From: Dan Cross
Sent: Monday, January 20, 2014 6:37 PM
On Mon, Jan 20, 2014 at 12:20 PM, Al Kossow <aek at bitsavers.org> wrote:
> Just throwing this out to see what other people
think.
I, for one, think that it's a great idea.
> I suspect we're at the tail end of the usage
life of devices that don't
> speak IP. I'm mostly thinking about networking devices 80's >>
00's
> So, what needs to be preserved? How much of this
does CHM need to do?
> Is any other collecting institution already covering this? How much is
> within scope?
Within scope? All of it. ;-)
> We've been having curatorial discussions about
this for years inside
> CHM, and have been doing some directed collecting since before the big
> exhibition was locked down in 2010, but it is a big topic and there
> were a lot of evolutionary dead-ends.
As someone else pointed out, many of the dead ends are important for what
they taught us about what did *not* work. (Not unlike EDEs in palaeontology :-)
> What got me thinking about this is I've been
working with someone who
> has storage units full of mid-80's IBM SNA stuff and it's taken me
> months to scan a fraction of it. I know there are huge swaths of
> telephony and networking that I've never even looked at. It's pretty
> overwhelming, actually, to get my head around from the software side.
There are telephony museums. Here in Seattle, the Museum of Communications
is a good example. Located in the unused upper floors of a working central
office, they have working examples of switches from the very early stuff to
a 1ESS, and documentation to go with it.
Non-circuit switched networking is a lot less well represented.
My short list would include any available materials
related to the
following, in addition to ARPAnet and Internet history:
RSCS:
The obvious IBM stuff, but also things like URep and
JNet; both of which
seem to have disappeared entirely.
Some of the BITNET history is interesting and has had lasting impact.
I was at the SHARE presentation at which BITNET was presented to the world
at large, where it was touted as an alternative to both the ARPANET and the
UUCP-based CSnet for universities that wanted to provide e-mail and file
sharing to their users. ("Because It's Time" was the credo of the project,
enshrined in the name.) As Lynn Wheeler is (perhaps overly) fond of
pointing out in various Usenet groups, the IBM-internal VNET used the BITNET
software to good advantage.
DECnet:
We got things like ISIS from DECnet.
Xerox XNS:
This was really influential; even implemented in BSD.
XNS grew out of PARC Universal Packets, of course, and PUP had a great deal
of influence on the implementation of TCP/IP. (At least, that was the view
I was provided with in Silicon Valley.) PUP was implemented in TOPS-20 and
WAITS for the MEIS ("Massbus Ethernet Interface System") at Stanford, and in
the terminal servers and routers based on SUN boards (also used in the SUN-1
workstation), before TCP/IP was available.
I think everyone knows what came out of the MEIS and that other gear.
The ISO OSI protocols:
These were also really influential, even if they never had much direct
impact.
I'd say that X.25, at the very least, had a large--and continuing--impact.
Chaosnet:
An oft overlooked footnote in networking history; this stuff even made it
into Lisp machines.
I think "even made it into" rather misses the point, since as far as I can
tell it was *created* for the purpose of connecting CADRs and PDP-10s together.
Rich
Rich Alderson
Vintage Computing Sr. Systems Engineer
Vulcan, Inc.
505 5th Avenue S, Suite 900
Seattle, WA 98104
mailto:RichA at
vulcan.com
mailto:RichA at
LivingComputerMuseum.org
http://www.LivingComputerMuseum.org/