Virtualization, High Availability and VMWare
I came across some excellent posts by Jeff on the Windows Server Virtualization blog that explains how high availability for virtual machines is achieved in Hyper-V and VMWare.
Virtualization is a wonderful technology – running more workloads on fewer boxes is a great idea – especially when it comes to cutting costs on rackspace, floorspace, power, cooling, hardware maintenance and improving resource utilization. However, IT Managers usually balk at the decision to virtualize because of one major reason – putting all eggs in one basket!
Believe it or not, this is one of the most common arguments that I get when I talk about virtualization. At the recent Microsoft Technology Day at Kuwait, I was asked the same question by the attendees during my session on System Center Virtual Machine Manager.
If you have 10 workloads running on 10 different boxes – even if one box dies, it just affects the workload running on that box. But if you run 10 workloads on one box, and if that box dies, will it bring down all 10 workloads?
The solution to this dilemma is Failover Clustering for virtual machines. Hyper-V integrates with Windows Server 2008 Failover Clustering and in an unplanned downtime scenario, when one server crashes, the virtual machines automatically failover to the passive node and each of the guest machines restart on its own – without any user intervention.
Fussy about the restart? Well, as of today that’s the best it gets – even if you have High Availability configured on VMWare, the guest machines WILL restart if a failover occurs.
So what’s the difference? Well – if you buy Windows Server 2008 Enterprise or Datacenter edition, Hyper-V and failover clustering is available to you at no additional cost!
Shijaz Abdulla is a Partner Technology Advisor at Microsoft, helping their top tier partners build on their Microsoft practice. He is also a trusted advisor to medium enterprise customers in Qatar. He is a blogger, tech enthusiast, and a Microsoft evangelist.
In today’s scenario if I have 10 workloads across 10 servers and I need to patch one, it only affects 1 server.
With Virtualization, how can I patch my hypervisor without requiring any downtime on all 10 servers?
Hyper-V Quick Migrate – you will have downtime
VMWare VMotion – You will not
And I think we all know how often MS Likes to release patches…is it Tuesday yet?
VMware Update Manager for ESX 3.5 and VC 2.5(released Dec 2007) will allow you to patch update the ESX host non-disruptively to vm workload. It vmotions vms off host….patches host …then vmotions back.
Wrong.
With Hyper-V in Windows Server 2008 R2, you will have no downtime. You do a Live Migration to move running virtual machines off, patch the host, and do a live migration to move the machines back, still running and intact with no downtime.