[ih] NCP and TCP implementations

the keyboard of geoff goodfellow geoff at iconia.com
Wed Jul 22 09:54:27 PDT 2020


thank you tony!  this is EXACTLY the "situation" yours truly was trying to
recall... :D

On Wed, Jul 22, 2020 at 5:41 AM Tony Finch <dot at dotat.at> wrote:

> the keyboard of geoff goodfellow via Internet-history
> <internet-history at eli...:
>
> > can anyone recall WHO and/or WHERE it was that had their NCP (don't think
> > it was TCP) implementation apparently running in "user space" as a job or
> > process (as opposed to as part of the OS/Kernel) and "complained" how
> much
> > it was co$ting them/their departments budget to be connected to the
> ARPANET?
>
> Sounds like MULTICS. I read Mike Padlipsky's Elements of Networking Style
> a few months ago, and I'm fairly sure he had some choice words about
> Multics and ARPANET billing, probably the exact incident mentioned in
> RFC425? Despite the lack of index I found it at the end of chapter 4:
>
> > Again, the Multics NCP, in its reluctance to "waste interrupt time",
> > employed a specific process wake-up to handle the [ECO] command. When
> > the network bills tripled one week, the cause soon became apparent - all
> > those ECOs. Now even though it is not germane to detail just how much
> > trouble this one caused before it was finally resolved, the point for
> > present purposes should be clear: Even after protocols are properly
> > designed, documented, and implemented, problems can be engendered by
> > their use in ways not concieved of by the designers.
>
> Well that advice has aged well :-)
>
> Also, https://multicians.org/mx-net.html - section 2.2.1 NCP
>
> > The Network Control Program daemon originally ran in Ring 1 on the 645,
> > resulting in incredible amounts of overhead and $ charges to the ARPANET
> > group for running the NCP. This stemmed from the fact that the ARPANET
> > was considered a research project at the time and not part of the
> > System. Later, as the ARPANET became more accepted, execution of the NCP
> > daemon functions was moved into Ring 0, and the greatest consumer of
> > daemon resources in the NCP, the handling of ALLOC (data space
> > allocation) messages, was done at interrupt time in the IMP DIM (in a
> > violation of what would later be understood as strict protocol layering,
> > but, heck, it saved $$ AND improved performance).
>
> Tony.
> --
> f.anthony.n.finch  <dot at dotat.at>  http://dotat.at/
> South Utsire: Northwesterly 4 or 5, occasionally 6 in southeast at first,
> becoming variable 3 or 4 later. Slight or moderate. Showers. Good.
>
>

-- 
Geoff.Goodfellow at iconia.com
living as The Truth is True



More information about the Internet-history mailing list