Xen High-Performance x86 Virtualization Released

Xen Team writes "The University of Cambridge Computer Laboratory Systems Research Group is pleased to announce the open source release of Xen, a virtual machine monitor for x86. Xen lets you run multiple operating system images at the same time on the same PC hardware, with unprecedented levels of performance and resource isolation. Even under the most demanding workloads the performance overhead is just a few percent --- considerably less than alternatives such as VMware Workstation and User Mode Linux. This makes Xen ideal for use in providing secure virtual hosting, or even just for running multiple OSes on a desktop machine."

The Xen team continues: "Xen requires guest operating systems to be ported to run over it. Crucially, only the kernel needs to be ported, and all user-level application binaries and libraries can run unmodified. We have a fully functional port of Linux 2.4.22 running over Xen, and regularly use it for running demanding applications like Apache, PostgreSQL and Mozilla. Any Linux distribution should run unmodified over the ported kernel. With assistance from Microsoft Research, we have a port of Windows XP to Xen nearly complete, and are planning a FreeBSD 4.8 port in the near future.

"Visit the project homepage to find out more, and download the project source code or the XenDemoCD, a bootable 'live iso' image that enables you to play with Xen/Linux 2.4 without needing to install it on your hard drive. The CD also contains full source code, build tools, and benchmarks. Our SOSP paper gives an overview of the design of Xen, and evaluates the performance against other virtualization techniques.

"Work on Xen is supported by UK EPSRC grant GR/S01894, Intel Research Cambridge, and Microsoft Research Cambridge via an Embedded XP IFP award."

Pfff

[AKAImBatman (238306)]

> Xen requires guest operating systems to be ported to run over it.

Get me all excited, then pull the rug out from under my why don't you? This is still pretty neat, but it's hardly a replacement for VMWare or Bochs.

Re:Pfff

[JohnFluxx (413620)]

I got excitted that windows would be ported to use it.. but then:

"The Windows XP port is nearly finished. It's running user space applications and is generally in pretty good shape thanks to some hard work by the team over the summer. Of course, there are issues with releasing this code to others. We should be able to release the source and binaries to anyone that has signed the Microsoft academic
source license, which these days has very reasonable terms. We are in discussions with Microsoft about the possibility of being able to make binary releases to a larger user community. Obviously, there are issues with product activation in this environment which need to be
thought through."

It would be a bitch if it was ported, worked perfectly, but then nobody was able to use it.

Re:Pfff

[spektr (466069)]

The Windows XP port is nearly finished. It's running user space applications and is generally in pretty good shape thanks to some hard work by the team over the summer.

I'm surprised that this was possible at all with Windows XP. Microsoft's Shared Source program doesn't seem to be as useless as I thought.

Of course, there are issues with releasing this code to others. We should be able to release the source and binaries to anyone that has signed the Microsoft academic source license, which these days has very reasonable terms. We are in
discussions with Microsoft about the possibility of being able to make
binary releases to a larger user community.

I think there are two possibilities to do this. First, Microsoft incorporates the changes into their main trunk or releases patches for it. Second, Microsoft allowes a group of hackers to distribute modified Windows binaries. Both alternatives don't seem very probable.

Re:Pfff

[Fluffy the Cat (29157)]

I'm surprised that this was possible at all with Windows XP. Microsoft's Shared Source program doesn't seem to be as useless as I thought.

Microsoft Research UK is literally next door to the University of Cambridge Computer Laboratory. Microsoft Research UK desperately wants the university to like it in order to gain more academic credability.

xp licence/activation .

[leuk_he (194174)]

If they ever release the XP binary you will have to bet that it will be some hight priced version of xp that runs it. Or else you will run into all kind of license problems.

If it will be a concurrent of The virtual machine solution they bought from connectix [microsoft.com] This will not be released.

And if it will work with a standard XP home/pro you will have all kind of activation [vmware.com] loopholes like in vmware.

Re:Pfff

[rmohr02 (208447)]

Only the kernel needs to be ported, and they have ported Linux 2.4.22 for you, and they're working on NetBSD, FreeBSD, and have almost finished Windows XP.

Re:Pfff

[websensei (84861)]

read the article.

