|
Hello,
The report is at: http://www.pdos.lcs.mit.edu/papers/click:tocs00/
and the homepage of the click modular router project is at:
http://www.pdos.lcs.mit.edu/click/
This is where I took my figures from. I think, before reading the
interesting links you've just mentionned, that the main difference lies
in the packet size used for the benchmarks. At the Click project, they
focused on routing capacity (using 64-byte packets) and not bandwidth.
It might also be that their test configuration was simpler.
Still with these small packets, they managed to route up to 435Kpps on
there test configuration with a 700Mhz PIII CPU. Like the 'fast
forwarding bird', they had to modify the hardware driver : they changed
it from an interrupt-driven mode to a polling mode. The network card
used was a DEC 21140 NIC.
Regards,
Jacques THOMAS.
Roberto Nibali wrote:
>
> Hello,
>
> > Anyway, benchmarks conducted for the Click Router Project at MIT show
> > that a regular PIII 700Mhz Linux router (no customization in drivers or
> > whatsoever) can handle up to 84.000 paquets per second. Then, 250 hits
> > per second makes only 0.3 % of it. LVS code should not (IMHO) add much
> > to it.
>
> Could you please provide information or reports on that statement?. I've
> been following the 'fast forwarding bird' development by Jamal Hadi
> Salim, Robert Olsson and Alexey Kuznetsov. If you read their paper from
> the 5th ALS conference from November last year you can read that the
> Linux' congestion collapse is at around 60Kpps. This was not even a max
> but the packets wouldn't even leave the system anymore for
> understandable reasons (softnet is not designed for such packet rates).
> The MLFFR (Maximum Loss Free Forwarding Rate) was around 27Kpps with the
> CPU being used 100% to process networking (from their document).
> Problems mentioned where the shared backlog queue for multiple NIC's and
> not optimal hardware flow control algorithms and interrupt livelocks in
> parallel packet reordering.
>
> After they implemented their fast forwarding scheduling their were able
> to sustain 80Kpps, while input rate was 148Kpps.
>
> Using an e1000 they achieved 360Kpps using a single PIII 933 MHz CPU.
>
> You can check out the next three links if you're interested:
> http://robur.slu.se/Linux/net-development/jamal/FF-html/
> http://lwn.net/2000/0928/a/fast-forwarding.php3
> http://www.globecom.net/ietf/draft/draft-almesberger-wajhak-diffserv-linux-01.html
>
> Best regards,
> Roberto Nibali, ratz
>
> _______________________________________________
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to http://www.in-addr.de/mailman/listinfo/lvs-users
|
