In einer eMail vom 09.12.2007 04:56:28 Westeuropäische Normalzeit schreibt
rw@firstpr.com.au:
Heiner, I am not responding to your suggestions because I am only Robin, as I wrote before: My concept is incrementally deployable because
whoever can afford to collect 400 000 entries can easily collect 1K-2K entries
in addition whereby the "routing churn" for collecting this additional info
tends to zero.
The IETF may develop LISP, but the IRTF should do research,
imho.
I think a good solution would be one we are confident will work for Well, the algorithm I used in my pdf-file is almost 20 years old (it has
been published by Kurt Mehlhorn in 1988) and I used it to compute the best next
hop for any destination outside the current node's 1x1 square degreee geopatch.
It will still be working in 15 years from now just like my own algorithms
which go beyond Dijkstra and which are rather 50 years ahead of the current
IETF's state-of-art.
But I am afraid that the IETF community will never accept that
Dijkstra did not finish the job, and therefore it will stick to the stone-age
like TTL-mechanism forever. It will not accept that Dijkstra only provides the
absolute minimum of routing information and will ignore any progress on this
matter (e.g. by calling it hand waving ...)
As a matter of fact there could be much more ambitious goals:
Not only to abolish the scalability problem but also to provide "multiple
topologies inter-domain", or:
Multicast to ten millions of receivers (without using any multicast address
btw),
or real-time traffic balancing (of course without MPLS) etc.
etc., all goals I experienced to be rated even VERY POOR.
Robin, while following the discussion, I got the impression that your
contributions haven't got the appropriate attention either. We both are
loosers, aren't we?
Bonjour tristesse,
Heiner
|