Xen only requires the OS *kernel* to be ported, not apps. For most users this is more than "good enough"... and makes it a viable replacement for vmware, especially given the reported (tiny) resource overhead.

Re:Pfff

[kzinti (9651)]

In being so quick to chastise the guy for not reading the article, you missed the point. If anything at all needs to be ported, it means that an OS like XP can't be installed on Xen out of the box. It means that a port of the XP kernel has to be completed, and that MS has to authorize the ported kernel to be distributed. For lots of folks, this is not good enough... even assuming that MS ever would allow the ported kernel to be distributed (and then we have to worry about activation issues too). So, while Xen is great for allowing linux users to run multiple distros at max performance at the same time, it doesn't much help someone who wants to run XP, and might never help them. Those folks still have to resort to VMWare or Bochs.

Pfff...

How many licenses per machine?

[Glonoinha (587375)]

Silly question about activation .. actually not -about- activation, but inspired by activitation :

If you only have a single computer with a single CPU, how many copies of WinXP do you need? That one is rhetorical of course, and the answer is One.

Can you run whatever software on that legitimately licensed WinXP machine that you like, assuming it was also legitimately licensed? That one is also rhetorical and the answer is Yes.

Now install VMware on that machine, WinXP as the host OS. By adding VMware you have not increased the number of CPUs or physical machines. If you created three virtual machines (if you had enough RAM and hard drive space, not a stretch at all) and wanted to run WinXP in each of those virtual machines simultaneously - do you need 1 license of WinXP or four licenses of WinXP (one for the host OS, and one for each VM)?

Granted the activation and active license management in XP may not allow this to happen even if in theory it should be allowed according to the 1 license / physical machine license in the EULA - but swap it with Windows 2000 or whatever ... what are the facts?

I am just curious.

Re:How many licenses per machine?

[steve_l (109732)]

I think it is something they havent had to deal with yet.

I am running Win2K+visual studio in one vmware vm right now; I can bring up Win2K3 to run office 2003 when I need to go on exchange. But with the block MSDN license *and* a volume 'no activation' key for XP, office, etc. I get to skip activation.

But imagine if I did have to activate stuff everytime I rebuilt a new VM? Within a month I'd have the activation police complaining I'd activated onto 5+ systems, and that therefore I was violating some lice

Re:Pfff

[jdh28 (19903)]

No, VMware runs unmodified binaries.

What they do provide is ready made installations of various operating systems you can just install, although you have to pay for these (well, the MS ones at least).

Re:Pfff

[vrt3 (62368)]

Yes, but VMWare still needs to make concessions for each supported OS. Yes, you can install and run Solaris/86, pre-v6 Netware versions, plan9, and BeOS under VMWare (with varying success), but they're not supported. They need to tweak VMWare for each OS.

I could be wrong, but I didn't think so. Not supported means they don't support it, not that it doesn't work. The tweaking that is done when you select a specific OS only concerns the configuration, such as amount of memory and stuff. Things you can easi

Re:Pfff

[Martin Blank (154261)]

I could be wrong, but I didn't think so. Not supported means they don't support it, not that it doesn't work. The tweaking that is done when you select a specific OS only concerns the configuration, such as amount of memory and stuff. Things you can easily adjust yourself.

Correct. Debian's not supported, but it works. OpenBSD isn't supported, but it (sort of) works. Basically, you have to be prepared to deal with some frustration in things being slow or sometimes unstable, but this does not have to do

Re:Pfff

[kscguru (551278)]

I'm unsure why Xen needs guest OS mods, vs the way VMWare and Plex86 do things. I only skimmed the paper, so I might have missed something.

VMWare and Plex86 need to trap priviledged instructions in running code and replace them with their own. I believe they do this by literally scanning code-to-be-executed, which slows down the virtual machine of course. Xen solves this by replacing all such instructions with Xen system calls; somewhat the same as User Mode Linux, where low level instructions are replaced with system calls to the host kernel.

Not quite ... they just run the virtualized kernel in plain user-mode, then when a privilaged instruction comes up it trips a processor exception. The VMM (VMware or Plex86) then steps in, emulates the privilaged instruction, then gets out of the way. Privilaged instructions, for the non OS-enthusiasts, are usually hardware-related things like INP and OUTP, instructions to change the page tables, switch between 8086 and protected mode, etc. Since these instructions are so rare, VMWare is only ~10% slower than normal - and runs full speed on a large set of workloads.

