-----BEGIN PGP SIGNED MESSAGE-----
Iljitsch van Beijnum [mailto:iljitsch(_at_)muada(_dot_)com] wrote:
On woensdag, aug 27, 2003, at 13:18 Europe/Amsterdam, Jeroen Massar
wrote:
I totally agree with your current insight that we need to seperate
the routing from the host identifier. IMHO every host should have
one globally unique ID and could have multiple transports, even if
those are IPv4, IPv6, IPX or whatever based and going over multiple
links or not. Though we should limit to IP based protocols to not
make it too complicated. Such a mechanism could solve problems for:
"site-locals" constructions, multihoming, mobile-ip without hindering
the size of the routing table as people could continue to use current
routing, thus TLA based and fully aggregated to the TLA level in the
GRT.
The multi6 wg has been working on locator/identifier separation as a
way to solve the multihoming in IPv6 problem for a while now.
And ever since they haven't progressed much unfortunatly :(
The problems we're facing (apart from the fact that there are
many ways to skin this particular cat and everyone has a different
preference) is that additional mechanisms are needed to get the extra
information across, and there is a price to be paid in one or more of:
additional round trips, more dependence on the DNS or something similar
to DNS,
Indeed, one will always have this problem, my current idea relies
on DNS, but if a program currently tries to contact another host
it will usually already do a forward resolve and some apps
(ssh for instance) also like to do a reverse resolve already.
So these shouldn't be much of a problem. In the idea I have I
have also added the possibility to limit these lookups until
a failure occurs.
additional per-packet overhead,
Not required for my idea.
loss of backward compatibility,
I covered that in my idea, by simply adding functionality
and not replacing it in a manner which could cause
unexpected effects on the installed base.
The installed base won't have the benefit
increased complexity.
Not overcomable, but it should not be noticeble much in
the idea I have. It's also more of a Proof of concept for me
to see if I can get such a mechanism to work and to proof
that such mechanisms are viable. Query me at RIPE next week
if you want to hear more about it.
Greets,
Jeroen
-----BEGIN PGP SIGNATURE-----
Version: Unfix PGP for Outlook Alpha 13 Int.
Comment: Jeroen Massar / jeroen(_at_)unfix(_dot_)org / http://unfix.org/~jeroen/
iQA/AwUBP0zaOimqKFIzPnwjEQL+ZQCeIR0X0MEUnLsow6IGwKfFLsYTstEAn2SQ
zMvifMn7z9A12oPp82cjXmCF
=LdLQ
-----END PGP SIGNATURE-----