From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Fri Nov 22 2002 - 20:04:45 GMT-3
At 3:40 PM -0600 11/22/02, Treptow, Georg wrote:
>Hello everybody, I am back and need your help.
>
>Does anybody know what is the maximum number of BGP iBGP peering connections
>a router can handle is?
>The router is a 7206VXR NPE400 with a Hssi Card. I can not find anything on
>the web.
>
>Your help would be greatly appreciated.
>
>Thank you,
>
>Georg Treptow, CCIE 9720
>Sr. Engineer - Qwest Solutions
>gxtrept@qwest.com <mailto:gxtrept@qwest.com>
Unfortunately, there really is no simple answer to this, even if you
know the absolute number of peers. Cisco, historically, has
recommended a maximum of 20-30 peerings before going to iBGP peering
reduction methods, but this certainly can vary. It is going to depend
on a number of factors, both external and internal to the BGP
implementation. If you are doing this in an ISP environment, I
strongly recommend benchmarking.
External factors, even on an iBGP router, include the amount of local
policy, the number of peers, the rate of change of updates on the
peers, etc.
Internal factors include the order in which the updates are sent,
which may or may not be optimal even from the same router vendor.
I'm involved with a effort on this in the IETF Benchmarking Working
Group (BMWG), along with people from Cisco, Juniper, NextHop, and
Nortel. We started with basic eBGP (i.e., no filtering or other
policy rules), and have the terminology document
(http://www.ietf.org/internet-drafts/draft-ietf-bmwg-conterm-03.txt)
about to go to RFC once we made a few editorial changes. There is a
methodology document that technically expired, mostly due to about
half the authors being laid off. We really should get it finished, as
there really wasn't left to do.
iBGP was going to be our next project. iBGP is more complex as a test
setup than basic eBGP. We then planned to go back to eBGP but
consider filtering.
This archive was generated by hypermail 2.1.4 : Tue Dec 03 2002 - 07:23:09 GMT-3