Last year, Project VRC invited people around the world complete an extensive survey that asked them to describe their desktop virtualization environment. 662 people took part in the survey, and since it closed, Project VRC has been busy making sense of the data. That information was made available today, and after looking at it, I want to share some of my observations. There are so many useful data points that I can't talk about it all in one article, so I invite you to visit the Project VRC website and download a copy of the results.
Before I get into the data, I thought I'd share some background on Project VRC. It was started in 2009 as a collaboration between PQR and Login Consultants to develop best practices and collect information on the desktop virtualization space. Ruben Spruijt (@RSpruijt) and Jeroen van de Kamp (@TheJeroen) (of PQR and Login, respectively) are the founders of Project VRC, and, along with a team of like-minded geeks, have compiled and shared much information over the years through white papers, presentations, and community involvement. These guys–and I mean this in the best possible way–are crazy! I don't know when they sleep, or when they stop thinking about desktop virtualization. The end result, though, is some really awesome information. (So thanks, guys!)
On with the data
I took a ridiculous amount of notes when reading the survey, but I can't possible write about them all. It's broken down into many sections, including VDI, SBC, comparing stats between the two, and explaining how Oracle stacked the deck. Well, they didn't say "stacked the deck", I did. Let's just say once Oracle learned about this survey, the number of Oracle responses skyrocketed. This information has been sanitized, and the Oracle numbers in the survey are believed to be accurate. We had the same problem (not with Oracle, though) when we polled our readers for the vendors that we should include in Geek Week, when we had a sudden influx of votes from a single IP address that belonged to Symantec :)
Let's look at some key observations:
There's no surprise here that VMware makes up the vast majority of hypervisor usage, with vSphere 5, vSphere 4, ESXi, and ESX making up 63% of the responses. Surprisingly, Hyper-V 2 is in use in 9% of organizations, which beat out XenServer's 8% share. They also note that 10% of the organizations are migrating to Hyper-V, which I have to say isn't all that surprising. Microsoft has made many improvements, and has all but declared war on VMware in that arena.
While we don't know the percentage of people migrating away from Hyper-V, the key takeaway here is that while MS was bringing up the rear, they are now the third-place hypervisor, and people are actively migrating towards it.
47% of companies have no WAN optimization at all in their environments, which isn't surprising. Those that do are using Citrix, Cisco, and Riverbed, followed by F5 and Juniper. Citrix is the clear favorite there, but with almost half the respondents doing nothing at all, there's a lot of room for growth with regards to the vendors and a lot of room for improvement with regards to the user experience and WAN utilization.
It's no surprise that XenDesktop has a 44% share of the connection broker space, or that View is second with 27%. What is surprising to me is that Oracle came in at #3 with 13%, beating out Microsoft (6%) and Dell vWorkspace (just 3%). Perhaps this is due to Oracle's marketing efforts (that's the word Project VRC used), but it could also be the loyal following that Oracle VDI has. It will be interesting to see this same statistic next year, which we can use to gauge how Microsoft is doing with RDS in Server 2012 and to see what, if anything, Dell is doing with vWorkspace.
Probably the most interesting aspect of this is that 32% of respondents were using stateless VDI as the primary desktop platform, while another 36% had at least some stateless VDI in their deployments. Those numbers are way out of line with perception, and I can't wait for Project VRC to mine that information more. In the future, they'll slice up the data in other ways to try to get an idea of what's going on. I have personally seen an increase in the use of stateless VDI when I ask the question during the VDI Road Shows that I give, so I'll be curious to see the results.
Not shockingly, SBC is in widespread use in 85% of respondents organizations. 40% of respondents said that SBC was their primary method of desktop and application delivery, while another 35% said it was in use alongside traditional desktops. Around 2/3 of people said that they use XenApp 5 or 6, followed by Windows Server 2008 R2 with 13%. Oracle and Server 2003 both came in around 4%.
Just over half of the people said that they use Windows Server 2008 R2 x64, while Server 2003 R2 x86 was in second place with 23%. This isn't surprising, since Server 2008 x86 is essentially Vista Server. I don't expect to see that number drop until we get closer to the EOL date for Server 2003, which is July 14, 2015. Until then, we might even see increased use in Server 2003 as a way to extend the life of Windows XP applications past April 8, 2013.
33% of respondents are using their existing storage for desktops, which we've argued against in the past because desktop I/O is so much different than standard I/O on storage arrays (or even that of server virtualization). It could be that their storage was already optimized for desktops, though, which I imagine would work just fine for everyday use as well. 42% have a dedicated storage solution just for their desktops. The vast majority of them are using well-known vendors. Additionally, of the companies using VDI, almost 3/4 of them are using centralized storage as opposed to local. I wonder if it's possible to mine from the survey the breakdown of stateless/persistent combined with centralized/local storage?
Citrix Provisioning Services beat out VMware Linked Clones, full clones and Citrix MCS for the top method of deploying desktop images.
User Environment Management
The leading method of managing the user environment is simply GPO (35%), followed by custom scripting (18%), then Citrix UPM (14%). RES, AppSense, and VMware View Persona all came in under 10%, while Dell vWorkspace, Liquidware Labs ProfileUnity, VUEM, and Scense were each used in less than 3% of environments. While it doesn't look like there's heavy UEM adoption, there are so many vendors in the space that it can appear that way. The clear favorite among third parties is Citrix UPM, but when you combine it with the shares of all the other solutions, organizations are roughly split down the middle on whether or not they use third party UEM.
A common thread throughout the survey is that companies place a high importance on user experience. In fact, the top four "innovation areas" in the survey have to do with User Experience (WAN, mobile performance, unified communications, and rich media). Still products, technologies, or methods that improve the user experience aren't highly used. Almost half of the responses indicated that there is no WAN optimization, and over half of the responses show that companies are simply using GPOs and scripts (although you could argue that people don't see UEM as an innovation area). That can mean one of a few things:
- Those are enough and all the users are happy
- The users aren't happy but IT doesn't know/isn't willing to fix it
- The third party solutions aren't good enough to justify the expense
- Awareness of third party solutions is low, or organizations are still just getting the lay of the land
If user experience is truly king, then you would think these solutions would be in more widespread use. I guess there's room for both the vendors and the organizations to grow, and since the key innovation areas called out are more about mobility and connectivity, it appears the public perception of what is needed gels pretty nicely with what we think (hooray!).
There is so much more information in the survey results that I cannot encourage you enough to download it and read it yourself. This article has only looked a comparatively small amount of information. There is detailed stats on SBC, application virtualization, load testing, web applications, storage, networking, server hardware, antivirus, and the vendors associated with each of these aspects of desktop virtualization. You can download the white paper, and check out all the other stuff they do, on ProjectVRC's website. Thanks again to Ruben and Jeroen for all the work they put into this!