Slashdot Mirror
VMware-Microsoft Battle Looming
An anonymous reader writes "VMWare released a white paper detailing its concerns with license changes on Microsoft software that may limit the ability to move virtual-machine software around data centers to automate the management of computing work. Two choice quotes: '"Microsoft is looking for any way it can to gain the upper hand," said Diane Greene, the president of VMware.' And, '"This seems to be a far more subtle, informed and polished form of competitive aggression than we've seen from Microsoft in the past," said Andrew I. Gavil, a law professor at Howard University. "And Microsoft has no obligation to facilitate a competitor."'"
I wonder if he has judicial ambitions.
I'm confused here, maybe some of you poeple who use virtual machines (more than me) can help me out. I've posted a few questions and points I am either interested in, or do not understand..
=============
Where is the boundary between a "virtual machine" and a "real one"?
After all, the BIOS is definately part of the machine/motherboard and thats SW. If there is another layer of SW inbetween your OS and you HW why should that be any different? I would treat a "virtual" machine as essentially the same as a "real" one - surely in the eyes of the law they must be the same, no?
M$ changing the license restrictions seems as though they are essentialy stepping outside the OS box and determining the physical HW you are and are not allowed to run on. Whats the legal situation here, has this been tried and testing in a court?
Can they actually prevent any version of Windows from running in a VM if that version of Windows cannot detect it?
At the end of they day if a court rules a VM and a real PC are legally the same, where would that leave M$?
I suggest you to look into IBM's System i product line.
They've got a fancy Hypervisor in Hardware (called the FSP, flexible service processor). Linux is supported natively.
The Managment Console is running Linux, too.
(chorus) Switch to GNU/Linux.
You know, I toyed with the idea of modding your post down as overrated, but then I thought that it made more sense to explain why rather than just do it. To put it simply, not everybody can switch to GNU/Linux for their datacenters. In a lot of vertical markets the only products available (or at least the best products available) run on Windows. Even if there may be a reasonable GNU/Linux alternative available there can be a significant barrier to entry in the form of long-term contracts, or an entrenched user base numbering in the thousands. Let's be realistic here, for many companies (especially larger companies, the type that are most likely to use virtualization) it's not simply a matter of swapping out Vista and Office for Ubuntu and OpenOffice, and then tying it together with OpenLDAP on the back end.
In my case I work for a software company that develops enterprise application software that is used by most banks, insurance companies, and large manufacturers. We actually started as a Unix-only application, but eventually we had to start developing for Windows simply because that's what the market place demanded. Now we develop and support on both platforms. Our in-house datacenter is heavily virtualized, and our servers are split roughly 50% Windows and 50% Linux/UNIX. Phasing out Windows in our case would not only be incredibly stupid, it would literally kill the company.
Don't get me wrong, OSS is great. We use it a lot, and it has it's place. But it is not some sort of magic bullet, and it definitely is not the answer to every IT-related question.
I manage 100s of VMs with VMWare's Virtual Infrastructure, and I call bullshit on your whole post my good man.
... poor. On other side, the people who have developed their own management for cheap (e.g. blade) hosts farms - feel least urge to switch to VMs from real hosts. For well managed environment with redundant hardware it is really waste to burn CPU cycles of emulation. (*)"
"Actual state of management software is
The actual state of the management software is excellent. Have you used Virtual Center 3? Care to elaborate with your details? I can: it's easy to organize a huge farm of VM's, prioritize them based on their CPU and RAM needs, move VMs from one physical host to another on the fly, generate alerts on nearly anything and customize actions based on those alerts, on and on. There is only one thing I don't like so far about VC/VI, and that is how LUNs on the ESX hosts are managed.
"VMs solve no particular problem, but just propagate problem of poor OS management to another level - hardware/emulated hardware."
It solves two problems off the top of my head: hardware proliferation, and hardware failure.
We have a glut of CPU speed and RAM space, but the chasses surrounding those chips are still expensive. Not just expensive in terms of the box itself, but in the cost of powering and cooling the server as well as the spacial cost of having it in the rack. The solution is to have one physical server do more than one job. The way to do that safely and sanely is to virtualize.
Another major problem VMs solve is hardware failure. Right now, in my own virtual datacenter, I pretty much don't care about anything less than a whole blade going down. If there is a single blip in the hardware, Virtual Infrastructure migrates the VMs off of the faulty hardware automatically, *without interrupting the VMs*. In other words, for 99% of all hardware failures, my VMs keep running and I can swap out the bad physical server without disturbing anything. Zero downtime. I still cluster at the OS level, too, so in reality I am only worried about my whole datacenter being nuked. Similarly, I don't care too much about hardware replacement cycles anymore either. When a warranty expires, I push all the VMs off of the box and unrack it. We unpack the replacement box, rack it & install ESX, and that's it. The Virtual Center will automatically shuffle VMs onto the new hardware.
"In the end, when box is plain hardware you can always pull the plug. Try to emulate that with compromised/erroneous VM which started hogging all system resources from other VMs - and management interface too."
Hehe. 'You try to open the door, but *there's too much blood on the knob!*' You are defining an extremely specific problem and treating it like it's a general issue with virtualization, but I will play along.
Here is what would happen in my environment. First the VC would automatically migrate all of the other VMs away from the crazy VM. Once the crazy VM was isolated, I could pull the plug on the box if I felt like that was my only option. Then I would activate a clone of the VM (I clone every freshly minted server and put the clone on ice--try doing that with nothing but "OS management"), restore data from backup if necessary, and away we go on an fresh, uncompromised server. Meanwhile I can go back and examine the crazy VM at my leisure and have it completely sandboxed.
The state of the art in virtualization is not a single server running a few linux instances, sir. Check out where VMWare (and probably Microsoft) is heading and you will see that things are rapidly evolving. I'm talking a quantum leap in datacenter management here, the biggest thing since commodity hardware.