[788] in Commercialization & Privatization of the Internet
Re: IETF questions -- Internet growth
daemon@ATHENA.MIT.EDU (Milo S. Medin" (NASA ARC NSI Proje)
Fri May 31 01:49:15 1991
To: bzs@world.std.com (Barry Shein)
Cc: fbaker@acc.com, francis%zaphod@gargoyle.uchicago.edu, com-priv@uu.psi.com,
In-Reply-To: Your message of "Thu, 30 May 91 14:33:05 EDT."
Date: Thu, 30 May 91 22:46:02 -0700
From: "Milo S. Medin" (NASA ARC NSI Project Office) <medin@nsipo.nasa.gov>
Barry, in fact, Class D multicasting is being used. OSPF uses IP multicasting
instead of broadcasting it's updates. Work is going on right now to
add multicast routing capability to OSPF as well, and this should greatly
increase the number of applications that use multicasting.
Personally, I think it's rather silly to extend the notion of Class A, B, and
C nets to add more "natural" masks. It's just easier to do away with
the notion of Class in networks (Mask egalitarianism!) and just support
variable length masks. OSPF doesn't have the notion of Class A, B, and C
nets in it -- every net carries a mask along as well. If we had an
EGP that supported this as well, we could do all sorts of good things to
conserve number space, even including implementing a pseudo-hierarchy
ISO style. But we don't alas... :-(
Paul Tsuchiya has outlined a very nice approach where network allocation
could occur in this sort of environment with flexible use of masks. And
this can all be done without changing the hosts. This is the real problem with
a conversion to OSI - you don't want to change the OS's and all the applications
out there. What's done between consenting routers however is none of the host's
business! I'm sure Paul would be happy to point you to his paper on the
subject.
Thanks,
Milo
