Many debate in the field, hypervisors are commodity items in the datacenters today. So much so that even Gartner have decided to retire the Magic Quadrant for it, saying its mature technology. (Funny, how array based storage isn't mature tech but yet still a quadrant for it, but that's a discussion for another day)
How so commodity?
Well, to many, if the hypervisor had the ability to create some VM's, report some performance metrics, online migration and a decent GUI for management, it was good enough. Most importantly, if it was 'free' it would have been the icing on the cake. If that is the case, how is it that VMware & Microsoft are still leaving so many competitors way behind (as of the last Magic Quadrant)? Could there be a reason for it?
This is not necessarily a post to compare every nut and bolt of the hypervisor. Just wanted to relate what are some of the key items one should look out for when it come to picking hypervisors.
What have you planned for your data center today?
Are you considering modernising your storage systems or networks? Perhaps HCI or network virtualisation is something you are looking at long term? Ability to deliver compute, network and storage capabilities natively, and strong interoperability with partner ecosystems, are vital ingredients for success. I think we all agree that anything purpose built and native, will always outperform a solution that is made to integrate with each other.
This is all big boy enterprise talk, how about SME's? Modernisation benefits everyone! For example, vSphere + vSAN can be had in a 2-node configuration licensed by small VM counts, built specifically for our smaller customers. This allows our customers to not compromise on capabilities but still allow them to start small.
Is the public cloud part of your strategy?
Does your strategy involve bridging workloads into the public cloud? Agility to move workloads back and forth between multiple cloud providers? Most customers are beginning to realise that the long term strategy is hybrid cloud, and it is often not just a single cloud provider. For a long time, migrating workloads either to or fro, is quite the hassle.
Many customers can easily migrate workloads from Production data centres to DR data centres via vMotion live online. Why is this so hard to do with public clouds? Wouldn't it be nice to view the Public Cloud Data Centre as just another data centre on your management pane, and just migrate workloads to it easily? Many claim they can "connect" to the cloud, but we already do it today, not just "connect", but fully operational and running in the public cloud.
Again, SME's can benefit from it. Assuming you have a Production Data Centre with no DR site today. Now the Public Cloud can be your DR facility. Feels like a real DC, works like a real DC, managed like your own DC, but consumed on a pay-per-use basis.
How easily can you acquire skill-set or support to drive this?
Skill set is not limited to those in your internal admin teams today, but also the ability to hire similarly skilled techs. Availability of vendor support is also key in this discussion. I'm speaking more from a local Southeast Asian context, because unlike other mature regions, skill-sets are scarce and principal vendor support, may not be as per their US or EMEA counterparts.
For example, it is probably a lot easier to find a candidate with VMware or Microsoft skill-sets today, than it is for someone with a Xen or KVM skill-set. And we are not talking about just a basic click to create a VM type skill-set. Often time, when vendors taut their support scores, it is measured globally, but are resources distributed evenly globally? So hence, would customers be okay (specifically in my region), when a vendor says, we do not have a presence in country X, we can support you from Singapore, if need be, we will travel to you. To some it may be good enough, but if you had a choice, you would probably want in-country support as much as possible.
Is it future proof?
We are starting to see more and more new workloads nowadays Customers are asking, if we are able to support Cloud-Native-Apps and Next-Gen applications too. Some of the common ones being, Kubernetes, Dockers, Cassandra and the likes. This may or may not apply to many, and very much still in its infancy (in my region) but it is a consideration that should feature if there are already discussions as to how this will tie in with your overall strategy.
But it is FREE!
Or is it? KVM or Xen is one of the few open-source projects, hence it is available for free, but many vendors have built proprietary features on-top of it, and charge some level of support for it. Some bundle it together with their hardware and calls it free, but if its truly free, I guess you could also ask for "just the software" since its free? Could you? Perhaps you could get some form of community edition and have zero support for it.
And by the way, do due diligence on what features are available on the FREE Edition vs the PAID Edition. I have seen my fair share of surprises.
I would like to believe that there are NO FREE lunches, and there are always caveats. Nobody will be spending time building something for FREE with no way to monetise it in some form or another. To side track, just so you know, VMware ESXi Hypervisors can be had for FREE too, with only community support. :)
So where are we going with this?
It all begins by having a strong, robust and trusted platform to work off. Interestingly (you guessed it), it is the HYPERVISOR that forms this base.