Hypervisors and the features discussions, will this become obsolete?
With the upcoming release of Hyper-V 3.0 we see a lot of movement from both VMware and Microsoft on marketing level so all features are promoted, recently this type of marketing became a bit more ugly as VMware launched a ‘Get the facts‘ and older blogs from Microsoft on ‘Windows Server 8: Hyper-V 3.0 Evens the Odds with vSphere‘. Dan Brinkmann posted a very useful overview (from a Microsoft perspective) on his blog.
One of Microsoft’s arguments is that System Center can manage both VMware vSphere and Microsoft Hyper-V while VMware argues this:
- System Center Virtual Machine Manager lacks basic management functions for vSphere environments, like managing a vSphere host, cluster, or resource pools and provisioning storage and networking.
- System Center Virtual Machine Manager requires vCenter Server to manage vSphere environments, resulting in redundant cost and more complex processes
But is it still valid to argue on features? I guess these multiple hypervisor shops will grow, just a matter of choosing the right product for the solution. Being vendor agnostic gives the advantage to be able to choose the right hypervisor for the job but there’s management of different hypervisors, maybe different responsibilities, heck even different SLA’s on different platforms.
I had this discussion recently with a couple of specialists on vSphere (and all other VMware products) and the future of the hypervisor as they saw Hyper-V arising too, so we had the discussion on what would become important in a couple of years and we agreed that management should be the main focus.
So why would we focus on features as we can choose the best hypervisor for the job, we all know that vendors are able to handle certain workloads at a different level (I’m not claiming it’s ‘better’, just different)? I think the future of the hypervisor lays in management of these hypervisor layer, although SC maybe can’t manage vSphere like VMware can with vCenter or vCenter Operations Manager but there’s a start. For the larger deployment the management can be in vCD (vSphere only), CloudStack (no hyper-v support yet) or maybe even OpenStack.
As long as management platforms will support multiple hypervisors and translate a generic functionality to the hypervisor functionality we’re all good right? So point is that the hypervisor is/will become a commodity and key play in this setup is the management platform that can deliver the best feature of each hypervisor.