On Tue, 2008-04-15 at 08:34 +0100, Dean Smith wrote: > The constraint in my case is the MPLS product I can buy. Each resilient > access is configured as its own private AS for BGP. The core AS uses > the SP registered AS. I have no option but to have a route table full > of <private as>.<sp mpls as>.<private as>
I'm may have misunderstood your setup, but AFAIK "allow-as in" and "as-override" were designed for exactly the Customer-SP-Customer scenario. It's a shame that the provider doesn't want to use them then. > I have no option to impose a single consistent AS on my Supplier- and > realistically only the one supplier for size of network I need in my > market (Full national coverage of the uk). I noticed this in OPs configuration: On Fri, 2008-04-11 at 13:30 +0100, Gary Roberton wrote: <snip> > router bgp 2856 <snip> AS2856 is "BT-UK-AS" according to whois. I assume it covers most of the UK. Coincidence? ;-D On Tue, 2008-04-15 at 08:34 +0100, Dean Smith further wrote: > As it happens we do have our own registered AS we use for our internet > facing presence and a limited number of external peerings. But again > I have to jump through hoops to present all routes as that single AS > after all no-one wants to see which SP I use for my MPLS core or how > many private AS I have. Given I can change/amend/delete/add almost > every other metric in the BGP decision process it seems strange I dont > have full control to manipulate the AS path aswell. I completely agree. More control would be better. I'll try and mention it to the next SE/AM I see, even though I think my (medium enterprise) voice means little to them. > still life would be dull if it was all easy. Yes, the "Plug'n'Pray" revolution still seems a few steps away in core networking. :-) Regards, Peter _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
