[ih] Failed Expectations: A Deep Dive Into the Internet’s 40 Years of Evolution (Geoff Huston)

[John Day]

What Karl was referring to, wasn’t really the Session Layer. That had been stolen by CCITT for Videotex. By 1983, OSI had determined that the upper 3 layers didn’t exist, were one layer, and fixed it so it could be implemented that way. What he was referring to was ACSE, which was the common means for opening a connection to an application. Something the Internet doesn’t support. ACSE support what the Internet does opening a connection to an arbitrary instance of an application. (But it does it without emulating jump points in low memory as the Internet does.) It goes further an enables opening a connection to a specific instance of an application, and opening a connection with different protocols to a specific instance of an application. Neither of which the Internet supports.

I really don’t understand why this sudden interest in OSI. It has been gone for 30 years. (Although it does seem to be a recurring theme.)

Is that because it takes attention away from the Internet’s fundamental flaws?  Wouldn’t it be more productive to figure out how to rectify those mistakes? 

There are certainly enough of them. I have never seen a group with such a track record: 0 for 7. At every major decision point, and there are 6 or 8 of them, with the right answer and wrong answers known, the Internet consistently chose the wrong one. 

It sure is a good thing that with software you can make almost anything work. Then no one will notice.

I agree with Joe. I never even cover the OSI model when I teach the introductory course. It would be more productive to discuss the Internet’s mistakes.

Take care,
John

> On May 31, 2023, at 20:11, Brian E Carpenter via Internet-history <internet-history at elists.isoc.org> wrote:
> 
> On 01-Jun-23 11:33, Robert Stanford via Internet-history wrote:
>> On 1/6/23 05:30, Karl Auerbach via Internet-history wrote:
>>> 
>>> 
>>>  - An association (session) layer.  This would have made mobility much
>>> easier by allowing an "association" to span multiple transport
>>> connections as devices move and change their IP addresses.  It also
>>> could have reduced the web's addiction to cookies. (OSI made the
>>> mistake of making their session layer excessively complicated,
>>> incomprehensibly documented, and never explaining what it was good for.)
>>> 
>> This now somewhat exists in Multipath TCP.
> 
> And https://datatracker.ietf.org/doc/draft-ietf-quic-multipath/
> And for that matter in SHIM6 (RFC 5533) although that's failed to deploy.
> And the IPv6 flow label, e.g. RFC 7098.
> 
> But this all seems *slightly* too low a level for real session identification.
> 
>   Brian
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history