[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

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

On 2019-02-17 12:07, Dave Crocker wrote:
> On 2/16/2019 2:53 PM, Brian E Carpenter wrote:
>> And OSI had one convincing argument: bigger addresses, for those
>> who already believed that 32 bits was insufficient.
> When I was managing parallel development efforts of both TCP and OSI 
> stacks, I don't recall hearing from any of my customers that they were 
> concerned about address space.
> This was in the late 1980s, long before the Web.

I completely believe that. But for those of us thinking about
intercontinental-scale networking at that time (specifically, in my
case, for high-energy physics; but there was also the space science
community), scaling was an issue, and our experience of DECnet Phase IV
running out of address space was a wake-up call. So the address space
offered by CLNP was very definitely an attraction, coupled with the
theory that the GOSIP requirements would generate widely available
and affordable products. Duh.

> They were concerned about turn-key operations and interoperability. 
> They got that from the TCP stack.  They never really got it from OSI.

Indeed. And OSI products were in general ridiculously expensive.
Fortunately, we never wasted much money on them at CERN; the only
serious effort was deploying DECnet Phase V.
> Since OSI was the official strategic choice of countries and companies, 
> but some had been using TCP, we started asking about their needs to 
> tools to support their transition from TCP to OSI.  What we got back was 
> a resounding and strong request that, instead, we provide for their 
> accomplishing OSI to TCP transitions...

As far as I know that problem never really hit high-energy physics, except
in the form of switching off DECnet Phase V when VMS was eventually phased
out. But that was after I left CERN, so I don't know the details.

[If anyone is interested, there are several versions of how TCP/IP
got into CERN, before TimBL chose it for the web:
1. Ben Segal's version: http://cern.ch/cnlart/2001/001/tcpip
2. Olivier Martin's version in: http://ictconsulting.ch/reports/European-Research-Internet-History.pdf
3. My version in chapters 7 & 8 of: https://sites.google.com/site/bcabrc/network-geeks-book
4. Robert Cailliau's version in "How the Web was born" (OUP, 2000) ]