Re: Xen Socket API

Re: Xen Socket API

[Mark Williamson]

> Is there any code available for this or it is still a paper-concept?

I believe XenSockets and another Xen-oriented network acceleration technology 
have been coded, but I don't know of any releases of the code as-yet.  
Something like this is bound to be made available at some stage, but I don't 
know if the original authors want to release their code.

Cheers,
Mark

-- 
Dave: Just a question. What use is a unicyle with no seat?  And no pedals!
Mark: To answer a question with a question: What use is a skateboard?
Dave: Skateboards have wheels.
Mark: My wheel has a wheel!

Re: Xen Socket API

[Daniel P. Berrange]

On Wed, Aug 29, 2007 at 03:36:14PM +0100, Mark Williamson wrote:
> > Is there any code available for this or it is still a paper-concept?
> 
> I believe XenSockets and another Xen-oriented network acceleration technology 
> have been coded, but I don't know of any releases of the code as-yet.  
> Something like this is bound to be made available at some stage, but I don't 
> know if the original authors want to release their code.

The original code was very recently made available on sourceforge under GPL:

   http://sourceforge.net/projects/xvmsocket/

Medium-to-long term though I don't see a Xen specific socket impl as really
being a viable model. A generic solution based on the new Virt-IO paravirt
infrastructure will be more appealing to app developers because it'll be 
possible to make use of it across all the different impls of paravirt_ops 
in Linux (ie Xen, KVM, VMWare, LGuest). 

Dan.
-- 
|=- Red Hat, Engineering, Emerging Technologies, Boston.  +1 978 392 2496 -=|
|=-           Perl modules: http://search.cpan.org/~danberr/              -=|
|=-               Projects: http://freshmeat.net/~danielpb/               -=|
|=-  GnuPG: 7D3B9505   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505  -=| 

Re: Xen Socket API

[Mark Williamson]

> On Wed, Aug 29, 2007 at 03:36:14PM +0100, Mark Williamson wrote:
> > > Is there any code available for this or it is still a paper-concept?
> >
> > I believe XenSockets and another Xen-oriented network acceleration
> > technology have been coded, but I don't know of any releases of the code
> > as-yet. Something like this is bound to be made available at some stage,
> > but I don't know if the original authors want to release their code.
>
> The original code was very recently made available on sourceforge under
> GPL:
>
>    http://sourceforge.net/projects/xvmsocket/
>
> Medium-to-long term though I don't see a Xen specific socket impl as really
> being a viable model.

I think it would have very little chance of being accepted upstream.  The xway 
TCP-over-Xen implementation would have even less chance of being accepted *as 
is* due to making modifications to the Linux network stack.

> A generic solution based on the new Virt-IO paravirt 
> infrastructure will be more appealing to app developers because it'll be
> possible to make use of it across all the different impls of paravirt_ops
> in Linux (ie Xen, KVM, VMWare, LGuest).

Yep.  If a generic solution using something like virtio is available *and* 
there are multiple hypervisors which can leverage it then hopefully it would 
have a decent chance of being included.

Cheers,
Mark

-- 
Dave: Just a question. What use is a unicyle with no seat?  And no pedals!
Mark: To answer a question with a question: What use is a skateboard?
Dave: Skateboards have wheels.
Mark: My wheel has a wheel!

RE: Xen Socket API

[Kaushik Barde]

Dan/Mark:

Thank you, we are creating a proto-type project that may leverage some
of the work.
We will be more than happy to share our results with you and the
community.

Thanks.
-Kaushik

-----Original Message-----
From: Daniel P. Berrange [mailto:berrange@redhat.com] 
Sent: Wednesday, August 29, 2007 7:55 AM
To: Mark Williamson
Cc: xen-devel@lists.xensource.com; Kaushik Barde
Subject: Re: [Xen-devel] Xen Socket API

On Wed, Aug 29, 2007 at 03:36:14PM +0100, Mark Williamson wrote:
> > Is there any code available for this or it is still a paper-concept?
> 
> I believe XenSockets and another Xen-oriented network acceleration
technology 
> have been coded, but I don't know of any releases of the code as-yet.

> Something like this is bound to be made available at some stage, but I
don't 
> know if the original authors want to release their code.

The original code was very recently made available on sourceforge under
GPL:

   http://sourceforge.net/projects/xvmsocket/

Medium-to-long term though I don't see a Xen specific socket impl as
really
being a viable model. A generic solution based on the new Virt-IO
paravirt
infrastructure will be more appealing to app developers because it'll be

possible to make use of it across all the different impls of
paravirt_ops 
in Linux (ie Xen, KVM, VMWare, LGuest). 

Dan.
-- 
|=- Red Hat, Engineering, Emerging Technologies, Boston.  +1 978 392
2496 -=|
|=-           Perl modules: http://search.cpan.org/~danberr/
-=|
|=-               Projects: http://freshmeat.net/~danielpb/
-=|
|=-  GnuPG: 7D3B9505   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505
-=|