I missed the part about this being configured only on a nontransit AS. If it's only for a stub AS then it seems like the worst case would be the stub AS losing whatever service guarantees they have with their ISPs ... or are there ISPs that offer service guarantees for traffic that leaves their network?

Lee

|---------+----------------------------> | | Rodney Dunn | | | <rod...@cisco.com| | | > | | | | | | 12/01/2004 09:18 | | | AM | | | | |---------+----------------------------> >---------------------------------------------------------------------------------------------------------------------------------------------| |
| | To: lee....@census.gov
| | cc: Rodney Dunn <rod...@cisco.com>, "David J. Hughes"
<bam...@hughes.com.au>, "'cisco-nsp'" <cisc...@puck.nether.net>, Gert | | Doering <ge...@greenie.muc.de>, Brian Feeny <sig...@shreve.net>
| | Subject: Re: [c-nsp] Growing BGP tables
| >---------------------------------------------------------------------------------------------------------------------------------------------|

No. I was talking about this only being configured on the RouterA/Routerb pair Where that represents it's own nontransit AS.

That's why I asked what deployment scenarios are most of you asking about for this.

I see this more as a benefit for a stub AS rather than any AS that is transit.

AS1 AS2

|---------- AS3 --------|

I put in the proposal the next hop check but are you asking that is a less specific prefix exist in the RIB then just dont' install a more specific regardless of how/where that less specific was learned?

Rodney