[ih] Resource sharing (was: Re: First file transfer on ARPANET)

John Day jeanjour at comcast.net
Sun Dec 23 10:47:58 PST 2012


Klensin mentioned Illinois, so I was listing only what had been going 
on at Illinois. It was not intended to imply it was the only place 
that sort of thing was going on.  It wasn't.  There was considerable 
other distributed computing work going on elsewhere in those early 
days.  The Irvine ring being one of the big ones.  As I said, I think 
those today like to think that all we were doing was remote login to 
make themselves feel good about only doing the web, which differs 
from remote login more in form than substance.

>On 12/23/2012 6:11 AM, John Klensin wrote:
>>  IMO, the shared-resource
>>activities that John Day and I are describing were somewhat different,
>>involving having the local software resolve user interface issues,
>>determine what resources were out there that could best meet the
>>requirements of the particular user and application, and then seeking
>>those resources out, using them, and making them look local.
>...
>>  All of that, as John points
>>out, pretty much requires an application-oriented operating system
>
>National Software Works?

NSW was another major attempt.  My impression (which could be wrong) 
was that a) it was far ahead of what the hardware could support at 
the time, and b)  the design needed another iteration.  It took a 
not-quite-right approach to an extreme.  A more naunced design would 
have been better.  But it was a step in the right direction.

>
>Remote Procedure Call?

I have never been impressed with this.  First because RPC was always 
synchronous, i.e. blocking.  (To me this is distributed in name 
only.) Computer scientists seem to be afraid of asynchrony. Second, 
because it isn't a *procedure* call at all. At best it has the 
semantics of a Fortran function call, not even a subroutine call, let 
alone a procedure call.  At the time I would tease the RPC zealots 
(and they were) with either it was like co-routines in Cobol or it 
was an elaborate way to encode one bit (request/response).  ;-)

(I really detest the way we inflate terminology to make ourselves look good.)

Similarly, when I have asked what was the big deal about peer-to-peer 
[sic], a consistent answer I get as the speaker's eyes fog over and 
he says reverently, "A Host can be both a client and a server at the 
same time!"  Of course he is a bit crestfallen when I point out this 
has been true since the first IMP went in.  When you boil it down it 
turns out to be mainly a different name resolution facility, with a 
technique for hogging bandwidth that makes it look like a DDOS attack.

>
>>As far as I know, very little has happened with that area of work
>>since our efforts,
>
>SafeTCL?

Huh?

>
>>But I think we do ourselves and the community a disservice by
>>confusing those resource-sharing ideas with remote login, remote
>>access,

>+1
>

Ditto.

>d/
>--
>  Dave Crocker
>  Brandenburg InternetWorking
>  bbiw.net



More information about the Internet-history mailing list