| Main |

December 19, 2010

VDI vs. SBC: ROI case study

The huge buzz around virtual desktop infrastructure (VDI) may ironically drive more server-based computing (SBC) sales. Some organizations are considering, as an alternative to VDI, expanding and repurposing existing XenApp deployments used primarily for application delivery to instead facilitate enterprise desktop replacement. One such firm recently requested an ROI comparison for migrating its 1,700 (and growing) physical desktops to either virtual desktops or to an enterprise Server-Based Computing environment. The required investment and the 5-year Internal Rate of Return (IRR) was nearly identical for the two scenarios, but the VDI option had a slightly higher ROI along with a shorter  period required to payback the initial investment.

SBC and VDI Similarities

Server Based Computing (SBC) is a mature technology that has been utilized for over 15 years, but primarily for application delivery; it never really caught on as a mainstream desktop replacement. Virtual Desktop Infrastructure (VDI), although only four years old, has quickly captured the interest of IT professionals already familiar with server virtualization. IT is generally more receptive to the simplicity of the virtual desktop concept than they’ve previously been to the idea of enterprise server-based computing.

VDI and SBC both enable the hosting of desktops on central server farms and use the same protocols to deliver application screen prints to users (VMware View also utilizes a protocol, PCoIP, developed especially for VDI). SBC typically incorporates Citrix XenApp along with Microsoft Remote Desktop Service (RDS – formerly called Terminal Server). VDI utilizes the virtual desktop running in a hypervisor, typically either VMware View or Citrix XenDesktop.

Both VDI and SBC desktops intelligently separate the personality of the user from the applications and abstract both from the OS. They enable a model of personal computing where the information belongs to the user; the device becomes a “choice” whether a PC, laptop, thin client, zero client, Mac, iPAD, iPhone, Android device or Internet café terminal. Users can securely access their desktops from anywhere that they can get to a browser. The desktop remains exclusively in the data center where it is secure, managed, backed up and replicated for redundancy.

SBC enables the use of thin clients, while VDI enables the use of either thin clients or zero clients (devices manufactured for VDI without any local OS). Both thin clients and zero clients are fairly inexpensive and have no moving parts, local drives or fan noise. The devices are configured simply by plugging them in, enabling quick and simple replacement in the unlikely event one fails. They eliminate the requirement for upgrading PCs or laptops on a regular basis as well as the necessity for users to double as desktop administrators. User productivity is enhanced while IT support time is slashed.

Both SBC and VDI can reduce anti-virus costs, slash power consumption (by using terminals) and reduce downtime while users await new upgrades or help desk support. They empower users – giving them a new level of flexibility and agility. They facilitate acquisitions by enabling quick assimilation of new organizations into the existing desktop infrastructure while still allowing them to run their existing environments in parallel. They enable fast set-up of remote facilities and reduce time to market by accelerating application provisioning.

Server Based Computing

The SBC concept originated with Ed Iacobucci who worked for IBM on the OS/2 development team in the late 1980s. Iacobucci came up with an idea for a multi-user version of OS/2 that he unsuccessfully pitched to IBM. Microsoft liked the idea, however, and helped fund his new company, Citrix. Citrix’s first Windows based multi-user product, WinFrame, debuted in 1995. Citrix originally called the concept of hosting centralized desktops sessions “thin-client server computing”, but renamed the category “server-based computing” as Windows terminal manufacturers appropriated the thin-client moniker.

SBC relies upon RDS which is a version of Windows Server that supports multiple users. While RDS continues to improve, issues can still arise regarding application compatibility. Making applications work correctly may require specialized IT administrative knowledge of RDS along with registry hacks, trouble shooting or application streaming.

Because all users on a server share a single Windows Server session, one user can potentially affect all of the users on a host server. As an example, a user could receive an email attachment that installs malicious code on the server itself. Application packaging may also be required in order to resolve DLL conflicts resulting from incompatible applications installed and working on the same server.

SBC requires connectivity to the central server farm in order for users to work. User personalization needs to be added through the use of products such as Citrix Profile Manager, AppSense or many others. Users may object to working on an unfamiliar Windows Server desktop rather than the Windows XP or Windows 7 format with which they’re familiar.  It requires an approximate one-time $110 Windows RDS license.

