[c-nsp] Cisco vs. Juniper

[Richard A Steenbergen]

On Mon, Nov 02, 2009 at 04:52:54PM +0100, sthaug at nethelp.no wrote:
> > In the past I have compared M7i with ASR1k
> 
> The M7i is getting a bit long in the tooth, so a better comparison
> might be ASR1k vs MX80. One important difference if you need a box
> *now* is of course that MX80 has been announced but I haven't seen it
> in the price lists yet.

They're actually coming out with (or may already be shipping, I don't
follow these boxes that closely) a replacement CFEB for M7i/M10i which
uses the I-Chip (the same fwding hw as M120 and the current generation
of MX). This should give it a slightly longer shelf life, as it will add
a bunch of modern features and some additional fib capacity that didn't
exist in the old hardware. Still though, this is a very old box (it came
out in 2003, as a lower production cost refresh on the M5/M10 which came
out in 2000). The CFEB won't fix the very limited capacity, so it
wouldn't be a fair comparison against a modern box. MX80 would indeed be
a much closer comparison, though the feature set is still pretty
different.

> Here I'd have to disagree. Sampled netflow works very well without a
> services PIC. If you don't do sampling the situation is different.

IIRC the default limit for sampled netflow (at least on M7i generation
platforms, I can't speak to MX80 or the like) was 7000pps per FPC. So if
for example you sampled every 1:1024 packets this would be good for
7.1Mpps of analyzed traffic for FPC (i.e. more than the box will ever be
able to forward).

-- 
Richard A Steenbergen <ras at e-gerbil.net>       http://www.e-gerbil.net/ras
GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)