* [9fans] wirespeed processing @ 10 Gb/s
@ 2011-06-28 20:55 Bakul Shah
2011-06-28 22:22 ` erik quanstrom
2011-06-29 0:29 ` ron minnich
0 siblings, 2 replies; 6+ messages in thread
From: Bakul Shah @ 2011-06-28 20:55 UTC (permalink / raw)
To: Fans of the OS Plan 9 from Bell Labs
Something like this would make a nice hack!
http://info.iet.unipi.it/~luigi/netmap/sigcomm-poster.pdf
Our framework, called netmap, maps packet buffers into the
process' memory space ...
One core at 1.33 GHz can send or receive packets at wire speed
on 10 Gbit/s links (14.88 Mpps), with very good scalability in
the number of cores and clock speed.
netmap is available on FreeBSD.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [9fans] wirespeed processing @ 10 Gb/s
2011-06-28 20:55 [9fans] wirespeed processing @ 10 Gb/s Bakul Shah
@ 2011-06-28 22:22 ` erik quanstrom
2011-06-28 22:41 ` Bakul Shah
2011-06-29 0:29 ` ron minnich
1 sibling, 1 reply; 6+ messages in thread
From: erik quanstrom @ 2011-06-28 22:22 UTC (permalink / raw)
To: 9fans
On Tue Jun 28 16:56:34 EDT 2011, bakul@bitblocks.com wrote:
> Something like this would make a nice hack!
>
> http://info.iet.unipi.it/~luigi/netmap/sigcomm-poster.pdf
>
> Our framework, called netmap, maps packet buffers into the
> process' memory space ...
>
> One core at 1.33 GHz can send or receive packets at wire speed
> on 10 Gbit/s links (14.88 Mpps), with very good scalability in
> the number of cores and clock speed.
>
> netmap is available on FreeBSD.
plan 9 already does wirespeed 2x10gbe "processing", if by
processing you mean disk i/o. :-)
- erik
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [9fans] wirespeed processing @ 10 Gb/s
2011-06-28 22:22 ` erik quanstrom
@ 2011-06-28 22:41 ` Bakul Shah
2011-06-28 22:59 ` erik quanstrom
0 siblings, 1 reply; 6+ messages in thread
From: Bakul Shah @ 2011-06-28 22:41 UTC (permalink / raw)
To: Fans of the OS Plan 9 from Bell Labs
On Tue, 28 Jun 2011 18:22:51 EDT erik quanstrom <quanstro@quanstro.net> wrote:
> On Tue Jun 28 16:56:34 EDT 2011, bakul@bitblocks.com wrote:
> > Something like this would make a nice hack!
> >
> > http://info.iet.unipi.it/~luigi/netmap/sigcomm-poster.pdf
> >
> > Our framework, called netmap, maps packet buffers into the
> > process' memory space ...
> >
> > One core at 1.33 GHz can send or receive packets at wire speed
> > on 10 Gbit/s links (14.88 Mpps), with very good scalability in
> > the number of cores and clock speed.
> >
> > netmap is available on FreeBSD.
>
> plan 9 already does wirespeed 2x10gbe "processing", if by
> processing you mean disk i/o. :-)
By wirespeed the author means processing 14.48M pkts/sec, not
814K pps! AOE is a special case.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [9fans] wirespeed processing @ 10 Gb/s
2011-06-28 22:41 ` Bakul Shah
@ 2011-06-28 22:59 ` erik quanstrom
2011-06-28 23:08 ` Bakul Shah
0 siblings, 1 reply; 6+ messages in thread
From: erik quanstrom @ 2011-06-28 22:59 UTC (permalink / raw)
To: 9fans
> > > netmap is available on FreeBSD.
> >
> > plan 9 already does wirespeed 2x10gbe "processing", if by
> > processing you mean disk i/o. :-)
>
> By wirespeed the author means processing 14.48M pkts/sec, not
> 814K pps! AOE is a special case.
no. you don't get to redefine the term just because aoe does
things well.
wirespeed is not a packet rate, it's a bitrate.
- erik
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [9fans] wirespeed processing @ 10 Gb/s
2011-06-28 22:59 ` erik quanstrom
@ 2011-06-28 23:08 ` Bakul Shah
0 siblings, 0 replies; 6+ messages in thread
From: Bakul Shah @ 2011-06-28 23:08 UTC (permalink / raw)
To: Fans of the OS Plan 9 from Bell Labs
On Tue, 28 Jun 2011 18:59:52 EDT erik quanstrom <quanstro@quanstro.net> wrote:
> > > > netmap is available on FreeBSD.
> > >
> > > plan 9 already does wirespeed 2x10gbe "processing", if by
> > > processing you mean disk i/o. :-)
> >
> > By wirespeed the author means processing 14.48M pkts/sec, not
> > 814K pps! AOE is a special case.
s/wirespeed/wirespeed processing/
>
> no. you don't get to redefine the term just because aoe does
> things well.
>
> wirespeed is not a packet rate, it's a bitrate.
"wirespeed processing" is being able to handle any pkt size
for a given wire bit rate so the worst case is for the
smallest size pkts.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [9fans] wirespeed processing @ 10 Gb/s
2011-06-28 20:55 [9fans] wirespeed processing @ 10 Gb/s Bakul Shah
2011-06-28 22:22 ` erik quanstrom
@ 2011-06-29 0:29 ` ron minnich
1 sibling, 0 replies; 6+ messages in thread
From: ron minnich @ 2011-06-29 0:29 UTC (permalink / raw)
To: Fans of the OS Plan 9 from Bell Labs
hmm. 10 Gbits/s. I should read the paper, but the thought occurs that
the last time we talked about wire speed of 10 Gb/s in my world it was
2002.
ron
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2011-06-29 0:29 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-06-28 20:55 [9fans] wirespeed processing @ 10 Gb/s Bakul Shah
2011-06-28 22:22 ` erik quanstrom
2011-06-28 22:41 ` Bakul Shah
2011-06-28 22:59 ` erik quanstrom
2011-06-28 23:08 ` Bakul Shah
2011-06-29 0:29 ` ron minnich
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).