The maturity of SBC may also indicate a dwindling support structure as VDI continues to capture the major mindshare not only of new ecosystem partners such as Unidesk and Pano Logic, but of existing legacy SBC players. Neil Spellings, for instance, recently raised some questions about potential challenges Citrix may have in supporting the legacy XenApp product.

On the positive side, SBC is a very mature platform – particularly when utilizing Citrix XenApp, which includes advanced tool sets such as session shadowing along with years of proven ecopartner complementary products. SBC tends to scale much better that VDI in terms of compute resources and also requires less storage IOPs. Tasks such as patches and upgrades are easily applied at the server level which then instantly propagate to all users – administrators do not need to manage a lot of desktops.

Virtual Desktop Infrastructure

VMware coined the term "VDI" in 2006 in response to the increasing tendency of its customers to run desktop operating systems within virtual machines on their VMware ESX hosts. Gartner refers to the concept as Hosted Virtual Desktops (VDH) while IDC calls it Centralized Virtual Desktops (CVD). Cisco’s recent entry (building upon either VMware View or Citrix XenDesktop) is termed VXI for Virtual Experience Infrastructure. The VDI industry has huge momentum and now includes not only VMware and Citrix, but also Microsoft, Red Hat, Quest and many smaller niche players along with a rapidly growing ecosystem. 

VDI refers to hosting desktops on hypervisors. Persistent desktops are run as a complete image in the data center on a 1:1 ratio of desktop to user. While this approach provides the exact look-and-feel with which users are comfortable, it still requires managing the individual virtual machines as well as multiple identical copies of operating systems and user data. A non-persistent desktop creates a user’s desktop each time she logs in utilizing techniques such as parent/child pointers. While this approach enables consolidated management and storage, it loses the ability for users to customize their desktops. Ecosystem partners such as Unidesk and AppSense as well as market leaders VMware and Citrix themselves increasingly offer different approaches to enable the combined advantages of both approaches.

The ability for users to work on a familiar Windows XP or Windows 7 desktop is one of the most compelling aspects of VDI along with no worry about application incompatibilities. Other advantages include better security, fault tolerance, superior load-balancing, easier backups, off-line sessions and a potential pristine desktop every time a user logs in. VDI administrators do not need to be specialists in RDS or registry hacks, and can utilize standard Windows printing.

On the negative side, VDI requires a $100 yearly Microsoft Virtual Desktop Access license for thin-clients, zero clients, Macs and other non-Windows devices or for Windows clients not covered by Microsoft SA. It scales roughly about half as well as SBC on the compute side, although ever faster processors and optimized compute platforms such as the Cisco UCS increasingly will mitigate this negative. VDI also can potentially require significantly more expensive storage in order to supply the required IOPs.

VDI vs. SBC Matrix

 Virtual-Desktop-Infrastructure

ROI Comparison

The organization mentioned earlier is a financial services firm that will migrate 1,700 physical desktop users (growing to 2,200) to either a Citrix XenApp SBC or a Citrix XenDesktop or VMware View VDI deployment. The organization is running a small XenApp implementation today – primarily for application delivery, but the staff has significant XenApp SBC experience gained at other firms. The organization currently purchases quality PCs and laptops along with Microsoft SA, and refreshes them every three years. It maintains a server in each of its 20 remote offices that can be eliminated under either a VDI or SBC scenario without requiring increased costs for bandwidth. It runs VMware vSphere in the data center, but requires new shared storage for either solution.

 

6a01156f01861f970c0148c6e30d0d970c-800wi 

The cash flow comparisons for years 1 – 5 reflect the difference in total expenses (CapEx + OpEx) between the existing physical desktop scenario and the proposed VDI or SBC scenarios, discounted for the firm’s cost of capital. The SBC solution includes Citrix XenApp Platinum Edition while the VDI solution utilizes VMware View Premier along with the required vSphere Enterprise Plus licenses. Both solutions assume identically priced thin-client devices that gradually replace PCs/laptops as they come up for refresh. VDI assumes increased storage costs of $410K while SBC incurs a little less than half that amount. Neither solution incorporates user productivity savings.

