In Bernd Harzog's recent article, Microsoft Hyper-V Virtualization: Resistance is Futile (Almost), an anonymous user posted a very interesting comment detailing five reasons why Microsoft Hyper-V might not be so hard to resist compared to VMware's true enterprise virtualization solution. I liked the comment so much I figured I would share it with all.
The comment starts by saying Bernd made some good points but that he, the commenter, felt that the article neglected to point out some serious issues Microsoft has and/or complete misconceptions that is floating around the blogosphere along with the following five reasons Microsoft Hyper-V will not live up to the hype.
1. Hyper-V is marketed as free but really isn't.
Yeah it's $28 bucks but that's on top of a Windows 2008 License paid in full. This is nowhere near the realm of free, in fact if you take a straight Windows 2008 Standard License which is $999 and add Hyper-V you have a $1027 hypervisor that doesn't even have management ability.
VMware ESXi starts at $495 and comes bundled with most major hardware vendors soon (which can be perceived as free similar to hyper-v if apples are to apples). ESXi also has far more functionality and stability over Hyper-V.
Perceived pricing is all fine and good but when it's time to actually fork over the cash, the reality can (and will) be painful. There is no way to obtain Hyper-V without paying MS for full Licenses of Windows 2008. Even if you have a software deal, you still paid, and don't kid yourself that you didn't.
2. Hyper-V is not actually bare metal.
What we have here is a competitor to VMware server with Hardware assist and Enlightened devices. Hyper-v looks good on paper but is nothing more than a hosted hypervisor with paravirtualization support with dated features.
3. MS target market is low margin with little to gain because SMBs aren't ready.
Going for the entry level market is fine strategy that has worked in the past but Virtualization does not follow standard paths. There are reasons why SMBs don't use Virtualization as much, if they really intended to do server consolidation en masse they would have done so with the plethora of free software already out there.
The truth is that SMBs do not want to put their eggs in one basket. Placing 10-15 servers on one physical box is not ideal for them. In this situation a single outage is massive and may bring down the entire company. Hyper-V only increases this risk further. There are no killer features in Microsoft's roadmap right now that can address high availability at the costs necessary to achieve SMB adoption. Free is not going to work here otherwise VMware server would be absolutely everywhere already.
The SMB market is slowly coming around but just any old Hypervisor is not going to break them free. What is really needed is an affordable solution that can bring functionality like VMotion and High Availability at a price that is highly affordable.
4. Hyper-V ties the hypervisor back to the OS.
I don't see a server that is running Windows 2008 being vulnerable to viruses, patches and so on being a palatable platform to house 10 or more machines. Need a security update? Time to reboot...yeah that's not going over well. Virus took down a windows component? Now you have a massive outage.
The OS being part of the hypervisor is a big mistake. It has to be transparent, the user can't know it exists at all, VMware is right that it should be in the hardware completely protected from the elements. A hypervisor is supposed to be running virtual machines, not Virus Scanners and other 3rd party junk screwing up the OS.
MS will have to release a true bare metal product to compete with VMware and they need to do it soon. I highly doubt they will though as Windows is integral to Hyper-V's ability to do half of the things ESX does.
5. Virtual Machine Density is a fraction of what is possible
Memory Overcommit is underestimated by Microsoft. If you give 4GB RAM to a Virtual Machine, does it really actually always need that 4GB? No! If it uses 2GB 90% of the time, why not use that 2GB?
ESX has at least double the capacity for VMs over Hyper-V in the overwhelming majority of situations. And yes, it performs admirably in reality. Higher VM density = major cost savings.
Microsoft has a long way to go. They have a 1.0 product, good for them. But this product is going to have an uphill battle for a long time in a new market. Virtualization is a different animal than SQL server and Netscape. Price is not the only factor, especially when just implementing Virtual Machines gives immediate ROI regardless of the solution chosen.
I for one will choose 99.999% uptime and spend 2x the money to get it over a solution that can't offer anything close when I'm already going to save hundreds of thousands of dollars a year on top of licensing fees to implement it (or millions).
I must admit, I tend to agree with our anonymous poster but then I would never put a bet against Microsoft, never have and never will. So, it ought to be interesting... What do you think? Do you agree or disagree with our anonymous friend?