[c-nsp] Advice - c7200VXR with 2 bgp tables and peering fabric

[Mark Tinka]

On Thursday, June 06, 2013 06:02:03 PM Eric A Louie wrote:

> Erik, I noticed that typo after I sent the message away.
> 
> Thank you for the bandwidth warning - I'm only about
> halfway there.  From the  old information at
> http://www.cisco.com/en/US/prod/collateral/routers/ps341
> /prod_qas0900aecd80471791.html it appears the NPE-G2
> doubles the packet forwarding rate and the RAM
> capacity.  I'm not adverse to upgrading the 7206VXR with
> the NPE-G2

As I've mentioned a few times on this list in the past, the 
NPE-G2 or 7201 is good for about 950Mbps of traffic when 
operated in "core router" mode, i.e., no or limited ACL use, 
basic IP/MPLS features, basic or no QoS, e.t.c.

At those traffic levels, the CPU was sitting at about 90% 
utilization, but began dropping packets if there were 
microbursts or if CPU started to hit around 93% utilization.

In terms of BGP performance, I have 2x full feeds running on 
an NPE-G1 now. It can get slow especially if policy changes 
are taking place, i.e., the BGP Router process. I'm testing 
RPKI on these as well, which can slow things down a little 
further until the entire table has been walked.

The NPE-G2 is much better with full feeds, and you can stick 
about three or four full feeds there before you start 
shaking things up. But I'd be cautious about running 
anything larger than two full feeds either on the NPE-G1 or 
NPE-G2, as even though you might have lots of RAM, your main 
issue will still be CPU performance during route churn.

I'm on IOS 15.2(4)S3 code across these units.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20130816/541f07d0/attachment-0001.sig>