RFC1597, CIDR

[jmorriso@bogomips.ee.ubc.ca (John Paul Morrison)]

> 
> Date: Fri, 02 Dec 1994 01:35:20 +0100
> From: Geert Jan de Groot <GeertJan.deGroot@ripe.net>
> Subject: RFC1597 
> 
> 
> 
> > I have occasionally heard the suggestion that we should have instead
> > gotten a number of Class-B or Class-C networks.  I think that is a
> > short-sighted view, as it trades the possibility of an independent
> > radio-based IP network for immediate connectivity over commercial
> > internet providers.
> 
> True, though the whole idea of class-A-B-C is fading out. Think about
> CIDR, where one works with a network prefix and a host suffix, which
> both have variable length. I.e. 44.137.254.128/30 is a 30-bit
> network prefix, and a 2-bit suffix (i.e. for 4 IP addresses).
> 
> This should not sound strange to people who were introduced into
> TCP/IP using Phil's package as this package has been classless
> from the past (good foresight, phil!)
> Most of the Internet world speaks CIDR these days, so it should
> be OK to announce a subnet of net 44 now. Anybody interested in
> an experiment or two?

This sounds nifty. How would this be set up? It's silly for packets
destined for 44.0.0.0/8 to be routed down to California when the
actual host is much closer. It sounds like it would be possible to set
up regional versions of 'mirrorshades' for each country/province,
possibly even do away with all this encap business. Can you post some
more info on this (tech/administrative)?

> 
> 73 Geert Jan PE1HZG
>