good thing, more frames means better experience).  There are occasions that it skips, but on average Citrix seems to be giving a better experience.
  • Performance
    • Performance in a lot of ways is how you measure it, if we are measuring in frames per second, Citrix wins.  If we are measuring in picture quality, hands down it goes to VMware. 
    • Throughout the video there are scenes that just get really fuzzy/blurry with Citrix. I tried to zoom in on this throughout the video but my camera just wasn’t good enough to display these artifacts.  What I think I’m seeing here is just the nature of Citrix.  Citrix doesn’t have a pixel perfect state like VMware does, this isn’t a bad thing, its a good thing (see bandwidth below).  However, in the day and age of HD videos for everything, its only fair to point out that the video is blurry/fuzzy for Citrix.  PCoIP on the other hand looks like an HD video.
    • Summing up: Citrix appears to have a higher FPS, so it is less jittery.  PCoIP is more jittery but better video quality.  
      • This can actually be adjusted on the PCoIP side, but as this is an out of the box comparison I didn’t disable “build to lossless” which would increase the FPS and decrease the video quality which in turn would lower the bandwidth.
    • Which video did I prefer… Citrix.  Which just goes to show VMware should have “build to lossless” disabled by default.  In a manner of speaking, Citrix does.
  • Bandwidth:
    • My bandwidth results were kinda worthless.  I didn’t realize at the time that Citrix didn’t have audio going and VMware did.  Citrix had impressive numbers, 10-20% less bandwidth, but honestly high quality audio could have been the difference.  Also, I know from experience if I just disable build to lossless on the VMware side, both videos would be fuzzy and the bandwidth would be about the same.
    • Aren’t we all past this bandwidth argument by now…  there are better discussions to have than the subtle differences in bandwidth between these products.  For instance where are the WAN accelerators for PCoIP (Riverbed that comments aimed at you.)
You have no idea how much I hate Big Buck Bunny.  Its not the video, it’s that I’ve seen the video thousands of times.  We use it for demos because its copyright free and its animated which always looks good when you are doing video posts.  Well, lately I’ve been moving into using this Landscapes Volume Two video by Dustin Farrell.  (http://vimeo.com/dustinfarrell) The reason I use this video is becuase its stop motion which means every frame is in perfect focus, basically its the perfect video for testing picture quality.  Plus its much less annoying than BBB, so far I’m not sick of it.
Landscapes Volume Two
  • Frame Rate:
    • PCoIP rocked this video.  I’d expect the first 30 seconds in this video to be just as bad as BBB, and honestly they probably are, but Citrix chokes so bad on this video its hard to compare it to anything.
    • I will most likely change this part of my post after talking with some Citrix engineers about it, but the LV2 is an AVI and I’m wondering if Citrix is just having a hard time with the video format.  Normally I’d pawn it off on client side rendering (that’s normally why you see such big differences in video for Citrix), but I know that BBB was host rendered (an easy test is to take a screenshot inside the VM).  BBB was in MP4, LV2 was in AVI, if you know something I don’t know let me know.
  • Performance:
    • I mentioned some artifacts in BBB, but I still said I preferred the Citrix experience over the VMware experience.  In this video the artifacts were horrendous on Citrix, but VMware did a good job with it.  I think it might be due to the video being shots of stars and being at night.  Blacks are hard for video codecs like DIVX so I can imagine they would be hard for ICA/HDX as well.
  • Bandwidth
    • Again, I didn’t really get into this much.  At a quick glance it looks Citrix wins, but Audio was working on VMware View and I didn’ thave it working on Citrix, so my numbers are worthless.

In the end, I was very impressed with the performance bump on the Citrix side.  I’ve done these comparrisons before and Citrix never looks passable on an iPad.  Whatever changes they made to the last client made all the difference in the world.  I think VMware is more versitle in video formats, but when Citrix can handle the video it does an outstanding job.  That’s not to say View does a bad job, it doesn’t, View does good, but Citrix just does it a bit better.


As always, I think both products are great.  My conclusion is to test this for yourself on videos that your company may need.  I think you’ll find your results will be about the same, for some video Citrix will do great, others not so much.  VMware will be consistent, but not perfect.