🚀 go-pugleaf

Peter Flass <peter...@yahoo.com> writes:
> Everyone was trying to get into everyone else's business, AT&T and Xerox
> into computers, IBM into copiers and communications (SBS and Rolm), Xerox
> into communications (Western Union), Kodak into copiers, etc.

other folklore ... guy doing M&A didn't bother to check ROLM's books
they quarter that they were acquired ... turns out it was the quarter
ROLM went into the RED.

HSDT was doing T1 and faster. IBM "officially" only had 2701 that did T1
... so I had to deal with boxes from other vendors. 2701 was no longer
produced and the existing ones were really getting old.

I was told that to continue with HSDT I had to have at least some IBM
content. The only thing I could find was that FSD was producing T1
ZIRPEL cards for the Series/1 (replacing some number of gov. 2701 T1 on
special bids). I was told (not longer after IBM acquired ROLM) only
thing I could find were the S/1 ZIRPEL ... however ROLM was all data
general, but after the IBM acquisition, ROLM put in order for a year's
manufacturing of series/1. The person that had been running ROLM
datacenter I had worked with in their prior life at IBM ... so in order
to get some S/1, I had to do some horse trading (part of continuing
HSDT).

Later, some IBM people brought in to look at issues at ROLM. One was it
was taking more than 24hrs to load new test kernels into the data
general boxes over 56kbits/sec links. I was asked to help them with
upgrading to T1 (1.5mbytes/sec) links ... to significantly reduce the
test/debug cycles (nearly 30times faster to load new test/debug kernel)
from 24+hrs to less than hour.

For some of HSDT, had equipment being built on the other side of the
Pacific. The friday before a visit, somebody in Raleigh communication
group sent out an announcement for a new internal discussion group on
high-speed communication with the following definitionw

low-speed               <9.6kbits
medium-speed            19.2kbits
high-speed              56kbits
very high-speed         1.5mbits

Monday morning on the wall of a conference room on the other
side of the pacific

low-speed               <20mbits
medium-speed            100mbits
high-speed              200-300mbits
very high-speed         >600mbits

...

the communication group had prepared a presentation for the corporate
executive committee that customers weren't interested in T1 for another
6-8 yrs. They had used customer data about 37x5 "fat pipes" ... number
of parallel 56kbit links operated as single logical link dropped to
zero at 5 or 6 links. What they didn't bother to include was that
typical telco tariff for T1 link was about the same as 5 or 6 56kbit
links ... customers just switched to non-IBM boxes supporting full
T1 when it got to that point (trivial survey found 200 customers
with T1 links supported by non-IBM boxes).

Later the communication group was somewhat forced into coming out with
box that sort of supported T1 terrestrial ... 3737 with lots of 68k
processors and boat loads of memory ... that simulated local
channel-to-channel adaptor (CTCA) ... immediately ACK'ing packet as soon
as the local box received it (because VTAM/SNA couldn't handle T1
terrestrial latency) ... and then using non-SNA between it and the
remote boxes ... even at that, max a 3737 could handle (partially
because of all the simualtion overhead) was 2mbits/sec
aggregate. full-duplex US T1 is 3mbits/sec aggregate (1.5mbits/sec in
each direction) and full-duplex EU T1 is 4mbites/sec aggregate
(2mbits/sec in each direction)