Technology, Cloud

Navigating the Virtual Desktop Challenge

Blog-post by,

We’re waist-deep in a virtual desktop project here at Menlo College, and are about to jump into the deep end with both feet. This week marks the official public launch of our Virtual Desktop Infrastructure (VDI) initiative – deemed by some to be amongst the most comprehensive VDI initiatives in higher education because of the desktop types and use cases we’re tackling. We started our pilot only 5 months ago and within the next 3 months will increase our virtual desktop environment five-fold, providing us with an innovative desktop infrastructure befitting of Silicon Valley’s business school.

Our project is an ambitious one that will:

  • give students greater access to lab applications,
  • reduce the operational costs of desktop management,
  • minimize the capital expenses needed for PC refresh, and
  • reduce our college’s carbon footprint.


Do these goals sound familiar? I suspect they might. Many institutions are searching for the “holy grail” of desktop infrastructure and support – improved user experience and reduced overall costs. Well, we think we’ve found it.

The core of our technology selection involves Unidesk and VMware. Our brag sheet includes simultaneously deploying persistent and non-persistent virtual desktops for all users and use cases across campus. And, we’re the very first college to leverage Unidesk’s ability to create virtual desktops using only local storage to bring our annual cost per desktop in line with the cost of a physical PC – well under $300 (excluding staff costs).

So how did we do it? I thought you’d never ask…

We began with a substantial pilot, as well as an ambitious one—hitting all use cases at once with virtual desktops in two student labs, the Library, and administration; plus, we wanted to include faculty and student access and knew these desktops would be highly personalized. It is unusual to have VDI pilots move beyond labs, where deployment has been a bit easier because the desktops are almost always non-persistent. But we’re not a huge college, so undertaking VDI for just a few of our use cases wouldn’t let us get the maximum return on our investment to go further with the project. The ROI for VDI only makes sense the more virtual desktops you deploy…so we tried them all.

With our new virtual desktops, we can quickly deploy applications and Windows updates to personalized faculty and staff desktops and to non-persistent lab and library desktops at a moment’s notice. From a pure hardware and software perspective, we found that costs are about the same relative to physical PCs, but we’re already seeing that the amount of staff time and effort required to manage our VDI environment is much lower. And, importantly, access has improved – student labs are no longer bound by the sheer number of physical PCs or by lab space, rather, virtual labs can now be expanded as needed and accessed at any time from students’ own PCs in their dorms or from any thin client or PC on campus.

Now, with a successful pilot under our belts, we’re jumping into VDI feet first.

I suspect many of you are either examining or are already underway with a VDI pilot of your own, and may be able to benefit from our experience navigating these waters. We’ve learned a lot of important lessons along the way and I thought I would share a few with you.

Lesson #1: Know your environment
This may seem like a no-brainer – for any project it’s important to analyze your needs, match your solution to those needs, and so on. And while a VDI project is no exception, I would argue that a VDI implementation requires a much more expansive needs analysis than usual. Among other things, you’ll want to:

  • identify your various use cases,
  • review your software licensing,
  • look into your server and storage capacity, not to mention network capacity, and, of course,
  • fully understand your capital and operating costs.


Once you have a set of assumptions, challenge them. Use cases can be blurry, for example – what you thought was non-persistent may actually be persistent, and vice versa. Not every VDI management software or architecture can handle both types of desktops, so knowing what you *really* need (and want) is critical to your design process.

Lesson #2: Don’t forget the learning curve
VDI can be game changing and, if done properly, can produce tremendous value in the form of improved access for users, better security, or reduced operational costs, or all of the above. But not without some initial pain. Your IT staff will most likely spend more time managing desktops initially, not less, as they overcome the learning curve associated with creating, managing, and supporting virtual desktops. They may have to learn new tools in order to create desktops, and start to understand what symptoms pair with which components of the VDI infrastructure – network, broker, client, or desktop itself. Users, as well, may find interacting with a virtual desktop to be confusing, especially if it is launched on a traditional, physical PC. There’s nothing insurmountable here, but expect that the learning curve will be there, and allocate additional time in your project for it.

Lesson #3: Pick the right technology
There are a lot of VDI vendors out there, with more appearing every day. Make sure you pick the one that’s right for you (see lesson #1). We believe the key to our successful VDI initiative lies firmly with the enabling software we’ve chosen: Unidesk and VMware View.

With Unidesk, both persistent and non-persistent desktops can be supported and managed from the same console. We can also update desktops without “breaking” user installed applications or their desktop customizations, using Unidesk’s “desktop layering” approach. No one else can do that. By sharing the same operating system and application layers across many virtual desktops, we can “package-once, patch-one, deploy-once,” enabling us to update the OS or key application across all desktops in a few hours (or less!). Unidesk’s ability to provision desktops using only local storage and still offer full desktop recoverability was also a key factor in enabling us to implement VDI at such a low cost. 

Where Unidesk enables us to create and manage the virtual desktops, VMware View gives us flexible access to them. It works with pretty much every client we have – repurposed PCs, Pano Logic and 10ZiG zero and thin clients, and even the Apple iPad. I, myself, have been using my iPad more and my laptop less. I open the View client for my iPad, and I have access to my fully customized, Windows XP virtual desktop and all of my apps, including our ERP…all from an iPad! Our students and faculty can also download the View client for their personal machines and access a virtual desktop from anywhere in the world they have an Internet connection – and they love the 24x7 access and View’s ease-of-use.

 

I would like to say that a cross-campus, multi-use case, simultaneous deployment of virtual desktops has been smooth sailing, but that would not be true. But all in all, we have found VDI to be a great adventure and one definitely worth taking. Bon Voyage!

