Virtualization | Feature

Calculating the TCO and ROI of Desktop Virtualization

Cost savings is one of the main arguments made for desktop virtualization. Here, the CIO of Lone Star College System shares how to calculate the TCO and ROI of your virtualization project.

The TCO and ROI of Desktop Virtualization
Photo: iStockphoto.com

This is the final installment in a three-part series on desktop virtualization. Part I looked at the potential benefits of desktop virtualization, while Part II focused on planning, design, and building a business case.

You cannot get to a realistic TCO and ROI for desktop virtualization without first identifying its strategic value to your institution, your goals, and the expected outcomes (Part I). After that, you must still design the architecture of the new system and decide who will be served by it (Part II). Only then can you move on to the task of calculating the TCO and, eventually, the ROI.

I recommend two ways to develop your TCO:

  1. Do the math: Identify what items need to be included in the TCO.
  2. Conduct a scenario: This is often the best way to show the TCO.

This story appears in the September 2013 digital edition of Campus Technology. Click here for a free subscription to the magazine.

Step 1: Do the Math
While all of us regularly calculate our maintenance and operations costs, very few of us drive to the cost per service. The reasons are simple: It's often unnecessary, and it's very hard to capture. Over the last three years, I have focused on determining the actual cost to deliver services at the Lone Star College System (TX), and it's not easy. To date, we have clearly calculated application-delivery costs for our ERP and desktops--both physical and virtual.

So how do you go about calculating the TCO for the desktops on your campus and the virtualized system you are considering? First, you need to figure out your current cost to deliver a computer--hardware and software--to a student or employee. We started with a simple checklist approach, and the number of items on the list grew fast. Next to each item on the checklist, we noted the associated cost per year. Small items often get overlooked, but they add up. Here, for example, is a short list of items related to deploying a physical computer:

  • Physical computer cost (including monitor) based on your current system lifecycle. Since Lone Star's lifecycle is four years, we have to include any replacement costs related to the desktop's projected lifecycle.
  • Operating system cost, even if you have a current enterprise agreement
  • All software licenses associated with the deployment of the system:
    • Antivirus software
    • Client-management tools
    • Actual applications deployed
  • Electricity consumed by the computer
  • Support costs per unit. This is a challenging calculation. You need to calculate all staff hours and costs associated with supporting a computer on campus. Since support costs increase as hardware ages, you should calculate these costs based on the number of years the computer has been in service. It is important to include all work orders. The need for software support often stems from old operating systems and patches. Consequently, you might want to break this calculation into two components:
    • Computers 1-2 years old have x number of service requests.
    • Computers 2-4 years old have x number of service requests.

Once you've done all this, you should have an excellent picture of your current TCO for the physical computers on your campus.

Your next job is to calculate the TCO of the planned virtualization project. To do this, you will draw on the vendor quotes you received based on your proposed design. Your biggest challenge is the fact that you must make some assumptions based on the experiences of others.

A typical short list of virtualization costs is similar to the one used to calculate the TCO of physical computers:

  • Cost of thin client, including monitor (if applicable)
  • Cost of required data center hardware
  • Cost of annual maintenance contract for data center hardware
  • Cost of virtual desktop software per seat
  • Operating system cost: You may have to change your current enterprise agreement or purchase licenses applicable to virtual desktops.
  • All software licenses associated with the deployment of the system:
    • Antivirus software. This may not be needed on each client, and will depend on the overall solution design.
    • Client-management tools. These also may not be needed on each client and depend on the overall solution design.
    • Actual applications deployed: If you are virtualizing your applications, you may save money on the licenses you need, but first read your existing licenses carefully: Virtualizing an application is prohibited by many licenses or requires a different license.
  • Power utilization of the system:
    • Client side
    • Server side: You have to add the cost of the power needed to run the servers.
  • Support costs per unit. These costs must be estimated based on others' experiences and projections, since a typical virtual desktop is based on a gold-state image (i.e., the operating system and software are regularly updated, tested, and start on demand in a virtual environment), and virtual applications' support costs do go down. If you are using thin clients, support costs will also decrease. In Lone Star's first deployment, for example, we estimated that the number of support calls would be a quarter of what they were when we utilized traditional computers. Four years later, I would estimate a 90 percent reduction in calls related to virtual desktops.