The organization would have to purchase either RDS licenses up front if going the SBC route, or SA licenses annually as it replaces PCs with thin clients if going VDI. Two additional full-time administrators were assumed to be required under SBC rather than just one under VDI. If only one additional administrator were required under SBC, the payback period would drop to 15.9 months while the 5-year ROI would increase to 228% and the IRR to 63%.

Choosing the Optimal Platform Requires Context

As with every key virtualization platform decision, I recommend taking a strategic approach. Start with an ROI analysis to first evaluate the economics of moving to a hosted desktop architecture whether VDI or SBC, but supplement it with an evaluation showing how the new technology will impact corporate objectives such as responsiveness, employee empowerment, disaster recovery, green initiatives, etc.

While industry analysts often contend that VDI does not offer a positive ROI, I have found that generally it does. Additionally, there may be benefits in terms of improved security, time to market and user productivity that warrant a move to hosted desktops even without a compelling ROI. A law firm with which I worked, for example, estimated increased billings of $1.6 M per year by enabling partners to bill more hours through remote desktop capabilities – an amount that easily overshadowed the projected savings.

Assuming the ROI and other benefits warrant a move away from traditional physical desktops, the best architecture can then be determined based upon environmental conditions and staff experience/expertise. An organization, for example, with a significant deployment of XenApp, even if used primarily for application delivery, might find it an easy transition to implement XenApp as an enterprise desktop replacement solution without requiring a significant investment in IT staff training or back-end infrastructure. A highly virtualized organization, on the other hand, may find that it makes more sense to extend its virtualized data center down to the desktop leveraging the existing infrastructure, licensing, management console and staff expertise.

Cisco VXI

Cisco’s recently announced Virtual Experience Infrastructure (VXI) has the potential for significantly weighting the VDI/SBC debate toward VDI by combining virtual desktops with unified communications. VXI takes a network-centric approach to addressing the gaps in enterprise scalability, performance and security that have contributed to a lack of wider VDI adoption. By both enriching the user experience and improving the economics of virtual desktops, VXI will make it easier for many organizations to commit to a virtualized desktop platform. The ROI analysis featured in this article was started before the advent of Cisco VXI; the Cisco convergence and other advantages were not considered.

The Decision

The financial services firm’s IT staff has yet to decide which hosted desktop solution they will implement. Given the similar ROI results of the two options along with the very positive experience the CIO had with an enterprise SBC deployment of Citrix MetaFrame at his last company, my guess is that SBC is going to win out.

 

Contributors:  Mark Vaughn, INX and Alan Kaplan, NetBlaze. Although Douglas Brown www.dabcc.com and I don’t always agree, he provided fantastic feedback (most of which I incorporated) and corrections.

 

References:

Server Based Computing vs. Desktop Virtualization Jeff Muir – Citrix Blogger

VDI vs SBC, like Gas Stove vs Microwave? James Rabev – The Citrix Blog

VDI is Dead, Long Live VDI-SBC! Christophe Corne – Business Computing World

VDI vs TX – What will you choose? Daniel Feller – The Citrix Blog

Terminal Services versus VDI: Brian’s presentation from VMworld Europe 2009. Brianmadden.com

Cisco targets virtual desktops with VDI. By The Bell

Cisco Unveils Virtualization Experience Infrastructure…  Cisco Press Release

The desktops may be virtual, but the ROI is real. By The Bell

 

Author Disclosure: I work for a leading Cisco/VMware partner.

TrackBack

TrackBack URL for this entry:
http://bythebell.com/2010/12/vdi-vs-sbc-a-roi-case-study.html/trackback

Listed below are links to weblogs that reference VDI vs. SBC: ROI case study:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

6 Responses to VDI vs. SBC: ROI case study

  1. Very interesting analysis. It would be interesting to understand the cost of the virtualization software in the case of VDI. How much was allocated for vSphere (and what was the pricing model) and for each VMware view seat?

  2. David,
    Thank you for your comment.
    I used $225 per user for VMware View Premier (+ SNS) which includes the back-end vSphere component. For Citrix XenApp Platinum, I used $300 + SA. I did not allocate additional monies for vSphere in the SBA example.