(2) (2)

Discussion
Would you like to comment on this content? Log in or Register.
Tech Marketer
Ali L 0 Points | Mon, 06/11/2012 - 23:03

Interesting article. Doubtlessly, Hosted Virtual Desktop saves money, energy, and space. After you’ve decided to go virtual, take steps to make implementation easier: Get to know some important terms about Hosted Virtual Desktop, types of Hosted Virtual Desktop, and leading companies and products in Hosted Virtual Desktop.

Info. source:

http://www.dincloud.com/hosted-virtual-desktop

Jerry
Jerry Bishop 100 Points | Thu, 07/14/2011 - 01:33

What a great post. I sincerely hope for your continued success. But I lost my enthusiasm for VDI some time ago. And yes, I am the guy who posted Will VDI Get You Fired?

I know a lot of schools are seeing real gains from VDI but the devil is in the detail of the true economics. The reason for commenting is not to be negative but to highlight that many believe Microsoft will be bundling several new features in its next release that will crush the VMware financials for anyone on academic pricing. It is believed MS will be folding App-V into the O/S allow all apps to run virtualized natively without the overhead of VMware's licensing or storage burden.

For now we just need to wait and see what finally happens. Meanwhile we also can celebrate the gains you have made for yoru students.

jdodge
John Dodge 1472 Points | Thu, 05/12/2011 - 19:19

Raechelle, Welcome to the Enterprise CIO Forum. You provide a lot great insight and advice such as one step back with the IT department managing more desktops and learning new tools -- to take two steps forward.

Could you quantify the number of virtual desktops, users,  the savings and tick off the benefits in order of their importance? Where are you in the evolution of this project? Is there an end point to this "adventure?"    

rclemmons
Raechelle Clemmons 8 Points | Sun, 05/15/2011 - 21:43

Thanks John, for the welcome.

Wow, a lot of questions...let's see if I can answer them all. We have about 365 desktops on campus today, and are growing about 5% per year. After our initial project we have about 50 virtual desktops, and are expecting to increase that to as many as 250 virtual desktops this summer. Next summer we'll try to virtualize the remaining desktops (our abilty to move to virtual is based on budget, and we're a July-June fiscal year). So as far as "endpoint", it would be Summer 2012.

We've done all sorts of cost analysis on our VDI project, looking at it from a 3-year perspective -- startup costs, and then 2011, 2012, and 2013 costs. Over 3 years we see that the capital costs of VDI are roughly the same as our capital costs pre-VDI (although in year 3, for example, the capital costs of VDI are greatly reduced over the same period in a physical desktop world). We're looking at abot $266 per desktop with VDI, and $242 per desktop without. These numbers are all based on our environment, actual refresh rates and costs, lour licensing costs, etc. So I don't want anyone to think it'll be exactly the same for them. [Perhaps there's another blog post in here, on how we calculated and compared costs!] OUr real savings, of course, is at the operating cost level, where we see our costs drop by nearly 50% over our 3 year projection. In total, with all costs factored in, we expect our virtual desktops to cost about $735/year compared to $1120/year for physical desktops, if we continued along the path we were on.

As far as benefits, there are a few, but for us the important ones were:

  • Improved access for students and faculty to computing labs and specialized software. With our implementation students can have 24x7 access to a virtual desktop, running the software they need to do their projects or homework.
  • Improved desktop support function. I have such a small team that reducing the amount of time it takes to manage and support our desktop environment is really, really important. Before VDI we were managing 16 "gold" images for different types of hardware and environment confirgurations (e.g., specialized lab, faculty desktop, classroom, etc.), so this is going to save us a lot of time and improve our desktop environment overall.
  • Reduced PC refresh rates and costs. This is clear from above.
  • Go green. Because we're implementing VDI we're able to refresh with thin and zero clients instead of traditional PCs, and save power in the process. Coservatively, our most power consuming thin clients consume 1/4 of the energy that our physical desktops do, and some zero clients we're testing consume less than 1/10th of the power. We figure that by year 3 of our project, based on how many PCs we project we'll refresh with thin/zero clients, we'll be saving over $4k per year in power alone!
jdodge
John Dodge 1472 Points | Mon, 05/16/2011 - 12:02

That is a great cost analysis, Raechelle. I am beginning to think higher education is showing us the way w desktop virt. You should say hello to Jerry Bishop on our site. He's very active in posting and a higher ed CIO like you. He's in the middle of a big cloud project.

http://www.enterprisecioforum.com/en/article/why-one-cio-nixed-new-data-...

Any chance you could post this as a blog? I will give great play....a lot of terrific info here.

rclemmons
Raechelle Clemmons 8 Points | Tue, 05/17/2011 - 03:26

Sure, I can post. Just the cost analysis or other pieces, too?

rclemmons
Raechelle Clemmons 8 Points | Sun, 05/15/2011 - 21:45

Hmmm...the HTML and the bullets didn't hold well in that reply. Makes it hard to read. Not sure how to edit it, either, now that I've posted it. Sorry about that!

blaberis
Bill Laberis 154 Points | Thu, 05/12/2011 - 15:49

Very good points both on the learning curve and that there really is no one-size-fits-all strategy with desktop virt. VDI is an excellent solution. It is not the only solution, by a long way.

rclemmons
Raechelle Clemmons 8 Points | Thu, 05/12/2011 - 17:19

I couldn't agree more that there is no "one size fits all" approach. When I left my last institution they were moving agressively into virtual computing labs (VCL), and couldn't believe I wasn't considering that for Menlo. It just didn't make sense for us, given our cost structure and size. It was only when we were able to address most or all of our use cases that it started being feasible. And then we had to decide if it was the best approach for us, because, as you said, it's one solution but not the only one. We're happy with our decision, but each institution needs to decide for themselves whether its right for them.