[ih] When did "32" bits for IP register as "not enough"?

[Jack Haverty]

On 2/13/19 11:08 AM, Dave Taht wrote:

> So, it seems obvious that address size problems plagued the arpanet
> and earlier versions of IP. When did the writing show up on the wall
> that the classful design wasn't working, and secondly that 32 bits
> wasn't enough?

I don't recall hearing any concerns that the ARPANET address space was
too small, other than the time when the "leaders" were expanded from 32
to 96 bits.

This was the era when computers were big and expensive, so there weren't
many of them.   What changed was the advent of minicomputers,
workstations, and PCs, along with the various kinds of LANs that made
computers cheap enough to have millions of them.

Despite that evolution, I don't recall much concern about address size
in the early IP days.  Remember, at that point the Internet was an
Experiment, and it was supposed to eventually go away when the
CCITT/ISO/Industry deployed the "real" infrastructure system for
computer communications.   The IPV4 address space was plenty big enough
for all the anticipated experiments and early deployments, e.g., by
DARPA and NSF, while waiting to transition to the "real" system.

Of course that never happened, and the Internet evolved as the only
system which you could actually buy, install, and use.  The IPV4 address
space only started to be a problem when ISPs started to proliferate, and
each needed a block of addresses for its customers.

When NAT appeared, IMHO it took a lot of pressure off the address space,
since the millions of small LANs could then share the same reserved
address spaces like 192.168.x.x or 10.x.x.x (in memory of the ARPANET
which had been retired).

/Jack Haverty