  3. At first when i saw the title I thought great, someone doing a comparison. After reading your background I thought you might also be capable of a real life comparison. Unfortunately after reading through your article it’s clear you’re really here only pushing VMware View and since it’s a one trick pony of course virtualized desktops come out looking pretty good.
    VDI is great but SBC still has a huge place in “desktop virtualization”. Consider the call center with 4 apps…you’re going to use a full desktop for all of them…I think not.. I can give you a few ROI’s on that.
    XenDesktop is the licensing structure you would use for either VDI or SBC…which is $225 list! You really killed your credibility there…come on…at least know what they sell.
    Also, you just put some numbers on paper, no specs, no detail…seriously?
    As to your comment about how many administrators it takes to maintain the environment… what is that based on? Do you have any references to base that on?
    I’ve read your blog for a while, this article was extremely disappointing. You should just say this is a VMware View ROI article…don’t mislead people into thinking you know what a Citrix solution costs…and Citrix does VDI too…christ! Citrix is about choice, do what is right for the user group you are deploying Desktop Virtualization to.
    To remain relevant with our peers and our customers it’s important we remain as impartial as possible when it comes to the solution…the second we sell a solution because it’s the only one we sell we become irrelevant…maybe a good salesperson, but not a good business person…let alone a technologist.
    A lot has changed with the Citrix solution set, might be time to brush up before you write another article comparing it.

  4. Dan,
    Thank you for your comment.
    Regarding XD vs. View, you will note that the blog states it was a choice of XA vs. either View or XD. I used View for the ROI analysis as it is a less expensive solution for this customer, but if VDI ends up winning, I’m sure it will then come down to another comparison between XD and View.
    You are correct in that I am not up to date in the latest XA pricing or features which is why I solitited external feedback. I relied primarily, however, upon the information from the customer who is already running XA and which has a great relationship with the local Citrix rep. The price they were told they would pay for XA Premium was actually quite a bit higher than the one I used for the ROI. And the staffing requirements were something that the customer discussed with our team, but given the customer’s background with both Citrix MF and XA, I felt comfortable with the determined levels. All of that being said, I do want to make the analysis as accurate as possible. I am going to go back and contact the Citrix rep directly, relay your comments and see if a revision is warranted. If so, I will update the blog post accordingly.
    As far as relevance goes as a technologist, I disagree with you there. I think that solutions providers sometimes try to appear as unbiased experts in several solutions, but that is not my approach. The CIO in my post, for example, contacted me because of an ROI analysis I did for his firm many years ago that resulted in a very large (and very successful) enterprise SBC deployment with Citrix MetaFrame. When he contacted me for an ROI for his current firm, I disclosed immediately that I was both VDI/VMware biased and far more familiar with View virtual desktops than other alternatives. He was more interested, though, in building a compelling business case for moving forward with the concept of hosted desktops; the particular architecture and products was secondary. And if he does happen to go with VDI, given the requirements of some of his users, I will recommend purchase of a profile management package unless View has incorporated such by that time. This will make the cost comarison between XD and View more accurate.

  5. Update comment to Dan:
    Dan,
    I spoke with the Citrix rep who confirmed that you are correct – customers can purchase XenDesktop for $350 vs. XenDesktop for $600, and still get XenApp! The difference, though, is that XD is licensed by named user while XA is licensed by concurrent users. In this case, the customer is already aware of the option. The rep concurs with my feeling that the customer will choose straight XA.

  6. Sometimes in IT people get hung up on what’s hot currently, and focus on “one or the other” strategies. The fact is that Terminal Server has advantages over VDI, while VDI also has advantages over Terminal Server. That’s why most organizations are best served by adopting a hybrid approach, with an optimal mix of Terminal Server (for task-oriented users), VDI (for power users), and Blade PCs (stock traders, graphic designers, etc.) which delivers the most benefit and platform flexibility to the organization.
    Ericom Software’s PowerTerm WebConnect facilitates this hybrid approach by managing access to Terminal Server, VDI and Blade PCs, all with one management console, one product.
    For more info on the hybrid approach visit:
    http://www.ericom.com/vdi_wts.asp?URL_ID=708
    Adam

Post a comment

Leave a Reply

Your email address will not be published. Required fields are marked *


*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>