[c-nsp] 7304 NPE-G100 vs. NSE-100

[Mikael Abrahamsson]

On Mon, 12 Dec 2005, Cory Ayers wrote:

> Does anyone have statistics on a 7304/NSE-100?  The data sheet states it
> can push 4Gbps bidirectional throughput and 3.5Mpps using PXF
> architecture.  I would like to know how much growth the NSE-100 will
> really give and if there are any caveats to running the NSE-100 vs the
> NPE-G100.  I have looked through the PXF information and believe most of
> our traffic will be PXF switched.
> http://www.cisco.com/en/US/customer/products/hw/routers/ps352/prod_maint
> enance_guide09186a008057410a.html

Our experience with the NSE-100 over the past two years have been varied. 
Performance is up to spec, but there have been quite a few PXF related 
problems.

But generally I'd say that the NSE-100 has at least a 5:1 performance 
advantage over the G100 in real life. The PXF performance is also pps 
without much (if any) mbps limitation, so with imix you'll see in excess 
of 4 gigabit/s, and it seems very little affected by how many features we 
turn on. Doing almost a gig with shaping/ratelimiting, we generally don't 
see PXF usage above 10%.

The G100 is generally the same as your 7200 NPE-G1 with the same 
performance limitations, which I think is very limited in todays world.

My gut feeling says that the NSE-100 could very well work as a 2*STM16 
2*GE router taking transit traffic thru it and dropping it on GE, it 
wouldn't handle a fullblown DoS attack with small packets, but it would be 
close.

-- 
Mikael Abrahamsson    email: swmike at swm.pp.se