[Discuss] PS: Most virtuous virtual boxen?

brendan brendan at ming.org
Mon Feb 9 13:35:32 PST 2015


We run a huge number of xen instances at work. Performance is great, although we've ended up compiling our own kernels and not using the stock kernels for it.

If you are interested, you could swing by my home office sometime and I could show you how things are set up and how it works.

We might be giving a presentation about the setup at LISA next year. I'd mentioned I could give a presentation about it sometime to the group, though I've been lazy about actually doing so. The upcoming Lisa thing might provide me with some motivation to actually do so.

> On Feb 9, 2015, at 2:00 AM, Patrick <Nixniq-sneaking at sneakEmail.com> wrote:
> 
> Sorry, just realized I'd asked about this almost exactly a year
> ago...
>    http://lists.vlug.org/pipermail/discuss/2014-January/000082.html
> 
> ...And got some really good answers.  Thanks for those.
> 
> If there have been any noteworthy developments since then, I'd
> still like to hear about them.  I'd be especially interested in
> comparisons between Xen and KVM -- supported OSes, performance
> degradation [and hopefully, the lack thereof] in guest systems,
> etc;
>    http://en.wikipedia.org/wiki/Kernel-based_Virtual_Machine
>    http://en.wikipedia.org/wiki/Xen
> 
> Currently putting together an eight-core machine with 12GB
> memory, so multitasking will certainly be an option.  8-)
> 
> 
> Thanks... again,
> Patrick.
> 
> 
> On Sun, 8 Feb 2015 21:57:19 -0800
> I wrote:
> 
>> Anyone here have any experience with Xen?
>>    http://en.wikipedia.org/wiki/Xen
>> 
>> If so, what do you think of it?  If not, or if your opinion is
>> not favourable, what sort of virtual boxen would you recommend,
>> and why?
>> 
>> I kind of like the *idea* of Xen, because it runs as a kernel in
>> its own right, portioning out hardware access to various guest
>> OSes -- instead of as a user program which runs *inside* a host
>> OS, which in turn builds virtual machines within itself for
>> various guest OSes to live in.  Cuts out a few middlemen, that
>> way, abstracts the hardware a little less.  It also makes the
>> overall system less like a Russian doll and/or onion, avoiding
>> layer upon layer of containers holding containers.
>> 
>> But that's only in theory.  What about in practise?
>> 
>> Also, a little off-topic, but not entirely because it might
>> occupy a virtual box, is it true that Windows XP can only
>> utilize one CPU, no matter how many cores it's given to play
>> with?
>> 
>> 
>> Thanks,
>> Patrick.
> 
> -- 
> Barker's Proof:
>    Proofreading is more effective after publication.
> _______________________________________________
> Discuss mailing list
> Discuss at vlug.org
> http://lists.vlug.org/mailman/listinfo/discuss



More information about the Discuss mailing list