Re: [Xen-devel] Testing Xen for CERN

[Ian Pratt <Ian.Pratt@xxxxxxxxxxxx>]

> Testing Xen for CERN
> 
> Hi, I am currently testing Xen and other virtual monitors for CERN and
> have some ?abstract? questions. We might want to use it in the
> Grid project if the benchmarking tests give good results.
> 
> 1. Will there by any port to IA64 or other platforms in the
> future?

x86 and x86_64 are the core architectures, but IA64 is in
progress as a background activity by one developer. I'd like to
see a PPC port at some point. Virtualization of RISC
architectures is generally much easier than x86...

> 2. When will Xen upgrade to 2.6 ?

Very soon. 

Now that common distros are beginning to upgrade to 2.6 we'll
move over. XenLinux 2.6 already booted some time ago, but hasn't
kept pace with other Xen changes.

> 3. Does Xen support dual processors?

Xen supports multiprocessor machines (including hyperthreading),
but individual guest operating systems are currently uniprocessor
(i.e. you have to run multiple guests to make use of a
multiprocessor box). We intend to add SMP guest support in due
course.

> 4. Benchmarking a virtual monitor can give some strange results. Which
> method do you use for Xen?

Xen has been independently benchmarked by a team at Clarkson:
 http://www.clarkson.edu/class/cs644/xen/files/repeatedxen-usenix04.pdf

There's also our SOSP paper.
 http://www.cl.cam.ac.uk/netos/papers/2003-xensosp.pdf

Xen's new IO architecture will have changed some of the numbers,
but the overall result will be similar.
 
> 5. How would you say Xen has an advantage over other virtual monitors
> like
> vmware, plex86, uml etc.

Xen's big plus points are:
 * Performance. Xen is very close to native performance.
 * "live migration" of VMs between nodes in a cluster
 * Released under GPL

The negatives relative to some of the other VMMs are:
 1 requires guest OS to be ported
 2 poor documentation 
 3 weak support for CoW sparse virtual disks

Point 1 you'll have to live with. We hope the community will fix
2. A solution to point 3 already exists but needs to be
integrated.

Xen has a good track record for stability. We're hoping to do an
Alpha testing release of "Xen 2.0" in the next week, once the new
control tools have settled down.

Ian







-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/xen-devel