Not having read the Xen site yet, what they _probably_ do is replace all such instructions in the source code with their own function calls. VMware did something similar at one point, but didn't turn it into a product because they felt it didn't have commercial potential.

Re:Pfff

[kma (2898)]

Someone was bitching "why can't XYZ OS run under VMWare if it runs perfectly well under generic PC hardware". Someone responded to the effect that VMWare was tuned to make certain concessions for the OSs they supported.

Right. The OS'es we "support." Not the OS'es "we're able to run." And those "concessions" are mostly performance trade-offs, not correctness.

It's still true that some OS'es don't run; but that's because our software has bugs. OS'es sometimes have bugs, too, though if the OS is important enough, we'll work around it. The bugs that prevent you from running the OS that some drunken Swede cooked up for course credit are admittedly less important to fix than the bugs that, say, prevent you from running Linux. However, in the long run, we try to fix even the bugs exposed only by drunken Swedes.

That's why AtheOS, OpenStep, BeOS, NetBSD, FreeDOS, B-Right, Plan9, QNX, and myriad other commercially unimportant OS'es run ok. Not because they're important to our customers (man, oh man, they aren't), but because they enable us to be sure our x86 virtualization layer is reasonably correct. That way, when Ingo Molnar decides to start using 80286-style 16-bit tasks with lots of grow-down and conforming code segments to do system calls in Linux 2.6, we won't get caught out too badly.

Re:Pfff

[kalidasa (577403)]

MS is one of their sponsors. How much you want to bet that MS licenses the technology and distributes it under their own logo? We already know from the VirtualPC purchase that this is a future direction for them.

Funny Guys

[tomblackwell (6196)]

... to publish a paper called "Xen And The Art of Virtualization".

MOL for x86?

[Realistic_Dragon (655151)]

In the same way that Mac On Linux makes moving people to Linux on a Mac pretty painless (just give them an icon for OSX in a window) this might do the same for migrating Windows users.

The biggest problem with emulators/virtualisation has always been speed. If a system can be set up that runs Linux but can boot XP easily and run fast, that will be a big improvement.

Of course it's not going to be much good for gamers (doesn't look like it can use hardware accelaration) but it's still pretty promising.

Re:MOL for x86?

[tgd (2822)]

This would not allow you to run Windows under Linux... it would allow you to run Windows and Linux under Xen, which is not nearly the same thing.

Contrary to the submitter's comments, this product is nothing like VMWare.

Interesting ..

[Anonymous Coward]

I'm really surprised they got assistance for windows XP. You'd think that Microsoft would want to stay well clear of anything linked to the GPL.

I guess such assumptions are incorrect, and quite probably a result of reading Slashdot too much!

Re:Interesting ..

[Gingko (195226)]

This is Cambridge computer lab we're talking about - where Microsoft research have the very next building over.

I live with one of the guys in the systems group and hope to go back there to do my PhD soon, and they do do very cool things there. Microsoft give them all the help they need, because academia is an excellent ideas feeder for the real world. Cambridge, being in a position of power with its serious reputation and fantastic set of minds, gets the benefit of the Microsoft help without any of the assumed costs.

Henry

Not really like VMWare

[javatips (66293)]

We have a fully functional port of Linux 2.4 running over Xen

With assistance from Microsoft Research, we have a port of Windows XP to Xen nearly complete, and are planning a FreeBSD 4.8 port in the near future (volunteers welcome!).

If one need to port an OS to make it work within Xen, then I will NOT compare it to VMWare. VMare can run your stock OS on a VM whithout the need to tweak it.

The performance advantage it has over VMWare is probably related to that. By having a few restriction on the OS, they can probably offer better performances.

Re:Not really like VMWare

[rmohr02 (208447)]

They've ported Linux 2.4.22 and are nearly done with a port of Windows XP. If it's relatively easy to change these, I will definitely try it.

Re:Not really like VMWare

[borgheron (172546)]

This depends on what is meant by "port". Is it possible that they mean that they have written drivers to work on those operating systems to make it work?

