[772] in Commercialization & Privatization of the Internet
IETF questions -- Internet growth
daemon@ATHENA.MIT.EDU (Brian Lloyd)
Thu May 30 00:14:59 1991
Date: Wed, 29 May 91 21:06:31 PDT
From: brian@napa.Telebit.COM (Brian Lloyd)
To: ietf@ISI.EDU, com-priv@uu.psi.com
Reply-To: brian@napa.Telebit.COM
It seems to me that we need to get away from the concept of only three
classes of network. I would like to see a scheme that allows
assignment of network number on any arbitrary bit boundary within the
IP address. Basically you could get the address space you need but
you wouldn't require gross overkill. This would require a mechanism
to propagate the size of the network number along with the network
number itself but the interior routing and gateway protocol(s) could
do that. Quite frankly, it is the subnet mask that is more important
today than whether the network number is for a class A, B, or C
network.
Let me use Telebit as an example. Telebit has a class-B network
because we have quite a few subnets and the NIC found it easier to
assign us a class-B network rather than many class-C networks. But
Telebit has only about 400 hosts -- a gross waste of address space.
Out of an address space capable of supporting 64K hosts we are using
about 400 addresses. The likelyhood that Telebit will use all this
address space is small (unless all you people out there buy
NetBlazers, make Telebit fabulously successful, thus saving the
address space from being wasted :-). We currently could grow by a
factor of 150 within the current address space. On this other hand,
using the scheme of arbitrary length network numbers I would opt for a
20 bit network number, a six bit subnet, and six bits for host number.
I think that a factor of 10 is more than enough growth capacity and we
can always apply for more address space if the need arises.
Brian Lloyd, WB6RQN Telebit Corporation
Network Systems Architect 1315 Chesapeake Terrace
brian@napa.telebit.com Sunnyvale, CA 94089-1100
voice (408) 745-3103 FAX (408) 734-3333
