Slashdot Mirror
VMware, XenSource Join Forces For Linux
porjo writes "Peace has been established on at least one front: XenSource and VMware are working together to improve virtualization in the Linux kernel. Their original disagreement has been displaced by a commitment to work on a solution together, says Simon Crosby, CTO of XenSource, the company that builds products around Xen virtualization software. The two are trying to come up with a common approach to virtualization support in the Linux kernel. [snip] The work now under way would let hypervisors from Microsoft, VMware, and Xen work together in the same data center. Under such a scenario, it would be possible for a Xen virtual machine, trapped on a piece of failing hardware, to be automatically moved over to a VMware hypervisor on another piece of hardware."
If this is true, then its a great news for standards, since atleast I hope that now anyone (possibly) could write a hipervisor for unknown/research systems with a lot of testing and deployment testing available.
They called me mad, and I called them mad, and damn them, they outvoted me. -Nathaniel Lee
It is great to see that big software manufacturers can help each other and build up a better product. Though, with all my reservations, I might not know enough about this debate to say this. I will do that anyway and you can read it the way you want. Consider the fact that every company making a major product would start helping each other. As far as this drive the development further it's great! This might though carry the risks for having an oligopoly driving the prices forward which is not so great. Also consider the fact that the companies that have the most use for virtualization probably have such a great win in testing that they could pay high prices for a product worth less. All this brings virtualization away from people who just need it for private use. I should also tell you that I don't even know in what extent these products do cost anything.
+1 Agree -1 Disagree
Maybe some day we'll only power down our computers after uploading them-still running-to an off site server to continue their existence. An then some day those never ending processes can mate and have children.
Suck a lemon?
Now this is great news! As VMware user on production systems, I am very pleased with such news. Now one of many thing that can go wrong in such "alliance" as that hypervisor interface will get bloated with vendor specific extensions. And we will end up with non-compatible interfaces as it was before.
I'd really like to run a virtual machine environment, so I've got Linux or OpenBSD underneath and a Windows client OS on top for when I want to run Windows applications. There seem to be a range of choices if I want Linux client OS, including Xen, VMWare, User-Mode Linux, etc., and some for BSD client OS, but is the VMWare server for Linux the only free choice if I want to support Windows clients? Free-beer is good enough, and I'm not a gamer, so accelerated graphics performance isn't critical either. If platform matters, I'm running Intel Celeron, so some of the newer hardware tricks probably aren't available.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Run VMware to sandbox MS DRM crap and Xen to sandbox VMware!
Under such a scenario, it would be possible for a Xen virtual machine, trapped on a piece of failing hardware, to be automatically moved over to a VMware hypervisor on another piece of hardware.
This is just like that episode of Star Trek where Professor Moriarty succeeds in escaping from the holodeck!
It's only a matter of time before the hypervisor/virtualization layer is a commodity - and with standardized interfaces, the vendors can focus on infrastructure management software.
VirtualCenter is imho way ahead of anything else available - and will be VMware's most important product going forward.
Is this article talking about the paravirt_ops API that is being discussed on linux-kernel lately, or something else?
Offtopic? Some reading suggestions for the mods. Mannie and Mike and Lazarus will provide you with some info on how the parent relates to the topic. Robert A. Heinlein was an award winning author who many nerds felt an affinity with. If your from a family of nerds your ancestors would probably have recognized the references. Perhaps a Wiki or two will help.
At one point Heinlein sends Lazarus on multiple time travel trips rescuing his various heros and heroines of his numerous novels. This includes (at least by my less then perfect recollection which I can't check cause the ex got custody of my library) a failed attempt at rescuing Mike the sentient computer of the Lunar revolution. "Under such a scenario, it would be possible for a Xen virtual machine, trapped on a piece of failing hardware, to be automatically moved over to a VMware hypervisor on another piece of hardware." See? Method of escape for Mike. Levity for those who recognize the reference in the parent post.
Please mods, if you don't understand a reference then don't mod it down. Especially on an AC post where it will lanquish at 0 anyway if no one finds it funny, interesting, informative or insightful and also has mod points they think it's worth spending on, which raises the question of what was so bad about it that it rated wasting a mod point on moments after it was posted? IMO modding a non-offensive attempt at humor by an AC to -1 is a waste of a mod point.