The next step is to calculate the infrastructure costs over time. In our planning, we assume that the virtual desktop infrastructure will last eight years, with minor upgrades needed around year four. I will note that we have had a virtual environment since 2009, and our initial Generation I servers are still in production. While we have upgraded many server blades to Generation VII blades, there are still valid production uses for the original hardware. It will take a long time to actually determine the full life of a virtual infrastructure model.

And don't forget to include the startup costs, too. These should only include one-time outlays associated with the project such as:

  • Data center upgrades: power and cooling
  • Network upgrades
  • Team-training costs
  • Consulting costs

These expenses should also be broken down into per-unit costs based on the expected life of the solution. In our case, we determined that the base solution (with a midpoint upgrade) had a full life of eight years. We made this determination based on the enterprise hardware we selected and how long we could keep hardware-maintenance contracts on the equipment.

At the end of all this, you should have a simple summary that outlines the following:

TCO Cost Comparison

Item

Per-Unit Annual Cost Lifecycle in Years Total Lifetime Cost
Physical desktop Actual cost 8 Actual cost x 8
Physical desktop refresh Estimated cost 8 Estimated cost x 8
Virtual desktop Estimated cost 8 Estimated cost x 8
Virtualization infrastructure Actual cost 8 Actual cost x 8
Virtualization startup costs Actual Cost 8 Actual cost x 8

The final calculation is to multiple the lifetime costs by the total number of desktops or virtual desktops that you plan to deploy. If your goal is 5,000 virtual desktops, then use 5,000 as the multiplier and compare against your actual costs for a physical desktop environment.

Step 2: Conduct a Scenario
The last task is to conduct a scenario to determine the value of the solution based on the project goal. A scenario brings soft costs (OPEX) and capital costs (CAPX) together, and is the best way to determine if virtual desktops are good for your organization.

Here's a typical scenario: University U has determined that the strategic value of virtual desktops is to improve services (support costs) and lower equipment-replacement costs. The goal is to provide Windows 7 across the campus on current hardware that is not capable of running Windows 7. The architecture will be a single, scalable converged infrastructure design. The virtual desktops will be based on a gold-state image that is nonpersistent.

The savings in capital expenses would be higher than might otherwise be expected, because University U plans to use existing computers for the endpoints. On the flip side, the school should expect to pay more in hardware-related support costs. Based on this model, the strategic value would be presented as follows:

  1. The move to virtual desktops will result in a savings of $___ over the next eight years.
  2. The startup costs, included in the total solution cost, would be $ ___.
  3. Support requirements will be reduced 85 percent from [insert current data] by improving system availability for students, faculty, and staff.
  4. We will be able to deploy Windows (7 or 8) on existing hardware, providing up-to-date applications for students, faculty, and staff.
  5. A future benefit of the system could be the support of a BYOD practice for students. This will be reviewed after the initial deployment.

The Return on Investment
At the end of this long process, you have developed a plan that identifies:

  1. The strategic value of desktop virtualization (For more information, read Part I.)
  2. The architecture of the system (For more information, read Part II.)
  3. The total cost of ownership over the life of the system.

The final step is determining the ROI. While often not necessary, calculating ROI is a best practice that will help in the adoption of any new technology. Unfortunately, calculating ROI is a challenge in higher education because, unlike the private sector, we don't have "loss of revenue" or the ability to reduce soft costs (personnel). It can be done, however, by comparing your physical deployment costs with the cost of the virtual desktop deployment, and using your scenario summary as a supporting element of the ROI.

In general, businesses expect a return on investment within three to five years, depending on the capital outlay. At some point, your school should experience savings through desktop virtualization, but don't expect the savings to be quite as high as in the private sector.

Furthermore, a virtual desktop approach might not be right for everyone. You will only know after completing a detailed review of the technology available, industry trends (BYOD), and strategic value: Is there a reason in my institution to provide this service?

comments powered by Disqus