CASE:
VDI environment at customer site with an average user base of 300 clients.
Issues: performance (duuuh!)
Consultancy:
January 2011: dear customer, buy more spindles please!
January 2011: dear customer, can we please change from view4.5 to 4.6 so we can use PCoIP?
The customer:
January 2011: $$$$$ …. NO
February 2011: $$$$$ …. NO
March 2011: …
December 2011: please solve our performance issues before the end of the year!
Consultancy:
(thoughts) ARE YOU KIDDING ME?
(answer) yes mr customer, we will give you our own hardware and prove you wrong. On the other hand, we are allready one year further so we offer a new solution to your issues: 2 new blades wth FusionIO cards and VIEW5.
============================================
Results of POC (Proof-Of-Concept).
VIEW5 setup on vSphere5 with Win7 linked clone image and no more tweaks then necessary. For the VIEW we used the VIEW 5 evaluators guide (download here). For the Win7 image we used the VMware View Optimization Guide for Windows 7 (download here)
Test 1) go to www.wechoosethemoon.org and launch the Apollo 11, just 40 years later. you can view the result of both sessions below.
Session 1: VIEW4.5 with RDP
Session 2: VIEW5 with PCoIP
Session 2: VIEW5 with PCoIP
Test 2) the test above is still user experience so I did the test twice and added a powerpoint show of over 5Mb. I captured these sessions with Wireshark (filter: host ip) and this is the amazing result!
Disclaimer: this is “only” a proof of concept on really old hardware (10 VDIs on 2 x HP BL460G1 with 12Gb RAM each) so it only shows the difference for this configuration and is not a benchmark to be used. There are too many different parameters that are changed. The result here is a sum of all enhancements:
* Client Operating system from Vista to W7
* hypervisor from vSphere4.1 to vSphere5
* VIEW 4.5 to VIEW5
* RDP to PCoIP
Lessons learned: should have done this a year ago …