I’ve been considering the idea of multiple hypervisors for some time now. The primary challenge that comes up in this discussion is the ability to manage the environments seamlessly. I even had an extended conversation with a vendor about this just a couple of days ago and general challenges of managing multiple hypervisors. The idea that you can manage your vSphere environment and KVM environment from a single pane of glass is great in concept. What’s interesting is that we don’t have the same expectation when we talk about managing heterogeneous OS environments. Or should I say, I haven’t seen a great operating model for seamlessly managing both Windows and Linux.
We love to compare things in IT. For instance, there was the whole Windows vs. automobile industry meme that was going around a few years ago. That proved to be more of apples to oranges comparison, but I think the OS example is a good comparison.
When collecting requirements for an application, we compile a list of functional requirements to determine the best OS platform. To this point we’ve built the infrastructure needed to manage heterogeneous OS environments. Normally, the touch points between managing the two environments are at a very high level such as hardware and application management. The underlying OS is normally supported by a separate set of management tools and even personnel. The commonalities are things such as configuration management, change management and incident management. For instance, in the case of hardware, we may have a single pane of glass to monitor hardware related events.
I’ve seen people attempt to implement solutions such as a single platform for patch management. I’ve also seen how difficult it is to manage both the product and operations model. The two different operating systems and the types of applications are so different that you end up with patch management administrators for each type of OS. You do get the advantage of unified reporting for compliance, but you lose the functionality of a purpose built solution. A better investment; a cross platform reporting tool.
Hypervisors should be treated the same as Linux and Windows management. There are requirements that drive selecting Windows over Linux and vice versa. For example, a test environment may be more suitable for KVM while production workloads run on VMware vSphere. Sure, you don’t get the industrial sized strength of vSphere management for the KVM environment. But, that’s also part of the determination for selecting the environment. In the OS world we don’t (solely) base the decision on which OS to use based on tools so, why would we use it as criteria for selecting a second hypervisor?
Where are you in discussions for using a second hypervisor in your environment? Is integrated management a real concern for your organization and why?