-----Original Message----- From: Chrisy Luke <chrisy@flix.net> <snip>
Even in your simplistic view, you would have to transfer 16meg of data
Please note, I did not suggest that people implement the simplistic view. I suggested that people focus on routing and not forwarding. Also, if one did implement a brut-force approach 16meg would not be transferred. You could transfer exactly what you transfer now and the table would be built inside the device where the 16 meg is stored. For example, if information was obtained that all traffic for a /8 needed to be sent via a certain interface then 65,636 entries in the brut-force table would be filled in with that small amount of information from an external device. Keep in mind the premise was that we would look at the entire address space as /24s and have an entry for each. That does not mean that the forwarding device needs to be told about each of the /24s. It can be told about large groups of them. That does not require 16meg of data to be transferred. Again, this discussion started with someone claiming that /24s could be added to the routing tables. My point was that current technology does not encourage this and the Internet governance of the industry has been developed to support what the hardware vendors have to sell. In other words, we can not start with a clean sheet and ignore the way current devices work (or do not work). My other point is that I do not think that vendors will worry that much about fixing the forwarding technology because "routing" is going to become very key to the hardware vendor's survival. I do not think that many Internet operators have begun to address what is going to happen when they have to deal with the world of routing vs. forwarding. Jim Fleming Unir Corporation - http://www.unir.com End-2-End: VPC(Java)---C+@---<IPv8>---C+@---(Java)VPC http://www.ntia.doc.gov/ntiahome/domainname/130dftmail/unir.txt http://www.ddj.com/index/author/idx10133.htm