[c-nsp] ISP / MPLS "POP" design

[Mark Tinka]

On Thursday, November 07, 2013 12:29:24 PM Adam Vitkovsky 
wrote:

> Well I have even more perverse idea of true link-state
> BGP, I mean literally running version of constrained SPF
> algorithm taking into consideration all the current BGP
> path attributes.
> And it would advertise segment labels as well :)

That's the one that scares me. 

BGP works very well, now, because it's not a link state 
roting protocol. BGP-LS could give us some insight into how 
we could potentially break BGP by trying to make it a pseudo 
link state routing protocol (may not be such a big issue 
globally, since I suspect the benefits of BGP-LS are likely 
to be constrained to a controlled set of environments, 
either within the AS, or among tightly strung AS's.

For BGP to take over the duties of an IGP, I'd reckon a sub-
BGP protocol that does not interfere with the mature path-
vector BGP of old, or knobs that can enable the link state 
portion of mature BGP in ways that do not interfere with 
normal global BGP operations. 

I think the benefit - if I had to put myself out there - 
would be more for having a single dynamic routing protocol 
in the network (I doubt we shall ever run away from static 
routing, however), rather than the benefit being the scaling 
properties of BGP (since, by definition, link state routing 
protocols can't scale well as BGP-proper does).

The whole idea leaves a bad taste in my mouth, but as we all 
know, there is always that guy :-)...

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/20131107/53bbbac1/attachment.sig>