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

[Brian E Carpenter]

On 2019-02-23 11:30, Dave Taht wrote:
> Brian E Carpenter <brian.e.carpenter at gmail.com> writes:
> 
>> On 2019-02-16 08:23, Clem Cole wrote:
>>> On Thu, Feb 14, 2019 at 2:58 PM Craig Partridge <craig at tereschau.net> wrote:
>>>
>>>>
>>>> Dennis' decision to adopt TCP/IP for NSFNET was critical.
>>>>
>>> Hmmm.  It certainly did not hurt and I agree was cleafrly an amplifier and
>>> important.   But I'm not sure that was as important as the basic economics
>>> of the time -> Moore's Law and cheap cycles (i.e. I don't think the NFSNET
>>> choice, while helped the effect, was not the high order bit on the success
>>> function).
>>
>> I took the original question to be about when the success of TCP/IP became
>> formally recognized. You are undoubtedly right that its market success was
>> a result of availability in BSD, with a usable API, just when Unix was taking
>> off.
> 
> My original question was kind of overly vague - we'd found a netnews
> conversation in 89 between jon postel, vint and others that seemingly
> led to both CIDR and ipv6 ("toaster-net"), as "32 bits is not enough")
> 
>>From a formal perspective the class B exhaustion documents mentioned on
> these threads, at two ietfs, kicked restructuring the internet to handle
> more addresses into higher gear, and it was realized that OSI had no
> future benefit once that was addressed.
> 
> ...
> 
> The writing on the wall moment for IP over OSI seems to be from around
> the time of the "kobe revolt" from what I've been reading here. All the
> events in 1992-94 - not just the rework of address spaces, but the
> opening up of the internet, ISPs getting off the ground, cheap modems,
> al gore, etc, etc....

Ironically there was a lot of *literally* IP over OSI (where OSI meant
CONS, connection-oriented network service, i.e. X.25) during the 1990s.
(French-speaking readers will note that this was not necessarily the
worst acronym in networking history.) In Europe, in some cases X.25
was the only connectivity you could get or afford. Leased lines were
still monopoly-priced.

More seriously, we shouldn't ascribe too much importance to the Kobe
incident outside the IETF/IAB community. The community reaction was
really against the IAB being on the wrong side of history for once.
But it was technical and market factors that were driving history,
as others have described.

> 
> ...
> 
> going off topic yet again:
> 
> I do think people overstate the early impact of the web nowadays.

Well, as I said in my book, the key moment seemed to be around
the middle of 1995 when the .com phenomenon really started.
After that I'm sure it became a driver. I used to check the ads
in each week's Economist. Unfortunately I didn't log the results,
but somewhere around late 1995 URLs started to appear instead of
email addresses, and within a couple of years an ad without a
URL became unusual.

> 
> It was being able to talk to more people you had something in common
> with than ever before.
> 
> When I started my ISP in 94, it was netnews and email that were the
> primary draws, and the web thing was... well... a "browser". Yes, it
> turned out folk had a huge appetite for cat videos, but huge numbers of
> folk in college had been exposed to "real email" and netnews and wanted
> that at home and it was within reach of the average geek, finally.
> 
> Everybody already had modems to contact compuserve and BBSes, (didn't
> the first laptops also have integral modems?) so when slip started
> working, BBSes started to die. There were very few Unix based BBSes (I
> had a hand in one in 86, and this one in 1990, which amazingly still
> exists: https://cellar.org/faq.php ) and their draw became "being
> multiuser".
> 
> we had real apps also (client/server databases and so on), that helped
> drive corporate adoption of IP vs ipx. Dialup access to big databases
> for remote offices was done not via browser but a huge host of tools
> (now departed) that created a custom UI element over that sql database.
> 
> I've been sitting here trying to remember some of those. Borland had
> some, there was a really popular one (power-something) - hundreds and
> hundreds of dollars per seat that corps were willing to spend to get
> their remote offices online.
> 
> It was a heady time! 
> 
> ...
> 
> OK, here's a story that I rarely tell and is OT:
> 
> in late 1993 I'd been having a bad time on every front and hadn't slept
> well for a few months, and my network was under attack, and I was
> essentially seeing snakes under the bed. I had a discussion with a
> shrink, and at some point I said something like "my machines talk to
> millions of people every day". (I was trying to explain netnews). She
> wrote (and I read, upside down):
> 
> "The patient is grandiose and delusional".

That is possibly the best anecdote ever.

   Brian

> 
> (Thankfully I escaped from that particular interview relatively unscathed
> and later found a shrink with a physics background that I could actually
> talk to which straightened me out a lot)
> 
> A year later we'd found funding for ICANECT and a few months after that
> we had 16,000 people online doing the same things that clueless shrink
> had dismissed as fantasy.
> 
> Me, grandiose? *always*
> 
> Delusional? Sometimes I worry, but not very often, anymore.
> 
>