Server virtualization has been in the last few years and still is a main IT trend on which firms are focusing their attention and their investments. There are many good explanations for this interest that we analyze in this brief article. We shall classify the benefits of server virtualization in two broad categories:
- financial benefits (e.g., reduction of server HW maintenance costs);
- intangible benefits (e.g., increased flexibility).
We shall see that while the latter are the most important long-term benefits, the former are those that typically play a main role in the investment decision process because they are easier to quantify and their value can therefore be better grasped.
As every other investments those in server virtualization have their own risks. Although we recognize their existence and their importance we will not analyze them in this article.
Most of server virtualization projects are also server consolidation project in the sense that one of the main project results is an impressive reduction of the number of servers. Clearly there are server virtualization projects (e.g., desktop virtualization) that do not aim at reducing the number of servers, but those that focus on server consolidation produce impressive reductions of the number of servers. With current technologies it is common to see 30:1 (or even higher) consolidation ratios. The net result is that large server farms consisting of hundreds of servers are replaced by very few servers hosting hundreds of Virtual Machines (VMs).
We made recently for one of our clients an in-depth analysis based on real costs of the benefits of a server virtualization project. The most striking result of the analysis is that the sheer savings due reduced server HW maintenance costs offset all the other project costs (including new servers purchase, project implementation) over a four year time horizon when tax shield is taken into account. Let further clarify this point. One of the most reliable criteria used to assess whether it makes sense to invest money in a project is the computation of the Net Present Value of the project. The Net Present Value is a simple formula that takes into account the simple fact that 1$ now values more than 1$ in one year by discounting the future costs and benefits by means of the so called opportunity cost of capital. In our project we computed the Net Present Value of the project over a four years time horizon, namely by taking into account costs and benefits for the subsequent four years. We also took into account the fact that future costs and benefits need to be discounted; and that if you invest your money in a project you will have tax savings (tax shield) that partially offset your costs. Taking all that into account we realized that the HW maintenance savings were enough to make the Net Present Value positive.
Clearly this is striking because all other financial benefits can be summed up the HW maintenance savings thereby further improving the Net Present Value of the project. This is exactly what happened when we added the other main categories of financial benefits:
- Floor space savings
- Power consumption savings
- SW licenses savings
These latter benefits are largely latent in the sense that they are accrued under specific circumstances. For instance in a server virtualization project that is also a server consolidation project you may end up with significant datacenter floor space utilization reduction but that does not necessarily mean you have any savings. If your datacenter occupation rate is close to the maximum this may become a real, very important benefit that translates in significant savings; if not, its usefulness is disputable.
By the same token, while you will likely have an impressive reduction of power consumption, the IT department may be uninterested to these savings for the very simple reason that these costs are often not charged to the IT department. Although that may seem strange to someone, this is what we see most of the time with our clients.
A server virtualization project entails new SW licenses costs, e.g., for the Virtual Machine Monitor SW, but under some circumstances it can also deliver a significant reduction of SW licenses costs. In a sheer server consolidation project with no consolidation of OS or application instances this reduction is clearly not due to the simplification of the SW stack but on the contrary to the specific features of the SW licensing rules. Licensing rules in virtualized settings are particularly complex and it may therefore happen that the net result is not a reduction of SW licenses costs but actually an increase of SW licenses costs. In our project there were on the contrary significant savings for a specific set of important SW applications. It is important to emphasize that these benefits were latent in the sense that given the SW license contract the client could not get his money back for the already paid SW licenses; nonetheless the client would be able to increase the number of deployed application instances at no additional cost.