[c-nsp] downlink bgp interconnect best practices

[Mark Tinka]

On Tuesday, May 31, 2011 08:57:41 PM Gert Doering wrote:

> Well, it depends a bit how the connectivity between CR1
> and CR2 is built.  If you have two independent switches
> there, the direct link is not strictly needed.  Still,
> it has the advantage that if these switches should fail,
> CR1 and CR2 always are connected - otherwise you'll run
> into some weird problems.

We battled with this when doing our initial PoP design 
several years ago.

Since we had our core routers multi-homed to 4x or 2x core 
switches in each major PoP, we found it pretty useless to 
have a direct link between the core routers in the same PoP. 
Reasoning was simple - if both switches died, they'd go with 
all the edge routers anyway, so connectivity within the core 
would be the least of our problems. Also, if connectivity 
between your core routers and core switches were on the 
order of N x 10Gbps circuits, trying to replicate the same 
connectivity between core routers would expensive, and 
sometimes pointless, or just lead to congestion in the worst 
of cases (which is better than nothing). Operationally, it 
has worked out too.

But again, YMMV, and it all depends on your own network and 
budgetary circumstances.

PS: you certainly need direct links between your core
    routers if they're geographically placed.

Cheers,

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/20110531/fda6cc80/attachment.pgp>