Who cares about the hypervisor? part 1

August 28, 2013 at 7:24 am in Cloud, Hypervisor by mikeresseler

I have been following the announcements at VMworld San Francisco this week to find out what is coming to VMware 5.5. Those who know me a bit may wonder why I did that as I am normally very Hyper-V focused. I have (in my opinion at least) a very good reason for that; The more we advance in the virtualization and cloud world, the more I’m getting convinced that the hypervisor has become a commodity. A couple of months ago I had a (virtual) bar-type conversation with Greg Shields about this and it is becoming more and more true every minute. Before I start to explain why I feel like that, allow me to take a minute why I don’t like hypervisor comparisons…

I don’t like hypervisor comparisons on blogs, whitepapers, independent researches (if you can call these independent…) and more because they are always biased. It’s that simple. If I would need to create a comparison post or whitepaper, I’m pretty sure that by the end of the day Microsoft Hyper-V would be the winner of the day because I am biased for Hyper-V. Is it because I believe it is the best hypervisor? No, it is because I know that hypervisor best. Those who write those posts try to compare different features with each other but in the end they have a preference and depending on what they see as necessary/optional/nice-to-have the answer to the best hypervisor question will be different. It gets even worse when we try to compare more hypervisors.

With that said, here are my observations from the last month. (Please note I’m looking at VMware and Hyper-V here… I can’t judge about the other hypervisors as I have not enough knowledge about them…). Both Hyper-V and VMware can be easily placed next to each other. They both have an impressive set of possibilities and if VMware says they have A then Hyper-V probably has it also but it is called B. In recent past times it always has been Hyper-V that needed to catch up with VMware but recently I just see 2 hypervisors with a lot of capabilities that are pretty much equal. The time that one fan-base can laugh at the other for missing X is practically over (it will always remain since they will never be fully equal but hey… follow me here for a second.)

A simple example: Microsoft Hyper-V and its VHDX format supports 64TB of disk space. Announced with Hyper-V 3 (or Windows Server 2012) which meant that the Microsoft guys could start shooting at the VMware guys because of this. At VMworld San Francisco they announced 64 TB support for VMDK also (technically correct it is 62 TB but hey…). But do we really care? When I need to chose for a hypervisor in my environment, is this one of my requirements? The possibility to have one (1!) virtual disk that is 60+ TB large. Or what about the fact that somebody supports x number of logical CPU’s and x amount of memory… (PS: for the Microsoft guys, I could have easily taken an example that turned around the example Winking smile)

Whenever I talk to somebody about what hypervisor he or she needs, I ask them the same simple question. What are your requirements? 9 out of 10 they have not a single clue and they tell something like performance and stability and then they are lost. So instead of focusing on what product has more features over the other ask yourself the following questions:

  • What features do I need in my environment?
  • What will the cost be of training the administrators in the technology
  • What kind of support do I want (and how much do I want to pay for it)
  • What’s the amount of VMs I want to run on a single host (density)
  • Do I have different locations?
  • (And many more questions that need to be asked upfront…)

Once you have the answers to those questions, you can start looking for the solution that BEST fits your needs. And not mine or any other out there that has a preference for one of the hypervisors.

All right, you made it this far… now it is time to come to my point. While I don’t care about the hypervisor anymore (as long as it fits your specific needs I’m happy) it now comes down to the next part in the chain… Automation, management, monitoring, backup, security… Setting up an environment is one thing. Keeping that environment running, healthy, protected is much more work and cost a lot more than the initial cost of the project.

I strongly believe that you need to ask yourself those questions during the initial phase of the project. What are we going to do when we are live? How will our patch management look? How am I going to see and foresee issues without staring at a monitor 24/7. What in case of disaster? How am I going to protect my environment against security risks?

In the follow-up post I’m going to discuss an integrated approach for not only your virtual environment but for your entire environment as a whole. And yes, I’m biased again for a certain suite of management tools but even if you don’t like those, you might want to read the follow-up post to catch my vision about management.