Jeff through up this great post on Hyper-V quick migration. I think this is a really interesting post as it generally gets me quite frustrated when pep's dont compare apples for apples! and the whole Quick Migration, Live Migration (Vmotion) bit is something that is never compared in the way it should be!
Generally speaking, I like to focus my blogs on what we’re doing at Microsoft regarding Hyper-V virtualization and pretend the rest of the Internet doesn't exist. However, there’s some buzz on the web about a topic that I feel I need to address. Today, this blog is the first of a multi-part blog on the topic of Hyper-V’s High Availability/Quick Migration capabilities compared to VMware’s VMotion (Live Migration) capabilities.
Before I dive into details, let me take a step back and discuss why high availability is absolutely CRITICAL to virtualization.
Virtualization is an awesome technology. It provides numerous benefits for reducing overall TCO, one of the most obvious benefits being power consumption savings. If you have a data center with 10,000 servers and you cut that number in half with virtualization (2:1 consolidation) you will achieve very tangible power and cost savings by retiring those 5,000 servers. Just look at your monthly power bill. Honestly, 2:1 consolidation is dead simple. (In fact, our own internal IT has been using Virtual Server for years now IN PRODUCTION with over 2,500 virtual machines and easily achieves 8:1 consolidation with four 9’s uptime. With Hyper-V, we see those consolidation ratios climbing in a big way.)
However, virtualization isn’t perfect. Virtualization actually creates a major problem: single point of failure. Think about it. In the past, you may have been running 20 workloads each on their own physical server. When one of those servers goes down it’s bad, but probably not the end of the world. In a virtualized environment, suppose you have 20 workloads running a top a single server. What happens when that physical server goes down? All 20 workloads go down. That’s not bad, that’s catastrophic. In fact, I’ve talked to numerous virtualization customers that have told me point blank:
Virtualization is great in test and dev, but there’s no way I’m deploying virtualization in production without a high availability solution. If that virtualization server goes down and I don’t have a HA solution in place, I will lose my job.
That sums up the importance of high availability with virtualization pretty well.
With this in mind, we knew we had to provide solutions for both planned and unplanned downtime. Planned downtime is the easier of the two (because it's scheduled, not a surprise) and the most common. Generally, planned downtime is for hardware servicing (adding additional memory, storage or updating a BIOS) or software patching. Most people schedule this work off hours (early mornings or on weekends).
Unplanned downtime is the more difficult one, where a server is unexpectedly powered off and you want the virtual machines running on that server to automatically restart on another server without user intervention. This is the scenario that IT told us we have to solve first and that’s exactly what we did. Hyper-V integrates with Windows Server 2008 Failover Clustering so that if you pull out the power plug, all of the virtual machines will automatically restart to another node without user intervention. Furthermore, this capability is simply included (at no extra $$$) with Windows Server 2008 Enterprise and Datacenter Editions.
So, if there are a few points I want to make today:
- Virtualization & HA go hand in hand; If you’re virtualizing today without HA, you should re-evaluate that strategy.
- Windows Server 2008 Enterprise and Datacenter editions provide Hyper-V and integrated HA support at no additional charge.
In my next blog, I’ll discuss planned downtime and go into detail on Quick Migration and how it compares with Live Migration.
SIDE NOTE: For those of you who haven’t had the opportunity to checkout Windows Server 2008 new failover clustering, you should. It’s incredibly intuitive and easy-to-setup. You can setup a 16 node cluster in four steps. I kid you not.