[Info-vax] Rethinking DECNET ?

[Bob Gezelter]

On Saturday, September 6, 2014 4:32:58 PM UTC-4, Johnny Billquist wrote:
> On 2014-09-06 17:59, JF Mezei wrote:
> > On 14-09-06 07:05, Johnny Billquist wrote:
> >
> >> I would say that what happened was that already in the 80s, TCP/IP
> >> proved that it already existed, it already worked, and was already
> >> available on multiple platforms. In addition, it was actually a project
> >> funded by the US Department of Defence, and of course they were using it.
> >
> > One problem:  the IP stack and applications were not up to snuff to
> > replace security/functionality provided by the likes of DECnet. And IBM
> > may not have had much functionality with SNA, but it had and wanted to
> > maintain security levels.
> 
> I think we've already covered the security of DECnet, which is totally 
> non-existant.
> 
> Functionality wise, DECnet do have some things in there which is way 
> better (in my opinion) to what exists in the IP world, but also some 
> things that are horribly worse. (CTERM anyone...?)
> 
> > SMTP did not have reliable ability to have read receipts, and many other
> > functions that were part of X.400
> 
> Read receipts... I've never understood the point of them. It puts the 
> world in a very syncronous mode, which sucks. MAIL-11 is still no fun, 
> compared to SMTP. The receiving node have to be up when you write the 
> mail, or else you cannot send it.
>
> This is something I can understand for something like PHONE (or the 
> equivalents under IP), but mail is almost by definition a 
> non-interactive medium. Who cares if the receiving machine is up right 
> now? And the receiver might, or might not read the mail. The receiver 
> also might, or might not open the mail. The two are not the same, and 
> software can actually only detect the latter. And I don't really care if 
> the mail was opened or not. If the topic is something I need a response 
> on before a certain point in time, I'll either write in my mail that I 
> need that response, and have a default assumption that I believe 
> everyone can live with in case I do not get a reply, or else I use 
> another medium for my communication.
> 
> X.400, had it ever been put to the scale mail on the Internet is at, I 
> doubt it would even have worked. You'd all be very happy with no spam, 
> and no other mail as well, I'm sure...
> 
> > In other words, with OSI, they built a "cadillac" with plenty of
> > functionality but at the end of the day, people only needed a much
> > simpler Lada/Yugo.
> 
> OSI built a M2 Bradley, but people actually wanted a sports car...
> 
> > Another aspect to consider: it was not until early 1990s that the
> > internet became "commercial".  Before that, it was seen more as an
> > educational/military network not suited for widespread commercial
> > deployment.
> >
> > Once the internet was unleashed to the masses, it of course became a no
> > brainer that it should become the de-facto standard. But before that, it
> > wasn't so obvious. (even though IP could be used internally).
> 
> Agreed that it wasn't until 1992 that internet became "commercial". 
> However, at that point, there was really no alternative around anymore. 
> So it was just as much a no-brainer from the point of view that nothing 
> else could offer what was being done with TCP/IP. But it was only after 
> it became "commercial" that other companies and institutions decided to 
> recognize this fact, and everyone decided to give up on pushing OSI.
> 
> And yes, it was a research network before then, and the military cloned 
> it for their own purposes. It was definitely suited for widespread use, 
> but regulations prohibited this from being exploited for a long time.
> 
> 	Johnny
> -- 
> 
> Johnny Billquist                  || "I'm on a bus
>                                    ||  on a psychedelic trip
> email: bqt at softjar.se             ||  Reading murder books
> pdp is alive!                     ||  tryin' to stay hip" - B. Idol

Johnny,

I agree with you that X.400 was flawed, but the some of the history was not as reported.

X.400 had many issues, but as one who was in the industry before email had general traction, it was easy to see the tipping point: business cards. SMTP-style addressing gained traction when new collections of cards overwhelmingly used SMTP-style addressing, rather than X.400, despite the existence of gateways between the two addressing regimes.

As to the "non-commercial" restriction on the ARPAnet, it was not as cited. At all times, the limitation was that non-governmental, commercial traffic could not traverse the government-funded backbone network. The Internet protocol stack was always published and publicly usable. Indeed, the history is that there were several regional IP networks (with interconnecting connections) before the restriction was lifted. The lifting of the commercial use restriction coincided with the removal of government funding for the general backbone coincident with advent of common carrier IP backbone services.

If one were to open up the DECnet protocol suite, one could add encryption at various levels (IBM's SNA has the capability; I thought it would have been good for DECnet back in the 1980's). One could also implement higher-DECnet functions above an IP-backbone. One shortcoming of TCP implementations is the lack of what is normally classified as "Session Control". In TCP-based protocols, Session Control (e.g., authentication) is left to the higher-level protocol and receiving server processes, which far too often has lead to exploitable security holes. 

- Bob Gezelter, http://www.rlgsc.com