[email protected] (Tony Harminc) writes:
> For many years (decades, actually) there have been other products (IBM
> and non-IBM) that talk the NJE protocols. Most notably, IBM's RSCS on
> VM uses an overlapping subset of the same protocol, and is
> interoperable. There have been NJE implementations for UNIX and other
> operating systems over the years, long predating NJE over TCP/IP. One
> popular product in the 1980s was JNET, from Joiner Associates, which
> ran on the DEC VAX.

NJE originated as HASP networking at customer (source col. fields 68-71
use to carry "TUCC"). It defined nodes in the unused entries in the 255
table of psuedo (unit record) devices ... so number of nodes was
typically limited to 160-180. It also had designed that it tossed
traffic when it didn't recognize the original or destination nodes. The
internal network had relatively early passed 255 nodes (the internal
network was larger than arpanet/internet from just about the beginning
until sometime late '85 or early '86) some past posts
http://www.garlic.com/~lynn/subnetwork.html#internalnet

and so NJE nodes couldn't be trusted except as edge nodes (since they
were prone to tossing traffic).

NJE also was relatively dirty implementation ... intermixing network and
job control fields ... as a result NJE systems at different release
levels exchanging traffic had habit of crashing the host (MVS).

By comparison, RSCS was very clean layered network design ... done by coworker
at the science center ... some past science center posts
http://www.garlic.com/~lynn/subtopic.html#545tech
reference
http://en.wikipedia.org/wiki/Edson_Hendricks

it didn't have the number of nodes limitations, didn't toss traffic when
it didn't understand the origin ... and it's clean layered design was
straight-forward to have drivers that understood other protocols
... useful in supporting JES/MVS as edge nodes. In fact, as the number
of nodes exploded around the world ... and all JES/MVS systems couldn't
be kept at the same release levels ... involving JES/MVS systems in one
part of the world crashing MVS systems in other parts of the world ... a
library of RSCS NJE drivers grew up that translated NJE header fields to
canonical form ... and then a specific RSCS NJE driver would be used to
convert any NJE fields to the appropriate format required by the NJE
system at the other end of the link (as countermeasure to traffic
crashing the host MVS system) ... there was the famous case of traffic
from san jose (gpd) mvs system crashing mvs system in hursley ... and it
being blamed on the hursley vm/rscs people for not having the correct
vm/rscs nje driver started (to keep mvs from crashing).

VM/RSCS was also the basis for IBM-sponsored BITNET for educational
institutions (where this ibm-mailing list originated)
http://en.wikipedia.org/wiki/BITNET
some past posts
http://www.garlic.com/~lynn/subnetwork.html#bitnet

internal politics eventually stopped the shipping of native RSCS
drivers, just the NJE drivers ... even though they continued to be used
on the internal network, in part because they werre more efficient ahd
had much higher throughput.
then later, internal politics forced the move of the internal network to
SNA/VTAM, at a time when it would have been much more cost effective to
have moved to tcp/ip ... which BITNET did. Some old internal
network related email
http://www.garlic.com/~lynn/lhwemail.html#vnet

including this reference to a little of the internal politics behind
moving to sna/vtam
http://www.garlic.com/~lynn/2006x.html#email870302
http://www.garlic.com/~lynn/20011.html#email870306

it was in the same time period that they were spreading misinformation
internally that the NSFNET backbone (precursor to modern internet) could
be done on sna/vtam ... somebody had collected a lot of the internal
misinformation and forwarded it to us ... heavily snipped and redacted
to protect the guilty 
http://www.garlic.com/~lynn/2006w.html#email870109

other old NSF related email
http://www.garlic.com/~lynn/lhwemail.html#nsfnet

past posts mentioning hasp, jes, and/or nje
http://www.garlic.com/~lynn/submain.html#hasp

one of the other early NJE issues was they couldn't find a valid
business case to ship the product ... the standard internal process
resulted in price much higher than customers would pay i.e. IBM still
adjusting to having to charge for software (after the unbundling
announcement).  Eventually they came up with the hack that they would
announce a joint vm/rscs and JES2/NJE product ... business case where
they combined the costs & revenue for the two products (pricing vm/rsc
the same as jes2/nje ... where vm/rscs revenue was used to cover the
jes2/nje costs... and eventually eliminating shipping native vm/rscs
drivers helped with the facade).

-- 
virtualization experience starting Jan1968, online at home since Mar1970

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to