Ref: Your note of Thu, 14 Apr 94 19:39:42 EDT Elise,
Yes, we have again started trying to get folks to move to BGP4 -- heck we are still trying to get some folks to move to BGP.
Trying to get folks to move to BGP4 is a good thing to do and it should be encouraged, but given the time constraints this is unlikely to be a feasible solution to the problem in the near future.
The results are slow in coming...
It seems that some folks PERSISENTLY missed the point. The clock is ticking. The only way to slow down the clock is by reducing the volume of routing information (via CIDR). Nothing else is going to stop or slow down the clock. The clock cares less whether "the results are slow in coming" or why they are "slow in coming". And, by the way, when the clock handle hits the mark there is going to be a massive crash. Forgive me my bluntness, but given where we are today I just don't see any feasible alternatives to a medium-large scale proxy aggregation within a next month or so. So, given the finite amount of human resources we have, let me suggest that the CIDR efforts should be strongly focused on the following two FEASIBLE choices: (a) make sites who have BGP-4 TODAY to withdraw individual components asap (b) make sites who don't have BGP-4 TODAY to arrange proxy aggregation by their providers (who assumed to have BGP-4) asap (or the providers will arrange it for them). Yakov.
Yakov,
Ref: Your note of Thu, 14 Apr 94 19:39:42 EDT
Elise,
Yes, we have again started trying to get folks to move to BGP4 -- heck we are still trying to get some folks to move to BGP.
Trying to get folks to move to BGP4 is a good thing to do and it should be encouraged, but given the time constraints this is unlikely to be a feasible solution to the problem in the near future.
The results are slow in coming...
It seems that some folks PERSISENTLY missed the point. The clock is ticking. The only way to slow down the clock is by reducing the volume of routing information (via CIDR). Nothing else is going to stop or slow down the clock. The clock cares less whether "the results are slow in coming" or why they are "slow in coming". And, by the way, when the clock handle hits the mark there is going to be a massive crash.
Forgive me my bluntness, but given where we are today I just don't see any feasible alternatives to a medium-large scale proxy aggregation within a next month or so.
So, given the finite amount of human resources we have, let me suggest that the CIDR efforts should be strongly focused on the following two FEASIBLE choices:
(a) make sites who have BGP-4 TODAY to withdraw individual components asap
(b) make sites who don't have BGP-4 TODAY to arrange proxy aggregation by their providers (who assumed to have BGP-4) asap (or the providers will arrange it for them).
Yakov.
If you refer to the list of the top ten ASs, you can see that large gains can be attained without being heavy handed. If the top 10, several of which are BGP4 capable, cidrize the savings would be substantial. --Elise
Elise:
If you refer to the list of the top ten ASs, you can see that large gains can be attained without being heavy handed. If the top 10, several of which are BGP4 capable, cidrize the savings would be substantial.
Probably true. While Peter L.'s comment is of course stretching it way too much in terms of "where's a whip there is a way," I believe that in the interest of the survival of the network, having to hurt people's feelings may be the lesser evil. As such, there should be a rather strong encouragement towards making progress here. Also, I believe it would be good for someone to take a very strong lead on renumbering ("I want my net 35 back; I fought so hard with Jon for it! It's MINE!"). Are you and/or Jon as the RAs taking the lead there? What is your implementation plan? If you don't have one, what are your current projections of when and how things will become critical? What is your model to define criticality? Hans-Werner
According to NSF/ANSNET data, 23 aggregates ALONG with its covered 898 specific routes are co-exist in the routing table now. If the ASs withdraws all these more specific routes, the routing table could be reduced by 898. Given the urgent need of reduce the routing table size, would these ASs withdraw those specific routes as soon as possible? --Jessica
participants (4)
-
epg＠merit.edu -
hwb＠upeksa.sdsc.edu -
Jessica Yu -
yakov＠watson.ibm.com