[c-nsp] 2 full BGP feeds on 3750?...expanded

[Mark Tinka]

On Wednesday 06 September 2006 12:40, Mark Tohill wrote:

> I may be digressing here, but on the subject of the validity
> of full BGP routes in relatively small SP environments, is
> there anything else apart from the obvious loss of granularity
> in moving from full feeds to a default? We are in a
> 'muti-homed to same provider' scenario.
>
> Is there any strong argument to maintain these assuming CPU,
> memory are n't restricted?

If I may...

The fact that you are multihomed to the same upstream 
notwithstanding, I think it would be a great way to detect 
far-end link issues, i.e., beyond (or farther, within) your 
upstream's routing domain.

Your upstream could originate 'default-information' to you via 
eBGP, but that might not help if the link between the both of 
you is up, but that to their own upstream(s) isn't, or is 
partially operational.

Yes, there are ways of going around this, but if you can support 
a full feed, I'd think it's the least complicated method to 
maintain connectivity.

Cheers,

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 827 bytes
Desc: not available
Url : https://puck.nether.net/pipermail/cisco-nsp/attachments/20060906/08e167ce/attachment-0001.bin