Conference 7.286::digital

    In the early to mid '70s there was ARPA net. Digital (Stu Wecker)
    began documenting the DECnet protocols DDCMP and NSP sometime in the
    mid 70s. We implemented ANF-10 (a variant of DECnet), Advanced Network
    Facility on the DECsystem-10 in the mid 70s. When it worked ANF-10
    provided some pretty powerful capabilities including routing and
    sethost, I believe. There was an early and limited implementation of
    DECnet on the PDP-11s providing task-to-task as I recall. Then there
    was an effort to integrate all of these efforts on one DECnet program
    with one set of protocols. This was circa '77 and '78. There were a lot
    of slips and resets due to differing implementations of an imprecisely
    defined and changing set of network protocols. The "Design Review
    Group" or DRG was put in place under Tony Lauch to stabilize the
    protocols and get them under revision control.
    
    Meanwhile, IBM had announced SNA and was laboring to implement same.
    
    Most of us felt that Digital was the industry leader in that era.
    
    I'll bet  John Adams would be happy to fill in any gaps in this note
    and correct my errors.

    What is really annoying is SUN Microsystem's ad that talks about them
    INVENTING computer networking. What about truth in advertising?

    In pretty much the same way that Thomas Edison "invented" the light
    bulb or the Movie Camera. Nothing is new.

    .2
    
    It's not a game of truth in advertising-- it's MARKETING!
    

RE: .2

	Even so far as to re-use a slogan we used to use.

	"The Network is the System"

	My jaw dropped upon seeing that.

						mike

    Re: last few
    
    Add to the FUD. Even Sun admit that their network is a dog!
    
    assuming their ads in the Wall Street Journal are the same as those in
    the trade rags.

    Note they don't say networking, but network computing.  How much of the
    OSF DCE environment is traceable to Sun's technology (e.g., NFS)? 
    Enough for them to make a partially credible claim?  Although I suspect
    that others, including Digital, can make equally credible claims.

> How much of the OSF DCE environment is traceable to Sun's technology
> (e.g., NFS)? Enough for them to make a partially credible claim?  Although
> I suspect that others, including Digital, can make equally credible claims.

None of the OSF DCE environment can be traced to Sun's technology.  (NFS
isn't part of DCE, it's part of OSF/1.)  Digital provided much of the core
technology for DCE:

	Time service (from DTSS)
	Cell directory service (from DNS)
	RPC (with Apollo ... er, H-P)
	Threads (CMA became POSIX pthreads became DCE threads)

Four out of seven isn't bad, but that's ancient history.  As the joke puts
it, "what have you done for me lately?"

___Pete

    I used to work for an OEM from '73 to '87 and did lots of work with
    RSX11M networking and followed VMS and server market later very closely.
    
    '75, when RSX came out, it had good real time support features (at that 
    time), but no real networking. I did a bisync driver for it during that year
    for a railway reservation system using an RSX pre-release kit.
    
    DECnet Phase I was a difficult product, and we didn't really want to 
    touch it. It basically had a messy support structure for networking,
    task-to-task support, DDCMP and some management functions.
    
    DECnet Phase II was an excellent product with a simple and easy to
    handle communications executive, which basically was a support for
    protocol layering � la OSI before OSI model was there. It was '77-
    '78 timeframe, I don't remember more exactly.
    We used it as a basis for a remote power station and control system
    network.
    There was no routing in the first release, so we had to add that and 
    support for HDLC. Later, Digital was one of the firsts to have real routing 
    as well as part of a commercial product.
    
    --Ritva
    
    To my opinion, Digital was still in a good position to grow market in
    networking in early '80s. It was killed by stubborn protectionism.
    Networkin is a game, where everybody must co-operate, not to gather in
    corners and turn their back to others.

    Well, I remember attending the first DECnet/RSX course in Europe in
    Geneva fall '75. That was RSX-11M, of course (RSX-11D was released
    years earlier).

    In the same time frame that is being discussed, Digital implemented 
    a country wide network for the New Zeland Department Of Health which
    consisted of two dual processor PDP-10 sites running TOPS-10 (Auckland 
    and Christchurch), and a PDP-11 (70+, I think) running RSX-11D at every 
    hospital facility in New Zealand.  There was lots of custom code.  
    I still remember how proud the team was when they were able to pass 
    one error free message per minute between all nodes on the network.  
    I believe this was the first wide-scale use of "DECnet" starting with 
    pre-Phase I products, and evolving to incorporate Phase I (1975-1977), 
    Phase II (1977-1979), Phase III (1979-1983) and Phase IV (1983 ->) 
    DECnet implementations.  
    
    Oficially released Phase I products (1976) were limited to PDP-11 
    family machines running RSX, and were rude, crude, and indifferent.
    
    There is a good, if brief, history in the September 86 issue of 
    Digital Technical Journal.  

    .....in other words, well before SUN Microsystems ever existed. QED.
    

Since I may be the sole survivor of the original DECnet implementers, I 
thought I'd stick my two cents in.

The first round of DECnet products (Phase I) started development from about 
October, 1974.  Some of the wait came from the fact that several of us 
were working on RSX-11M until then, when it shipped V1.  In particular, 
John Gilbert and Roger Heinen moved over to DECnet, as did I.

