virtualization.info

2010? Is this not the definition of vaporware? VMware has been doing this since 2006.

Slight correction: the first version of VMware to include the VMotion (i.e. the "Live Migration") capability was ESX 2.0, which shipped in 2004.

Good news for VMware.

Good news ... 1.5 years until doomsday.

They have already lost the small server conversions.

VMware was focused on live migration - aka VMotion - in 2004, as indicated a few posts ago. I imagine they will continue to be years ahead of the competition 1.5 years from now - hardly doomsday.

2010 if you're lucky... i've also seen reports of 2011! Real good stuff, live migration

But what about DRS, Storage VMotion, SRM, Lifecycle Management, Continous Availability etc... are they gonna spend another 3 years on those?

I don't see how competition being six years behind in the technology equates to doomsday for the market leader. I guess we'll all find out when the event happens.

Sun though they had the server market sewn up too. They thought: "Why would anyone want cheap windows servers to run anything important".

It's another piece of the cost puzzle. Why pay continue to pay VMware a lot of money for VMotion (at least for Windows servers) in 2 years, when it's part of the OS you're already licensing to do the real work?

Because there are many more pieces in vmware then just vmotion that blow away hyperv this second. And where talking 2 years to catch up with only one feature when MS will need to match about 6 more. At that rate in 12 years MS may have storage vmotion.

Anon from 9:04:00pm is on the right track. Another example of this is VMware has flexible VMotion for dissimilar processors. Makes things easier for rolling ESX host hardware upgrades. We don't know the fine details behind Microsoft's hot migration. ie. What limitations or restrictions will there be? Users are praising now for a 2 year out solution that they may not actually be able to qualify for depending on their infrastructure. There is much to be ironed out before simply declaring death to VMware.

The most important thing Microsoft taught to me is to be prudent with their announcement... 1.5 years before, it's only a rumor !

Also - to get this functionality, you have to buy the Enterprise or DataCenter version of Server 2008, which will create a memory/cpu overhead on the cluster nodes (not to mention the price) AND you need a hardware-assist from the processor, which means you have to buy the new Intel VT or AMD-V cpu's.

did any of you guys think that the live migration demo was somewhat staged? Did you see how the console of the VM playing the movie just appeared out of no where. Like it magically faded into the screen. Plus you'll notice that the vm console doesnt even show up in the windows taskbar! Like it was superimposed onto the screen somehow. Youll only see Failover cluster manager in the taskbar. Also, the zoom into the video looked quite odd. I dont recall a VM console being able to expand into full screen so nicely, and then zoom out again so nicely. Hmmmm.....MS may have live migration, if so, give us the details on how its done. Why not do at least a ping!

>It's another piece of the cost puzzle. Why
>pay continue to pay VMware a lot of money
>for VMotion (at least for Windows servers)
>in 2 years, when it's part of the OS you're
>already licensing to do the real work?

Perhaps because VMotion at that point will be free as well and what you will pay VMware for is something that MS will promise to have 3 years down the road again? Perhaps?

As to the "Live Migration" demo, think about it like this: Why couldn't/didn't MSFT do "Live Migration" in Hyper-V 1.0? To do a Live Migration you need both the source and destination hosts to be able to access the storage where the VM resides at the same time. NTFS, today, isn't built to do that. You can get that feature today from partners such as Sanbolic with their Melio FS product, but not natively...so on that thread, one can surmise MSFT will also be releasing a shared NTFS file system with 2008 R2. It's also not a stretch then to realize MSFT most likely had to balance features vs. release dates.

There are obviously other points to consider, but something to chew on...

Well here we are March 2009 and the Hyper-V Beta has been released for over 2 months now, with the RTM easily being before the end of the year same with Windows Server 2008 R2. Hyper-V is already gaining ground over VMWare, as well as Citrix's XEN platform. The costs of maintaining contracts with VMWare are getting very expensive, especially considering the hefty and restricted hardware requirements of ESX and ESXi. Now you look at Server Core and it will run any Windows drivers built for the 6 kernel, run on virtually any hardware that supports hardware assisted virtualization. The costs for Small Business to support VMWare are way to large, but if you can run it on a cheap server with Hyper-V server or Server 2008 Enterprise core (to cut down on licensing costs), you can save power time, and money with the MS solution. If VMWare would come out with something that is just as good then they would become a force again, not just a company that makes money off of licensing renewals.

On this topic, I have made an observation rather than a prediction, because I'm not a virtualization Wizard. If, from the "VMware v. Hyper-V" discussion, you abstract the talk of specific features, the discussion is reminiscent of the "NDS v. AD" and "NetWare v. NT Server" discussions. NDS and NetWare were spectacular enterprise products which performed a very specialized set of actions (authentication, file serving, and print serving). Today we regard those services as nothing special. At the time, they were special, and Microsoft had competing products which didn't scale well and were comically short on features.

In time, Microsoft improved the products and they included their functions with the OS license and all the other services provided by a Windows Server computer. Microsoft's feature sets mostly eclipsed any possible application of NDS or NetWare, and Microsoft did it at no additional cost.

This sounds a lot like where Microsoft must be going with virtualization: capturing nearly every application where virtualization is needed, leaving a significant minority but, to be sure, a minority, of specialized, high-end installations to a competitor. (Of course, in the NDS comparison, I'm thinking about eDirectory as the specialized descendent of NDS.)