On my first blog post about Hyper-V R2 which you can find here I received a lot of reactions and I always appreciate it to get any type of response (except the 380 spam comments per day). Most of the people agree with the statements and some don’t. I’d like to start this article with stating that it was not my intention to bash Hyper-V as a product, I really think Hyper-V is a great hypervisor, it’s the management what is creating the issues.
So now with the upcoming version of Hyper-V 2012, which will be released as a beta version on the 29th, I’d like to look at what’s changing in regard to my earlier article. Of course the complete hypervisor management architecture consist of Hyper-V and SCVMM 2012.
The focus of this article is (again) mainly on what it means for VDI scenarios however I think that most of the conclusions will go for whatever scenario.
So what’s new in Server 8, Hyper-V and SCVMM 2012 and why does it matter!
Windows Server 8
- Native Dedupping on network and storage
- Storage on SMB 2.2 or NFS
- Live migration without Fail-Over clustering
- Network revised with true virtual switching
- Teaming based on virtual switches (just add physical NIC’s)
- Resource metering
- No need for clustering now that SCVMM will take care of server placement
- Live IP based storage vMotion
So how do all these changes make a difference. The best way to show this is by using the picture I created for my earlier post and create a new picture based on the new product features.
The pictures below show the components of a managed Hyper-V platform. On the left the old scenario with Hyper-V R2. On the right the scenario based on Hyper-V 2012.
I hope that the before and after shots make clear that the new features will result in a easier architecture. Less components leads to less complexity and in the VDI case a lot less things that could cause issues.
I can only draw a theoretical conclusion based on the new features because I haven’t tried Windows Server 8 yet.
My conclusion is that with the new features of the different components things will start to change and get better. With the removal of Fail-Over clustering and CSV’s there are just less components which can cause issues.
With the added dedup features and the possibility of live storage vMotion over IP I think that especially for stateless scenario’s we don’t need a shared storage solution anymore. We can use cheap SSD’s for VM placement and when we need to do maintenance we just vMotion the vm’s to another server. I think this will lead to cheaper and easier platforms!
I am really looking forward to Hyper-V v3 and expect that this will absolutely change the way how I look at Hyper-V in VDI scenario’s. The hypervisor was already great and now the management surely looks mature as well.