Who Wins, Now that the Hypervisor Wars Are Over?
Let's face it, the hypervisor wars are over. VMware clearly dominates the market and Microsoft is closing the gap with every release of Hyper-V, but the focus right now is no longer on the hypervisor.
We have reached a point where no single hypervisor will reign supreme in any organization. Furthermore, I believe that VMware will come under increased pressure from Citrix, Red Hat and other contenders. What we are beginning to witness in the enterprise is the compartmentalization or classification of hypervisors based on a workload perspective.
A year or two ago I probably would not have recommended Citrix XenDestop be run on anything but vSphere because of all the added benefits that VMware's hypervisor can offer, especially from a performance perspective. Today, I find that I can comfortably recommend the use of Hyper-V or even XenServer for XenDesktop workloads. Conversely, I can see situations where workloads of certain tier-1 applications can run on vSphere, whereas some other tier-1 applications can be run on Hyper-V, and so on.
The end of the importance of any particular hypervisor isn't a bad thing. You are probably thinking just about now that having all these hypervisors deployed will be a support nightmare, maybe even a logistical nightmare. It's inevitable, since the shift is already happening. What can alleviate the pain or "soften the blow" as it were are the management tools that will allow you to manage these different hypervisors from a single pane of glass. Microsoft System Center Virtual Machine Manager is a great example of a management console that carries with it a lot of promise.
The public cloud is another factor contributing to the new status quo. Enterprises will without a doubt use services from Microsoft's Azure, VMware's vCHS and others. So, it makes total sense for them to also leverage the hypervisor locally to facilitate interchangeability between the public cloud and the private virtual infrastructure.
This adoption of multiple hypervisors could have potentially been avoided had we been smart enough as a virtualization community to develop a standardized virtual machine format that works across hypervisors. But as long as each vendor maintains their own format, the enterprise will find itself using several hypervisors and several public clouds for different workloads to avoid vendor lock-in and to take advantage of the price wars that will take place.
So for those of you whose strategy is to standardize your company on a single hypervisor to streamline processes, support and training, I caution you to rethink this approach and take a closer look at the market circumstances. It might be cheaper for your organization to train in and support multiple hypervisors and multiple public clouds.
I am eager to hear from those of you that are running multiple hypervisors in production. What's the reasoning and justification that you used to win over your company. Please share your comments here or at firstname.lastname@example.org (my editor's e-mail).
Posted by Elias Khnaser on 01/22/2014 at 11:12 AM