I believe more information is needed on this before I make any decision as to whether it's better or worse than VMware.

GJC

Welcome

[show me (696663)]

I, for one, welcome our new Xen masters.

so its just an extra layer of abstraction?

[ikoleverhate (607286)]

So really, this is just an abstraction layer that means even the OS is unaware it's sharing hardware, so in theory theres no way for a malicious user to take advantage of other users. Pretty cool in a boring and limited sort of way. Kudos to the team who did it, I'm sure it's a real technological challenge. Not what the /. headline promised though ;)

ob. Beowulf cluster comment

[Vic Metcalfe (355)]

Finally I can create a Beowulf cluster without the clutter of all those machines!

The only thing I'm wondering...

[metroid composite (710698)]

...is if you can copy and paste between the OSs. That always annoyed me when I had Linux open remotely through a Windows machine. I had to leave a submission form open on my website as a "back door" to copy stuff in. This goes for things like Gnumeric to Excel data too.

Comparison with other free VMs?

[Laur (673497)]

So it looks like this is the third (or fourth) free VM for Linux, the others being Plex86 [nongnu.org] (and a different fork here) [sourceforge.net] and User Mode Linux [sourceforge.net]. Does anyone have a good comparison of these three? I know Zen compared UML on their site but not plex86. I'm not really sure of the differences between them, particularly the different versions of plex86 and UML (Zen explained their virtualization process pretty well on their site). Which is the best choice for different scenarios? It looks like Zen is the winner for running Linux as the guest OS, and the original Plex86 (first link) is the only one which offers a free choice in guest OS's.

Re:Comparison with other free VMs?

[emj (15659)]

Well:
User Mode Linux

1. Is linux Ported to the linux syscalls
2. Has been ported partially to windows as well
3. No hardware access only software modificatade to be hardware
4. Is just Linux
5. Extremly Easy do install 2 Files to download and then start NO configuration
6. Very Fast emulation of linux, 10% Slow down while compiling the kernel
7. Support SWAP partition as memory for virtual machines

Plex86

1. Is X86 & Bios ported to C
2. Runs multiple OSs
3. Runs on multiple OSs
4. Very hard to conigure
5. SLOW!
6. Thourgh Em

Re:Comparison with other free VMs?

[Laur (673497)]

Thanks for the info, however you seem to be confusing Plex86 with Bochs [sourceforge.net]. They share a lot of code, however while Bochs emulates the entire hardware (and thus can be used on any platform) Plex86 virtualizes many things and so can only be run on x86 hardware, but is supposed to have much better performance because of this. Also, AFAIK Plex86 is only available for Linux (as the host OS). The original Plex86 [nongnu.org] should run any x86 OS. The new Plex86 fork [sourceforge.net] simplifies the virtualization even more and only runs Li

Virtual Virtual Machine?

[Stiletto (12066)]

So when do I get a Virtual Virtual Machine, to allow me to run all these VM's at the same time? Because as we all know, it makes much more sense to do this, than, oh I don't know... Buy another $299 computer?

Nice jab at Mozilla!

[seanmeister (156224)]

"....and regularly use it for running demanding applications like Apache, PostgreSQL and Mozilla."

That's kinda funny, lumping a web browser in the same category as server apps designed to handle gazillions of users.

Come on, it's not THAT bloated!

Re:Nice jab at Mozilla!

[seanmeister (156224)]

No, you haven't been trolled. Not by me, anyway. The text I quoted appears in the article as posted on /. [slashdot.org] ... I R'd TFA a bit later and saw that the actual article didn't mention Moz!

For all those not just out of nappies/diapers...

[jantheman (113125)]

Oh, you mean this [old-computers.com] then?

(sry NYI)

User Mode Linux

[shird (566377)]

Sounds an awful lot like usermode linux. Get a vesion of the kernel/OS which makes calls to a host virtual machine rather than directly to the hardware for privileged instructions. Everything else can run directly through the CPU without being emulated because it is running in 'usermode' (or ring 1).

Why Microsoft+Intel? NGSCB backward compatibility!