Stu Wecker had been writing NSP (for logical links), DDCMP 
(for data links) and MOP (for maintenance/bootstrapping).  The group I 
was with was doing DECnet for RSX-11M/S, RSX-11D, and RT-11.  [Re: .10 - 
RSX-11D was _not_ years earlier than 11M].  I worked on FAL (File Access 
Listener) and the downline system and task loading to support RSX-11S.  
Many of the utilities were practically identical between  RSX-11D and 
RSX-11M/S.

We shipped, if I remember correctly, in April, 1975.  I happened to go 
on loan to a customer project until December of that year, the idea 
being to see how the stuff was actually used, so we'd know what to do 
next or what to fix.

I came back to the Engineering group at that time, and a couple of us 
produced some updates.  In the fall of '76, I joined the VMS development 
group, to work on DECnet-VAX, with Jim Krycka.  While I won't go into 
details, there were some interesting times then: the RSX DECnet 
developers were also the keepers of the DECnet protocols.  Some of us in 
the other groups (including 10s/20s, RSTS, RTS-8 [might have the name 
wrong]  would get annoyed when the protocols would be changed unilaterally.
I led the yelling and screaming, as a result of which, Larry Portner 
locked the chief implementers plus some architects in a room for two 
weeks and told us to come to agreements that would freeze the protocols 
for a year and a half.  He also commissioned the DECnet Review Group 
(DRG), to be chaired initially by Bill Keating, as his first job at 
Digital; Tony Lauck eventually took the group over.

The result was  the Phase II products, which had the mechanisms for 
routing in the protocols, but no implementation until Phase III.

I'll reminisce some more, in case anyone is actually interested. ;-)

- Scott

Scott:

> RTS-8 [might have the name wrong]

  You've got it right.

                                   Atlant

    re .13:
    
�[Re: .10 - 
�RSX-11D was _not_ years earlier than 11M].
    
    What I meant was the operating system, not DECnet. If I remember
    correctly, RSX-11D itself was released somewhere around '72-'73.
    

    re: .13
    
    I was the tech. writer working on those protocol documents, and I 
    remember being in shut up in that room, keeping track of all the
    decisions being made on network management protocols. It was more
    than 2 weeks. Also, George Cohen did the NSP design -- when I was
    there, at least.
    
     Liz                                   

    More - more. This is fascinating stuff
    
    Alan
    

>    More - more. This is fascinating stuff
    
I hate `me too' posts, but in this case - me too!

Chris.

I came in 1976 as a tech writer for DECnet on RSTS/E, and
the clearest thing I remember was that at some point in the
Phase II implementation, they called a screeching halt
and started over.  (That may be the where the guy in RSX remembers
being thrown in a room by Larry Portner.)

What I remember is standing in the Mill, totally confused, after
having produced a first draft which Jim Miller (DECnet/E developer
at the time) was not at all happy with (well, when they'e that
upset, at least you know they care) and hearing George Plowman 
tell the milling, muttering crowd something to the 
effect of, well, we've discovered we don't know quite what we're 
doing here, so we're going to rethink some aspects of these protocols.  
We're not real happy with this, but that's the way it is.

I thought, "Well, this is a strange place, but that *really* took guts", 
both from his standpoint as a manager telling his management it needed 
to be done, and then telling the troops it was going to be done, as well.

That's the most dramatic moment I remember.

Gee, this is like ol' home week.

Howdy, Scott!

Tom
(No, I wasn't a DECnet developer, just an engineering support grunt for VMS 
V1 - V2.)

    It's been a long time... I started with DEC in the MIll in Feb 1976 as
    an 'In House Field Service Engineer' (Not to be confused with a
    Multi-Vendor Customer Service Support Representative  ;-) ).
    
    I remember building the PDP11 DECnet LAB on 5-5. I believe the first
    node was officially named 'TheArk'. This was because it had 'two of
    everything'. 2, 1Meg DMR11s, 2, 56k DMR11s, 2, DUP11's etc. The reason
    that there were 2 of everything was because we had agreement between
    the development/operations/maintainence groups that no one would log a
    service call unless the software failed on 'Only One Controller'.
    
    As I recall TheArk Spawned a biblical theme and other nodes like Noah,
    Adam and Eve Followed along. 
    
    We also had one of the most interesting patch panels I have ever seen 
    located right smack in the middle of the room. I say interesting
    because it was actually one of the 'cleanest' patch panels I have ever
    seen. (Not to be confused with the Terminal Patch Panel in the main Lab
    which was one of the worst I have ever seen.)
    
    For you PDP11 buffs out there.... As I recall TheArk (and a few other
    systems had Interupt Vectors into the 700's. To put that into
    perspective the 'default' Yellow Zone stack violation was 400! I don't
    believe I ever saw an outside customer with a Unibus Interupt Vector
    above 400. (All PDP11 digits are Octal... Of Course!!  :-)  )
    
    Jim

	Sounds like a bunch of bus (geesh I used to hate these things) repeaters
to me!!

RE: <<< Note 4055.21 by CHOWDA::FAHEY "Are we having 'FUN' yet?" >>>
    
>    As I recall TheArk Spawned a biblical theme and other nodes like Noah,
>    Adam and Eve Followed along. 

Good memories. I used ADAM (an 11/70) and EVE (an 11/780) when I joined the
company, working down on 5-2.
						-- Tom