From: Swan, Jay (jswan@sugf.com)
Date: Thu Aug 23 2007 - 13:32:00 ART
Not currently an ISP guy here, but... you're on the right track. OSPF or
IS-IS are used for next-hop reachability. Obviously you can't
redistribute BGP into the IGP in a service provider environment
(although some have done so by accident and melted down their networks
as the routers started crashing). So either you need to run iBGP
everywhere, or run BGP at the edge/aggregation layers and use an MPLS
core to label-switch your traffic between entrance and exit points. The
MPLS core is appealing because your core doesn't need to know anything
about BGP, it scales easily to IPv6, and you can build lots of other
money-making MPLS applications on it.
On small customer-facing edge routers you can avoid BGP by using static
and/or default routing.
From what I have seen small/midsize ISPs run OSPF, and large/very large
SPs lean toward IS-IS, but there are some really big OSPF networks out
there too.
The Cisco Press BGP and MPLS design books have info on this. Also check
out the NANOG tutorials and Networkers presentations.
Jay (#17783)
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Gregory Gombas
Sent: Thursday, August 23, 2007 9:37 AM
To: Cisco certification
Subject: Question for you ISP guys
For those of you with ISP experience, can you tell me what routing
protocols do service providers typicall run within their AS?
Do you have every single router running BGP? I can't imagine
redistributing 225k+
routes into an IGP, so how do you pass these routes withing your AS?
If you are using iBGP what are you using to transmit next hop
information (as iBGP does not normally update the next hop of the
external AS)?
Can someone point me to some documentation showing typical ISP routing
design?
Thanks,
Greg
This archive was generated by hypermail 2.1.4 : Sat Sep 01 2007 - 11:32:12 ART