[NZheretic (23872)]

A quick glance over Xen group's paper [cam.ac.uk] leaves me very impressed with the performance these techniques can achieve. That the Xen group has decided to relase the code under the GPL leaves me very greatful. However, that both Intel Research and Microsoft Research has funded it, leave me somewhat concerned.

As I have stated before about Microsoft's purchase of Connectix's Virtual Server technology [oreillynet.com]

In my opinion Microsoft's acquisition of Connectix's Virtual Server technology has very little to do with running any other vendors operating system.

Microsoft needs a Virtual Server for backward compatibility for it's NGSCB ( Next Generation Secure Computing Base [microsoft.com] ) DRM ( Denial of Rights Mechanism [slashdot.org] ) platform.

Just as Microsoft's XP backward Win9x compatability opens up many locally exploitable API to gain SystemLocal privilege access, to the point where many programs need Adminstrator privilege to run, existing XP and win2k software would open up too many opportunities for helpfull hacker to bypass Microsoft's NGSCB DRM mechanisms.

Microsofts all too obvious solution is to provide a "Virtual" PC mode, running a modified XP and WinME, with the NGSCB providing virtual filesystems and hardware access. All, access of course, with the NGSCB DRM scanning and control.

Where do you want to go tomorrow?

The Electronic Frontier Foundation [eff.org] are about to publish a paper [com.com] criticizing a component of the "trusted computing" technology promoted by Microsoft, IBM and other technology companies, calling the feature a threat to computer users..

Hardware support

[Anonymous Coward]

According to the README, it requires special hardware drivers and is not targetted at desktops. Don't expect stellar graphics performance. VMWare *does* give you something for the money.

Hardware support
================

Xen is intended to be run on server-class machines, and the current
list of supported hardware very much reflects this, avoiding the need
for us to write drivers for "legacy" hardware. It is likely that some
desktop chipsets will fail to work properly with the default Xen
configuration: specifying 'noacpi' or 'ignorebiostables' when booting
Xen may help in these cases.

Xen requires a "P6" or newer processor (e.g. Pentium Pro, Celeron,
Pentium II, Pentium III, Pentium IV, Xeon, AMD Athlon, AMD Duron).
Multiprocessor machines are supported, and we also have basic support
for HyperThreading (SMT), although this remains a topic for ongoing
research. We're also looking at an AMD x86_64 port (though it should
run on Opterons in 32-bit mode just fine).

Xen can currently use up to 4GB of memory. It's possible for x86
machines to address more than that (64GB), but it requires using a
different page table format (3-level rather than 2-level) that we
currently don't support. Adding 3-level PAE support wouldn't be
difficult, but we'd also need to add support to all the guest
OSs. Volunteers welcome!

We currently support a relatively modern set of network cards: Intel
e1000, Broadcom BCM 57xx (tg3), 3COM 3c905 (3c59x). Adding support for
other NICs that support hardware DMA scatter/gather from half-word
aligned addresses is relatively straightforward, by porting the
equivalent Linux driver. Drivers for a number of other older cards
have recently been added [pcnet32, e100, tulip], but these are not
recommended since they require extra packet copies.

Microsoft Research

[non (130182)]

in case anyone forgot, not only is microsoft research [microsoft.com] their neighbor, but it was also the first microsoft research center outside the us. wired [wired.com] has more [wired.com] about what you can get for $80 million.

Source code not available

[avenj (673782)]

The source code is only available via BitKeeper, as far as I can tell from their site. The BitKeeper tools are not free software and cannot be used by anyone who has contributed to a competing product, according to the license for the free-as-in-beer version of BitKeeper. Is there somewhere else to get source and I'm just missing it?

Great Idea! Why all the whining?

[HopeOS (74340)]

The x86 architecture is not designed to be virtualized, at least not for code expecting to run in ring 0. The work that VMWare must do to emulate ring 0 without the guest OS coming unglued is simply daunting, including traps for every detail from setting and clearing interrupts to modifying page tables. A lot of that work requires predicting code execution paths and dynamically modifying the code at runtime. It's a real nightmare.

Plex86 promised to provide an open-source solution to this problem, but the last time I checked (which admittedly was awhile ago), they were planning on using bochs to emulate ring 0. This is not exactly satisfactory, but I don't blame them. It's a hard problem, one for which VMWare is entitled and deserving of reward for having solved.

