[ih] Some Questions over IPv4 Ownership

Noel Chiappa jnc at mercury.lcs.mit.edu
Wed Oct 13 08:24:53 PDT 2010


    > From: Ernie Rubi <ernesto at cs.fiu.edu>

    > On Oct 11, 2010, at 8:35 PM, Vint Cerf wrote:

    >> the initial address space was decided on in 1977

While Vint will have to answer this definitively, let me give you a sense of
what quadrant the answers will be in....


    > Do you remember who decided the space and who employed that person(s)

In answering this, I think it best to give you a little bit of background to
the development of IPv4 - because your question does not have a simple
answer. The choice of a 32-bit space was only the final stage in a longer
design/decision process, and prior versions of the design had a very similar
design,

To start with, you need to understand that in the beginning, there was no
separate IP protocol: TCP (the protocol that provides a reliable stream
service) had not been split apart from IP.

With that in hand, here is a table of the various versions preceeding IPv4,
with their dates, and what their address spaces looked like:

TCP 1 (documented in RFC 675)
December 1974 - Variable length (network number given in a field of up to 15
'nibbles' (i.e. 4-bit chunks), plus a 2-byte 'host number' on that network)

TCP 2 (documented in IEN-5)
March 1977 - Fixed 32 bits (1 byte network number, plus a 3 byte 'host number')

TCP 3 (documented in IEN 21)
Jan 1978 - Variable length (given in a field of up to 15 bytes)

TCP/IP 3.1 (documented in IENs 26/27/28)
Feb 1978 - Fixed 32 bits (as for TCP 2)

TCP/IP 4 (first documented in IENs 40/41)
Jun 1978 - Fixed 32 bits (as for TCP 2)

Today's IPv4 addresses (fixed 32 bits, but without any fixed boundary between
'network number' and 'host') came about in two even later steps; the
provision of class A/B/C addresses (circa 1980-81: they are not in RFC 760,
January 1980, and are in RFC 791, September 1981), and the adoption of CIDR
(circa September 1993 - RFC 1338 and RFC 1519).


So if you're asking i) 'who first decided on _an_ Internet host address
space', the answer is Cerf et al, in TCP 1, and they were all at Stanford at
the time. If you're asking ii) 'who first decided on _a_ 32-bit namespace',
it was again Cerf et al, in TCP 2 - but note that this was later replaced
with a different design for a while! If you're asking iii) 'who first decided
on the _current_ 32-bit namespace', it came in several stages: the first
(iiia) was Cerf et al, in TCP 3.1, going back to a 32-bit space - but it
didn't take today's form (with semi-arbitrary-sized chunks of that space)
until CIDR (which we can call iiic, with iiib being A/B/C).

When the particular engineering decision to go with a 32-bit address space
was taken, for both forms ii) and iii) of the question, Vint (whom I believe
has the largest share of the credit/blame for iiia), at least :-) was
employed by the USG, at DARPA (IIRC he was there from 1976-1982, after being
at Stanford previously). Others involved at that time were employed by a
variety of private entities, I think all under Government contract from DARPA.

As for all the others (e.g. the later CIDR stuff) - that would take a certain
amount of research to figure out (e.g. who was funding Cerf/Sunshine/Dalal at
Stanford in the early 70s).


    > where it was decided (at work, at home)?

It certainly wasn't a private decision (if that's what this question is
about). I would think that most of these decisions were taken at meetings
(most of the early ones all at DARPA-funded Internet project meetings; the
CIDR one at IETF meetings, for which the funding situation is infinitely more
complex).

    > Were you and Mr. Postel's actions under the control/direction of the US
    > Government or a agency of said gov't at the time? 

Vint was a US Govt employee during the ii)/iiia)/iiib) decisions. Jon, as an
employee of USC-ISI (from March of 1977 until his untimely death; before that
at UCLA, Mitre and SRI) was always, I _believe_, acting in the context of a
contract between the USG and ISI.

However, as others have noted, these contracts at that point in time did not
specifically say 'do this specific task, then that specific task', but were
rather more general - 'do good stuff in this area' kind of thing.

About i), I don't know. About iiic), most of those involved were neither
government employees, nor acting at USG direction under a contract from the
USG.

	Noel



More information about the Internet-history mailing list