So I am in a POC (like most companies out there) with a VDI deployment. One user group for the VDI pilot is located within branch offices of the company, connected via T1 (1.5 mbps) links. So the purpose of this blog is to reveal my WAN test results with VMware View vs Quest vWorkspace. This blog should not be considered an industry standard or a technical guide, it is just my observations from testing in my lab.
One thing I really love is a good virtualization challenge. Whether it's virtualizing RightFax with Dialogic SR140's, custom applications that need physical modems (which I easy EZ DialOut software for), or anything that requires access to physical devices via VM directpath. My new challenge is to use serial device converters to USB with VDI. Particullarly serial coin counters, cash counting machines and serial printers.
I'm sure you see that I mentioned serial devices and you might say "hey, you can't use serial over PCoIP!". And you would be correct. I am using SeaLevel serial to USB converters (which is the only supported converter for the app I am using on the VDI session). I don't think that SeaLevel officially supports the configuration that I have, but they have a great support staff that helped me out! Using a Serial over IP device was not an option in my scenario. I also did not want to revert my PCoIP connection to RDP so that I could use serial over RDP. Thin Clients with serial ports are not cheap!
I currently have a close relationship with Quest vendors, so I decided to give vWorkspace a shot along side my View 4.5 VDI tests. I have worked with VMware View for some time now, but I had limited knowledge of what Quest vWorkspace had to offer. I worked alongside a great engineer that helped me get the product off the ground so I could do my testing. My first impression of vWorkspace was "look at all the options"!
The user experience? Well, vWorkspace did seem a little choppy compared to PCoIP. But that was only true when trying to surf web pages. In this particular group I am testing VDI for, web surfing is not a business critical, so I'm not so concerned about that. My main concern in testing both products is "how much bandwidth is each session going to take?". Since vWorkspace leverages RDP, the USB traffic is in on 32111. In VMware View, the session traffic is over port 443 and the USB traffic is over port 4172.
From this graph, you can see that the USB traffic in VMware View is a monster! The session traffic over 443 is not that bad at all. I found it pretty odd that the non-optimized traffic was better than the optimized traffic for vWorkspace.
What did my tests include? I had 5 laptops with the client installed for each connection broker. All clients connected to one connection broker at a time. I surfed maybe 3 or 4 web sites on each session at the same time to see how responsive they were. I then concluded with testing my USB devices on one session. The app that accesses the USB devices is just a small 16bit app, nothing fancy.
So what's the next step? I'm going out in the field with both products to test. Then we will take a serious look at the cost of both products and also see which product fits our performance needs. Not all user group for VDI are the same, so we may end up with a mix of both.