[c-nsp] RE: Routing performance database

[Hank Nussbacher]

Unfortunately, until someone comes up with a Linpack/Whetstone/whatever
style benchmark for routers, we will end up comparing apples to oranges.

-Hank

On Sat, 2 Oct 2004, Michael Markstaller wrote:

> This is really a great idea ! I'd be happy to supply results, I have an
> excel-sheet with results from approx. 10 different lower-end platforms..
> (so not blaming anybodys 6509 ;)
>
> But I also see some problem wether there are legal issues ? (it's not
> only that we are fed with marketing-papers but also not allowed to talk
> about the real-world..)
>
> I also don't why it shouldn't make sense to build up such a database due
> to differences in environments and/or platforms as long as it can be
> documented which feature were used (i.e.
> IOS-version/Switching-path/NAT/CBAC/IPSec/COMP-LZS/GRE/L2TP etc.) and
> how it was tested (streams,packet-size etc.), at least it will be much
> more realistic than the pps and megabits of IMIX-3DES which is very far
> away from reality..
> An episode why I think this would be important to have such a database
> for anybody: I had a month-taking case/disussion with Cisco some time
> ago, that a 2621 was unable to deal with > 1Mbit IPSec-traffic under
> load and I wished something like that was available.. they told me that
> the routers are simply to slow for the job of dealing 1 MBit IPSec with
> CBAC instead of confirming that the packets shouldn't get
> process-switched.
> the problems disappeard after some 12.2(x)T improvement in CBAC
> magically without Cisco ever confirming me, that it the problem was
> CBAC-pkts getting process-switched, in my case only with IPSec and
> HW-crypto (which was the real reason for the slow performance)..
> some weeks later this appeared:
> http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122newft
> /122t/122t8/ftfirewl.htm
> and I had no more performance-troubles since then, the 2621 push through
> 2Mbit with ease, but it took me weeks to find out..
>
> when purchasing IOS-boxes it'd be helpful to see wether at least
> 1/6-1/10 of the in several whitepapers specified IMIX 3DES-performance
> is somewhat realistic with real-world features..
> BTW: the miercom-reports for the new x8xx platforms go the right
> direction but the most interesting details are unfortunately missing..
>
> Michael
>
> > -----Original Message-----
> > From: cisco-nsp-bounces at puck.nether.net
> > [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Sam Stickland
> > Sent: Thursday, September 30, 2004 1:33 PM
> > To: cisco-nsp at puck.nether.net
> > Subject: Routing performance database (Re: [c-nsp] 10 Mb
> > internet traffic ona 3640?)
> >
> >
> >
> > There's been a lot of questions lately (as always!) about the
> > performance
> > of various routers. Would there be any legal issues about
> > with setting an
> > a site where people can list their real world experiences with the
> > performance of various pieces of kit?
> >
> > I guess it wouldn't really be any different from a hardware
> > review, but
> > there's often clauses in licenses that prevent this sort of stuff.
> >
> > Anybody know? I don't mind setting it up.
> >
> > Sam
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>