When Microsoft Hyper-V Server 2008 shipped last October, it was clear that whilst it was a very performant and powerful platform, it was missing some of the key features and capabilities that you got with Hyper-V as part of Windows Server 2008. For me, the killer capability that was missing from Hyper-V Server 2008, was the ability to cluster the physical boxes, and thus create a more resilient virtualised infrastructure. This wasn’t an issue for Hyper-V as part of Enterprise and Datacenter Windows Server 2008. The reason this capability wasn’t in Hyper-V Server, was the fact that the Hyper-V bit of Hyper-V Server, was based on the Hyper-V bit of Windows Server 2008 Standard Edition, which, as most people know, also doesn’t have clustering capability. This wasn’t the only capability that was missing, due to the use of Standard Edition. Scalability was also affected, as Standard Edition of WS2008 tops out at 32GB RAM when deployed as x64, thus Hyper-V Server 2008 also had this limit imposed.
So, with these missing capabilities in mind, what was Hyper-V Server aiming to provide?
Well, as I blogged way back in October, I highlighted 4 key areas for Hyper-V Server 2008:
Why Test & Dev? Well, think about it – if you’re running VMs on a production, clustered, set of WS2008 Hyper-V boxes, would you want to be testing and developing on the same machines? Would you want to be creating VMs, installing OS’s etc, on those servers? That would consume resource, so why not use Hyper-V Server 2008 to stage those VMs, get them spun up and tested, and then, once the process has completed, export them over to your production environment, without needing to modify the virtual machine. Makes sense to me, and after all, Hyper-V Server isn’t exactly going to cost you anything.
Server Consolidation – yep, still agree with that one, especially if people are thinking about virtualising things like Small Business Server 2008 Premium. Using HVS 2008 to hold both SBS OS’s (the core, and the extra server) enables SBS customers to save some hardware costs and increase their flexibility as their backend infrastructure becomes a few VHD files.
Branch Office – I’m in 2 minds about this one – part of me thinks it’s great to virtualise those 4 servers down to 1 physical and 4 VMs, because I’ll save money through power etc, and I’m still providing the same level of resiliency for those 4 workloads as they had in the physical world, but then the other part of me thinks, it would be safer to have a clustered solution, but that would incur licensing costs in the form of Windows Server 2008 Enterprise or Datacenter licensing. I guess it would be up to the customer to determine the level of resiliency required.
VDI – I still believe that HVS 2008 is a good solution for VDI, especially for pooled vDesktops, but, I think what hits HVS 2008 hardest here is the 32GB RAM limit. This may not be enough for all folk, but for the smaller environment, it provides a simple, and performant solution.
So, that’s Hyper-V Server 2008 today – all I’ve tried to do there is help set the scene, so you can understand better, where Hyper-V Server 2008 R2 fits when it launches later in the year.
What’s changed in R2?
Not the price for a start. Still free, but then, I think it needs to be.
If I had to summarise the changes and improvements in HVS 2008 R2, I’d pinch Zane Adam’s statement, from the Microsoft Virtualisation Blog:
“Our core strategy is to ensure that our customers can virtualize their IT environment in the most cost effective manner, and at the same time, have access to enterprise features like live migration and clustering features for high availability. So in addition to scalability and performance improvements in this version, customers can get live migration and host clustering capabilities and high availability (up to 16 nodes) at no charge.”
First thing you’ll notice from that statement, is that Live Migration and Clustering (High Availability) are in there. In the industry, especially with Live Migration, these have been deemed ‘Enterprise Features’, and because Hyper-V in V1 didn’t have Live Migration, it couldn’t be classed as an Enterprise ready technology – something I’d disagree with wholeheartedly. That said – would I have liked Live Migration in V1 of Hyper-V? Yes! Do I think it would have accelerated adoption? Yes! Something we’ll never know I’m afraid! Let’s, for argument’s sake, say that Live Migration is still an ‘Enterprise Feature’. What Microsoft, with Hyper-V Server 2008 R2, are doing, is making this ‘Enterprise Feature’ available to a much greater market than ever before. Citrix, with their XenServer 5 platform, are doing the same, and have an equally strong story on that front, introducing an excellent array of features into their platform.
I’m not going to say that Hyper-V Server 2008 R2 or XenServer 5, have all the bells and whistles that something like vSphere from VMware has, but I think the Virtualisation space is really going to shake up over the next 12 months, as there are a hell of a lot of people who haven’t virtualised yet, and a fair few people waiting for R2 to come around.
Live Migration and Clustering are great additions, but what else is new?
The Cluster Shared Volumes capability is present, which means no more 1VM per LUN (unless you want to!) recommendation, but whilst solving one problem is great, CSV also provides component level fault tolerance, which means that SAN/Network connectivity for a particular cluster node can interrupted, yet the VMs will carry on running with no interruption. You can read a bit more about CSV here.
You can also hot add/remove VHD’s or Pass Through Disks to VMs, with no interruption, and we’ve increased the number of logical processors (cores) that are supported in the physical hardware, up to 32. Final improvement that I’m going to mention – 1TB RAM memory support – very scalable indeed.
Where does it fit?
It may be easier to start with where doesn’t it fit :-) It’s scalable, it’s got Enterprise Features (:-)), it’s got a broad hardware support ecosystem, a growing ISV support ecosystem, it can be deployed and managed with current tools, it integrates with your current infrastructure, it’s small, fast, secure, and suits Enterprises in the same way it suits SMBs. It’s great for VDI, Testing, Development, Hosting, Branch Offices (thanks to Clustering), and above all, it’s a fantastic choice to virtualise your Server workloads too. It’s WIM based, so you can easily deploy it from a Windows Deployment Server, Ghost it, ImageX it, use System Center Configuration Manager 2007 to deploy it – you’ve got loads of choice. One of the nice new features of vSphere (Enterprise Plus) is Host Profiles, to ensure ESX hosts are configured in the same way through deployment, or after deployment. Isn’t that what SCCM 2007 SP1 does with Task Sequences? How much can be achieved with unattend files? Time will tell but I’d argue they aren’t too dissimilar.
You may be thinking, if HVS 2008 R2 is as feature rich as WS2008 R2 with Hyper-V (Enterprise and Datacenter), why would I buy WS 2008R2 with Hyper-V? Licenses. You don’t get any free licenses with HVS, whereas you get 4 free running instances of Windows Server with each copy of Enterprise, and an unlimited number of free running instances with Datacenter. So, if you need Windows licenses for your virtual machines, you may choose to go down the WS2008 R2 route. That said, there would be nothing stopping you buying WS2008 R2 licenses, and running them on Hyper-V Server 2008 R2, with it’s slightly smaller footprint, and equivalent feature set.
Hopefully this has given you a better understanding of Hyper-V Server 2008 R2. If you want to read more about Hyper-V Server and R2, check out these resources:
PingBack from http://microsoft.linkedz.info/2009/04/24/understanding-where-microsoft-hyper-v-server-2008-r2-fits/