|
In einer eMail vom 29.03.2008 08:43:33 Westeuropäische Normalzeit schreibt
pekkas@netcore.fi:
Isn't this whole discussion irrelevant? Not even Microsoft cares about updating all its own older operating
systems, but the RRG does.
What's research got to do with it?
My impression: All the current discussion is rather lobbying than
researching. Arguments are used to talk down other people's solutions.
By research I understand questioning the existing paradigms as a first
step.
No single paradigm is questioned by any of the currently discussed
concepts. LISP is a two loose hops forwarding technique, not a loc /id
split.
I once mentioned a good example for loc/id split: The social security
number (=SSN) for US-citizens is most probably one of the best identifiers. It
will not change at all (whereas by getting a new passport you will of course get
a new passport number). The SSN identifies some US-citizen no matter how often
and how fast he moves from one place to another, i.e. changes his address =
location.
So, can't we have provider independent identifiers being identifiers
and AS-number combined with geographical coordinates being locators ?
Multihoming: the PI identifier is unchanged, the dest.AS.number may vary,
the geographical coordinates are unchanged.
Mobility: the PI identifier is unchanged, dest.AS may or may not change,
the geographical coordinates change.
I think Lars Westberg also suggested to include the AS number in the IP
header not long ago.
"Rekther's law" talks about two choices. Why aren't both analyzed? Why is
one of the two choices entirely ignored? Could it be that no one knows which
choice the current architecture complies with ?
Heiner |