In this post about VMware on the Benefits of Virtualization for the SMB, we concluded with a bit of analysis that suggested that if lack of budget was the major constraint that was preventing an SMB from pursuing a virtualization project, that this would tip the scales in favor of Microsoft Hyper-V due to its lower price of acquisition, and the fact that it would cost less in terms of training time for the IT staff for a typically all Windows shop to adopt Hyper-V than learn something this is at its core completely new (VMware) with a GNU and Posix management appliance or based on Linux (Xen).
Of course VMware can (and will counter) with the argument that a very feature rich, high performance, and scalable hypervisor is available from VMware in the form the free ESXi hypervisor. So what we have in the battle for the virtualization business in the SMB is a battle between two products (Hyper-V and ESXi) that are either “free” or “almost free”. However, the notion of free software falls afoul of some very important concepts rooted in the field of economics, that should be strongly considered when making this decision.
Opportunity Cost says that the cost of doing something (say activity A) is not just the money and time that you spend doing activity A. It is also the foregone benefit of what you could have realized had you spent that time and money on say, activity B. This is an extremely relevant concept to the decision for how an SMB, an SME or even an enterprise would choose between Microsoft and VMware. Let’s start with the SMB case.
Most SMB’s (if we use the definition in VMware’s study of an SMB having between 20 and 1000 employees), have an extremely limited IT staff. On the lower end of this scale there might be only one full time employee or just a consultant who works for a VAR and who takes care of the IT resources for multiple companies. If there are only a few people in the entire IT department, it is likely that the company is a Windows only shop since Microsoft has done a great job of pushing down the cost of administering Windows based servers over the years, and also done a great job of building up the supply (through training and certification programs) of Windows IT Administration resources. In this case, there are several factors that work in Microsoft’s favor:
- If the IT team already knows Windows Server, it is a small hill to climb to embrace Hyper-V.
- Embracing Hyper-V would not require learning an entirely new layer of data center infrastructure as would even the free version of VMware’s solution ESXi.
- For an all Windows shop, with obviously all Windows IT admin’s Hyper-V would not mean bringing in a set of people who are likely to be strong VMware advocates and who in many cases would bring with them a pro-VMware and anti-Microsoft culture and attitude. Keeping things simple would promote faster and easier decision making, and prevent the outbreak of the inevitable VMware vs. Microsoft debates within the company.
However for an all-Microsoft SMB or SME, going down the path of Hyper-V does entail some significant opportunity costs. Those costs are the benefits of the VMware solution that come in the form of features in the VMware solution that Microsoft does not offer and that are then foregone as a consequence of the Hyper-V decision. This least of features and their associated benefits includes:
- Transparent Page Sharing, and Memory Overcommitt. Together, these features allow for greater guest density on a VMware system than on a Hyper-V system. This directly affects the cost of the virtualization project as it directly affects the number of host that need to be purchased for the project.
- Mature interfaces for value add by third party vendors. VMware offers the VMSafe interface which allows security vendors to plug into VMware, the vNetwork interfaces that allow for monitoring appliances and value added networking products like the Cisco Nexus 1000V, and the vStorage interfaces that allow storage vendors to surface their value added features (like de-duplication) to the VMware system.
- Improved availability features (like SRM and Fault Tolerance). These are only available in the more expensive VMware version, vSphere but this is relevant nonetheless as there is at least an upgrade path to this version for the SMB that grows to require these features.
To wrap up the opportunity cost part of this analysis, it will certainly be easier and cheaper for an all-Windows SMB or SME to go the Hyper-V route that the VMware route due mostly to IT admin skill set considerations. However the decision to go down this easier and cheaper path should be made while considering the opportunity costs of the forgone benefits that a VMware solution would provide.
Before making a decision to go down a path with a virtualization technology an SMB, SME or an enterprise should consider the costs of reversing that decision and going with a competing solution. This means asking the question, “What if, for whatever reason, we needed to undo this and change directions”? Switching costs are something that software vendors are keenly aware of and that drive certain behaviors on the part of software vendors that may seem illogical at first blush.
Consider the following question. Why is it in the interests of VMware (and its shareholders) to give away VMware ESXi? ESXi contains a great deal of VMware’s market leading intellectual property. Why would a set of very experienced and very smart software executives (who used to run Microsoft Windows) at VMware ‘give away” something that cost so much to develop and is so valuable. The answer is that by getting their technology and the skill set embedded into a company, even for free, they create a set of costs for that company to switch to something else. The bet on the part of VMware is very simple. The bet is that if a company adopts ESXi, and if and when that company outgrows ESXi and needs a product with more capability, that it will be more costly to switch to a competing virtualization platform than to upgrade to a higher end of something that is similar to what the company has already been using. The cost to move to a competing or alternative platform is called the switching cost associated with that action and is economic manifestation of the degree to which you are locked into something once you have deployed it.
As a side note, it is switching costs that make the comparison of the current battle between Microsoft and VMware for the virtualization market with the battle between Microsoft and Netscape invalid. While many people adopted the Netscape browser in the early days of the Internet, Netscape failed to build features into its product that creating switching costs for its customers. Therefore once a competitive and free (packaged with Windows) browser emerged from Microsoft, people switched because it was easy to do, and because there were no painful consequences for doing so. The same remains largely true of the browser market today. Other than the ability to run certain Window centric applications in IE (something based upon an Active-X control by way of example), if you switch browsers everything should just work. This means that the barriers to entry in the browser market are low, as the switching costs from the incumbent leaders are low as well (something that Google intends to exploit with Chrome).
The switching costs for virtualization platforms are much more profound. The very first decision that you are making when you choose a virtualization platform (whether you know you are making this decision or not), is from which vendor are you going get the layer of software that interfaces to the hardware in your data center. If you choose Hyper-V then that layer will continue to be the Windows device drivers that you probably have been relying upon for years. If you choose VMware, then that layer will be the device drivers that VMware vSphere ships with either ESXi or ESX. Once you implement either and standardize on this layer of software in your data center you have incurred substantial switching costs. This is the reason why ESXi is free. VMware knows that if it can establish a toehold with this layer of software in your data center, that your costs of upgrading to other versions of their software will be lower than your costs of switching virtualization platforms.
Just as Milton Friedman (the Nobel prize winning economist) once said “There is no such thing as a free lunch”, there is also no such thing as free software. The minimum cost of a supposedly free piece of software is the opportunity cost of your time spent using it, and the forgone value of that time spent doing something else. To go back to the browser example, the cost downloading an using the free Google Chrome browser is really the forgone value you could have gotten by doing something else with that time. This holds very true for virtualization software (since there is much more time involved in this decision). Finally the choice of a virtualization platform also creates significant switching costs, which makes the decision a much more strategic one than the choice of a browser. SMB’s, SME’s and of course enterprises should carefully consider all of financial and non-financial (indirect costs like opportunity costs and switching costs) when making an important decision like which virtualization platform to standardize upon.