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

[Dave Taht]

The 0.0.0.0 thread has been fascinating and I now have more to read
than I ever imagined I would. Moving sideways...

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?

As near as I can tell, this netnews exchange

https://web.archive.org/web/20030718205943/http://www-mice.cs.ucl.ac.uk:80/multimedia/misc/tcp_ip/8813.mm.www/0178.html

ultimately resulted in CIDR and the next message, kicked off IPv6 - for
the toasters!

https://web.archive.org/web/20030913113707/http://www-mice.cs.ucl.ac.uk/multimedia/misc/tcp_ip/8813.mm.www/0237.html

Somewhere also in this fascinating month of internet history (morris
worm days) someone suggested starting to use the class-e space, with no
replies.... Were there proposals to use this as an extension of some sort?

Anyway it seems to me (in retrospect) that everybody *knew* 32 bits
wasn't enough even going back as far as 1981 (loved hearing about how
tcp went from v2 to v4) ?) but it didn't become
recognized as a serious problem til about this era. ?

>
> I personally (in my current naive state of mind) would like to see 
> daemons daemons source traffic from any of the other 65,535 ports 
> excluding than the destination port as a possible source port.

IANA suggests a smaller ephemeral port range than what linux uses.
https://en.wikipedia.org/wiki/Ephemeral_port

Given the profusion of NAT-like nasty things like CGNs and DS-lite and
the rise of QUIC I would favor extending the ephemeral port range as far
as possible. 

>
> Aside:  Why is port 0 special?  }:-)