[c-nsp] NPE-G1 PPS limitations

[dpinkard at AccessLine.com]

Would one expect gigE to gigE traffic local to the NPE-G1 to scale higher
than that?

-----Original Message-----
From: Ross Vandegrift [mailto:ross at kallisti.us] 
Sent: Thursday, April 17, 2008 11:27 AM
To: dpinkard at AccessLine.com
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] NPE-G1 PPS limitations

On Wed, Apr 16, 2008 at 01:51:48PM -0700, dpinkard at AccessLine.com wrote:
> I'm stuck with older hardware (VXRs, and GSRs) and trying to plan for its
> eventual demise... 
> 
> At what point should one expect the G1 to actually top out on PPS? I've
> heard anecdotal claims of 200-225Mbps with voip-sized (200byte) packets
for
> something around 125kpps, which is dramatically lower than Cisco's
> optimistically marketed number of 1Mpps. 
> 
> So assuming a 20 line ACL, how should one expect it to scale for traffic
> going to/from ports on the gigE interfaces? Should this performance drop
> when traffic is headed to the PCI buses?

7200 CPU starts to get pretty beat up and you climb into those numbers, but
it gets far worse if you're using PA-GE modules.  The PA-GE interfaces
top out at around 400Mbps before they start dropping packets, even if
the pps is low.  Your pps numbers are very close to what I recall,
though we were hitting the bps issue more.

The gig interfaces on the NPE are substantially beefier and scaled
beyond this limit, but I doubt that you'll get as much more out of
them for pps issues.


-- 
Ross Vandegrift
ross at kallisti.us

"The good Christian should beware of mathematicians, and all those who
make empty prophecies. The danger already exists that the mathematicians
have made a covenant with the devil to darken the spirit and to confine
man in the bonds of Hell."
	--St. Augustine, De Genesi ad Litteram, Book II, xviii, 37