Xen is a virtual machine monitor for x86 that supports execution of multiple guest operating systems with unprecedented levels of performance and resource isolation. A port of Linux 2.4 to run on the VMM works well, and *BSD and XP ports are in progress. Version 1.0 was released a few months ago.
Get to Know Xen
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
This is fairly old news. Has anything happened that would show this as a more viable platform.
Ok, so in plain English, what does Xen actually do?
Indeed old news. I do not see any development progress. Though i couldn’t find out so fast when 1.1 was released.
The homepage is:
http://xen.sf.net
Which links to:
http://www.cl.cam.ac.uk/Research/SRG/netos/xen/index.html
On the homepage an interesting (non 3rd party) benchmark:
http://www.cl.cam.ac.uk/Research/SRG/netos/xen/performance.html
which clearly points out it is _faster_ than UML. A demo CD can be downloaded from the homepage.
Another interesting thing is the note on the bottom: “Work on Xen is supported by UK EPSRC grant GR/S01894, Intel Research, and Microsoft Research”
Woah thanks, that’s imo the coolest thing MS has done in 2003.
I know it is old news. But I just heard about it at a conference yesterday and I couldn’t find any references on OSNews.
I think it is too cool a project to be ignored. I really couldn’t feel any diffenrence between plain Linux and Xen Linux… Ofcause it needs some more work.
You’re right, it is a cool project and it shouldn’t be ignored. I’ve been keeping my eye on this one for quite some time and though the sourceforge page shows activity, there is nothing else to indicate things continue to move.
Last year about half way through, they made the great anouncement about the project, and that they were close to porting to MS. It doesn’t look like anything has changed.
I did play with the project when it was first announce . . . it works but the tools aren’t there yet. Stll too much has to be done manually IMHO. At this time there are a number of projects that accomplish the same thing. Yes, UML performance isn’t so great but this project http://www.linux-vserver.org/ is very stable, gives good performance and is in steady development.
As to the question of what is it by salmacis: It’s for virtual servers, the holy grail for hosting companies and development companies (my company). It allows you to build perhaps 50 “virtual” servers on one physical server machine, keeping them all truly seperate and working alone. Less money spent on hardware = :-).
As to the question of what is it by salmacis: It’s for virtual servers, the holy grail for hosting companies and development companies (my company). It allows you to build perhaps 50 “virtual” servers on one physical server machine
What do you mean by ‘virtual servers’ … like running multiple operating systems at the same time?
>What do you mean by ‘virtual servers’ … like running
>multiple operating systems at the same time?
Yes, that’s what he means, since this is the whole purpose of Xen. “virtual servers” is a good practical use of Hardware Virtualizers, when you can’t (or don’t want to) afford the cost of real hardware, but still have to provide the whole system isolation layer.
IBM z/OS is a fine illustration of this practice (remember the IBM ads where the management guy asks where the whole servers hardware disappeared, and the tech guy shows him the single system replacing the whole set ?). You only have one real server, but several (they advertised thousand) OS running on it, providing full isolation to their customers.
you could learn a lot on this by having a look at the provided links.
I love virtual server technology for development and testing – but that IBM ad always bothered me. Being an IT manager the thought of a single point of failure for so MANY systems keeps me up at night…
Ahh, I knew there was a down side. 🙂
Actually, the IBM has covered A LOT of ground in that area. It’s been a while since I’ve looked at their solutions, but I was really impressed with the built in redundency.
Yes, no matter how far you go with it though, at some point it does come down to a single point of failure, it is all in one machine after all.
It comes down to risk vs benifit I believe. I think IBM has done much to increase the benifit and lower risk, but these other (low or no cost) solutions, come much closer to the single point of failure issue, IMHO.
Does this have to be limited to servers though? I mean, could I use it to switch between Windows XP and Linux desktop, or even two instances of Windows XP like I had two different computers on a switch box ?
Xen certainly took the best approach they could performance wise, but don’t seem to have fostered enough of a community to get a diverse selection of operating systems ported to run on top of Xen.
At this point, the two available operating systems, Windows XP and Linux 2.4, do not appeal to me in the slightest. Perhaps after we see a Linux 2.6 or FreeBSD 5.x port I’ll be interested…
With regard to the funding and provision of resources by Microsoft research, I could probably shed some light on this by saying that the head of MS Research UK is an ex Cambridge University Computer Lab member and works just across the way in the sister building to our Lab.
I gather that he gave the go ahead to release source code for the NT5 kernel (in order to port it to the Xen platform) without the knowledge of MS Redmond.
Incidentally, one of my friends is a very talented hacker with intimate knowledge of parts of the Linux kernel and was distinctly horrified by some of the nasty hacks he saw and had to work round in the NT code. The NT kernel gets away with it because it never has to run on anything but IA32 and the Alpha (it still has the ifdefs in there!). I’m told the XP part of the Xen system is very much a work in progress because it is incredibly difficult to eliminate the bugs with such spaghetti code, while the Linux modifications were (comparitively) simple.