So how do you get real virtual host performance out of your x86 machine? You design around the flaw. Operating systems that are written to run in ring 1 and call the Xen hypervisor instead of performing ring 0 functions will run at nearly full speed. Those that do not or will not make these compromises will see at best, an emulated ring 0. It's really that simple.

Personally, I don't see much value in being able to run any arbitrary operating system in a single environment if it's impossible to get real, sustained performance out of it. Cross-platform testing: sure. Kernel debugging: certainly. But host virtualization? What I want is the capability of running multiple hosts simultaneously, and if the operating system needs delibrate tweaking to make this possible, then obviously that's the direction to go.

To answer the gripes of people who want just that transparent hosting of unmodified OSes, it would be interesting to see a follow-on project for dynamically modifying the guest OS to thunk-out all the privileged calls to use the Xen extensions directly rather than trapping exceptions as they do now. This would probably not work well for page tables modifications, so a hybrid system might be employed. In this fashion, the best of both worlds may be achieved.


-Hope

Why Is This So Hard?

[Chasuk (62477)]

Is it really that fucking difficult to write an accurate summary?

A person reads the summary, and reasonably thinks:
"Great! An Open Source equivalent of VMware Workstation! Now I'll be able to run multiple OS's on my Desktop machine without the expense!"

Then he reads:

"Xen requires guest operating systems to be ported to run over it..."

This changes the picture dramatically, and should have appeared MUCH earlier in the summary.

Come on, I know this is only Slashdot, but stupidity and dishonesty like this get really annoying.

Xenify, anyone?

[leob (154345)]

With all the achievements in x86 binary code analysis, instrumentation and compilation (think of valgrind or Transmeta, to name just a few), wouldn't it be possible to read the kernel binary into some kind of a compiler, find all uses of the instructions that have to be replaced, replace them (and the adjoining code, if necessary) with the Xen interface calls, then reassemble and relink the kernel?

How extensive and how non-trivial are the necessary changes?

Re:I bet it's not Open Source...

[Zigg (64962)]

RTFA. It's GPL.

(waiting for the 20 second delay to expire... @#%$! Slashdot)

Re:I bet it's not Open Source...

[Trigun (685027)]

With assistance from Microsoft Research, we have a port of Windows XP to Xen nearly complete

They won't release the source for XP, but you can probably get a compiled binary. I just wonder if you'll have to re-register every time you change your virtual hardware.

Re:I bet it's not Open Source...

[Aspasia13 (700702)]

From the page (which hasn't bet /.'ed... yet)" you can read that Xen itself is GPLed:

"Modern computers are sufficiently powerful to use virtualization to present the illusion of many smaller virtual machines (VMs), each running a separate operating system instance. Successful partitioning of a machine to support the concurrent execution of multiple operating systems poses several challenges. Firstly, virtual machines must be isolated from one another: it is not acceptable for the execution of one to advers

Re:I bet it's not Open Source...

[sjwt (161428)]

If you had botherd to open the Xen link,
and read the page which was a whopping
8100 bytes in size you would of noticed
"Xen is Open Source software, released under the terms of the GNU General Public License. We have a fully functional port of Linux 2.4 running over Xen, and regularly use it for running demanding applications like MySQL, Apache and PostgreSQL. Any Linux distribution should run unmodified over the ported OS."

rather then just trolling off the two words "Microsoft Research"

Re:How is it ideal for secure virtual hosting?

[Zigg (64962)]

Give each customer their own machine that they can do pretty much what they please with, root rights and all. No customer gets rights over the other customers' VMs, though.

Re:How is it ideal for secure virtual hosting?

[__past__ (542467)]

Virtualization rocks for that kind of things. I happily use FreeBSD jails [freebsd.org] for that. They work on a different level and have different goals, but giving each (group of) user(s) their own sandbox to play with is definitly a really cool thing, both for admins and for the users themselves.

Re:Very cool indeed

[Rob Riggs (6418)]

Very cool, but why is Microsoft involved?

I would guess it is because Microsoft has a modest edge over Linux International when it comes to funding research grants.