[c-nsp] bgp in the "core"

[Matt Buford]

matthew zeier" <mrz at intelenet.net> wrote:
> I'm looking for examples of data center-like networks and whether or not 
> they
> run ibgp in the "core" or just on their transit routers.
[...]
> The "core" provides L3 connectivity for customer networks and each transit
> router is connected to each "core" router/switch.  The access switches are
> plain L2 switches.  Core and transit run OSPF.
>
> The problem I have is that the transit routers are the boxes initiating my 
> BGP
> routes and I contend that if they become disconnected from the "core", 
> having
> them continue to announce routes is a Bad Thing.
>
> I believe that the core should do the route origination.
>
> Am I off my rocker?

Nope, you are right on.  The datacenters I manage are built out in a similar 
design.

The core should run iBGP and originate the routes.  Also, I put a full BGP 
table in the core so that it can  make the proper decision and forward the 
packet directly to the right border router.  I don't want to see core1 -> 
border1 -> border2 -> transit, or something like that.  I want to see only 2 
hops in my datacenter (such as coreX -> borderX -> transit) for all paths.

6500s with at least sup2/msfc2 (sup720 recommended) work great for the 
"core" role here, able to handle the full BGP table and everything else 
necessary.