Nobody is really going to use this. When people talk about this, it's like saying, "if it's the 3rd Tuesday of a month that ends in 'ber', I'm in an important meeting, sitting in my assigned seat, and I spill coffee on my shirt but not my tie, I can totally switch my shirt without taking off my tie with only a small hiccup in the meeting agenda that we can train the attendees to work around as long as they're sitting in their assigned seats! Isn't that great?! Let's set up our systems to support this and assign the seats now! Move all critical meetings to the 3rd Tuesday of the month!" You'll pick your favorite VM engine and hypervisor and run all your VMs in it. You might have individual users like developers running groups of VMs under a different hypervisor, but you'll be hard pressed to find an excuse to transfer a running VM versus rebooting it. And you almost never "pre-detect" failing hardware and transfer a running machine. I'm constantly reminding people that vmotion and transfering running machines has everything to do with scheduled maintenance, and nothing to do with disaster response. You can't currently (and don't want to) transfer a VM off a local disk over the network, to another disk. Transfering depends on a SAN and fast uncongested network. If your disk controller is failing, you're not going to transfer. If your nic is failing you're not going to transfer. If your CPU or RAM starts glitching you'll be very lucky to successfully transfer a blue-screened OS. If your power drops and you're running on battery and want to transfer somewhere else, you need so many prereqs like a bridged network to somewhere where there's still power, and a mechanism to seamlessly switch your users over... Maybe? If you're this size and budget you probably have the same brand of hypervisor as a hot spare in the remote bridged, SAN replicated, alternatively powered site.
The really important news here is that they're not going to be forking the kernel. Xen and VMware were submitting patches that weren't compatible. If Morton and Torvalds went with Xen's patches, then they wouldn't consider similar but different VMware patches. It'd be redundant. So VMware would need a forked kernel to put their patches in.
Microsoft will never jump in and run a hypervisor on Linux, and if they wanted to they had wanted to last week, they'd need to submit patches to Linux to compete effectively. Extremely unlikely. With this news, MS could write a VM engine or hypervisor to run under Linux. The earlier announced partnership between MS and Xen is simply getting Xen to help make sure Linux VMs work in MS Virtual Server. (In other words, the patches Xen is submitting to the Linux kernel will help make sure that Microsoft can get a few snippits of info from a Linux VM running under MS Virtual Server. It's not really important.) I *WISH* MS would tweak Windows to run more smoothly on hypervisors. But I predict that they'll only be tweaking it to run better under Virtual Server.
I guess I'm just curious. Isn't Virtuozzo and its GPL'd counterpart, OpenVZ, also a legitimate virtualization contender? Does anyone know why they weren't included in these discussions? I thought that they were the first company to request that their hypervisor be added to the mainline kernel.
BTW, Qemu on x86 with the kqemu acceleration module works great with Win2k for basic software, e.g. it's still a bit slow for intensive video / sound apps.
how about creating an operating system that isn't more full of holes than swiss cheese?
Actually, it sucks that you had a bad experience with Linux, so much so that you'd rather use what I consider to be one of the worst operating systems ever perpetrated on the computing public, Windows 98.
I partially agree with you, though. Virtualization is a great tool, but I hope we don't start using virtualization as an excuse to write poorly written and insecure code.
I think that security and usability have to be balanced with each other, and using Win98 is the electronic equivalent of wearing a sign on your back that says "Kick Me" and a T-shirt with a bulls-eye on the front.
Sorry mate
Can someone explain to ignorant old me what linux-on-linux virtualisation is for? What problem does it solve?
QEMU is a dynamic translator, ie like bochs but MUCH faster. Not only, there is KQEMU that is a kernel module to enable true virtualization for QEMU (like VMWare). Unluckily, KQEMU is not open source, but the creator Fabrice Bellard is willing to release the source if he gets the money he deserves. I think QEMU/KQEMU deserves more attention...
Wondering why i am doing so strange posts? I am trying to get a "+5,Flamebait" or "-1,Insightful" rating.
It'd be really great if they'd also learn the lessons and great ideas from the mainframe, where IBM has been doing virtualization (VM, now z/VM) and a decoupled hipervisor (that's their term I believe) for more than 35 years now. It's clear that the hipervisor in System i and System p are all gearing up to really allow one to move VMs from one place to another as the situation demands.
Disclaimer: I do not work for IBM, but I think they've got great technology. I'd love to have the dough to buy a new System z9 and load it up with all things Linux.
Ich suche die Leidenschaft, die keine Leiden schafft.
this is sooo boring ...
Too bad :-) I'm running an old basic P4 Celeron, so I'll need something besides Xen. QEMU sounds like an interesting option, or else there's VMWare.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Parallels really is the counterexample to this paper, the speed is pretty much the same in hardware VT-X or software only, but the main difference is that the processor load is way lower once VT-X is turned on.
There is an 8139too network target too (which is 100mbps), which should be more than acceptable, but I think you need to build Xen yourself to get it.
Also, what of giving the guest OS it's own network card through a PCI mapping? I always have more onboard/outboard network controllers than I know what to do with in my servers. It's not the most transparent way of going about things but if you have a bandwidth-hogging virtual host to run...
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
Also it would be nice if we got a few more emulated devices supported. Maybe a gigabit ethernet adapter. A CDROM 'server' that allows you to attach and unattach ISOs (which might look like insert/remove events).
The holy grail would be M on N threading. But I don't think that fits into his emulation model very cleanly.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
... is old hat. We know how to do that, and we can do it easily.
But z/VM can't virtualize Windows.
That's what we're trying to do within an infrastructure that we control.
IBM can't help us there.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON