I think, this is a pretty challenging topic for any routing architecture and not implicitly solved by any one. Constraint: It should not multiply the worldwide routing churn. Particular not at those far remote regions where still the same path were taken. Also, the destination host is not the only one who has a service contract with someone else. ISPs, mutually, too.In principle the very first hop may have to be compliant with the very last hop like all other in between too. As I have expressed myself several times, proper done hierachical routing is THE way to match any scalability problem. And yes, the hierarchically zoom could even be extended as to conceive hosts even as nodes and therefore can do whichever TE routing using different last hops to the host via different ISPs. As a side effect the host address, though assigned by one particular ISP, may as well be used by some other ISP just as if it were a PI address. Heiner In einer eMail vom 25.04.2008 11:41:54 Westeuropäische Normalzeit schreibt [EMAIL PROTECTED]:
Thanks, everybody, for sharing your thoughts in this exercise. The discussion has certainly shed some light on the preferred approach of the research group in things routing control. Here is my conclusion: Solutions should preferably and by default give control over a particular resource to those entities who pay for that resource. In the specific case of ingress link selection, this means that the receiving edge network should be able to choose. Solutions may go beyond this and provide a means for entities to cede (part of) their control. Continuing the example of ingress link selection, a receiving edge network may announce a selection of alternatives from which sending edge networks can choose. - Christian -- to unsubscribe send a message to [EMAIL PROTECTED] with the word 'unsubscribe' in a single line as the message text body. archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg
