wrote:
> Good point. Most of the really gloomy geek outlooks come from the Big
> Enterprise Systems quarter. You've said before that these are the computers
> that do the world's work. And as you point out, embedded people think the
> work is done by machinery, and LAN people think the world is a network.
> If your area is where the big problems really lie, maybe we need a clearer
> idea of what these systems do, and what they do NOT do. Does power
> generation and distribution rely on them? Communications systems? Assembly
> lines? Train switching? Water treatment plants? I'd really like to know.
I may expand on this in a future WRP (this isn't a shill for WRP
subscriptions so much as pointing out that some of the WRP ramblin's are
reprints from articles here, essentially the same natterings, just more
of it. )
Enterprise systems, the big honkin' iron, really got rolling in the
1960s. The model then was a .5 to 1 MIPS box, a quarter to a meg of
memory, 500 nanosecond was considered fast, with a half dozen
or fewer channels and perhaps a half gig of rotating removeable storage.
Printer did 1,000 lines per minute (make it 15 pages/minute).
*********** skim zone **************
The first terminals showed up in this timeframe. Mostly these were
typewriters hooked up at 10-15 cps, make it 300 or 600 baud.
Interestingly, most peripherals were unbuffered. A characteristic of the
IBM S/360 channel was that it could sustain single byte transfers
directly into the host's memory without interferring with CPU
operations. Byte multiplexor channels handled single and two or three
byte transfers very efficiently. The Selector channel did the same for
blocked data from faster record oriented devices such as tapes, disks,
and drums (or fixed head store).
At the dawn of time, 1964, until the early 1970's, the only game in
town were mainframes. Large enterprises computerized by building systems
using chisels and hammers. Databases showed up toward the end of the
1960's, as DL/I, IMS, CICS/DB, VSAM, MarkIV. I've met and worked with
the architect of VSAM, he likes Startrek.
A high performance database was hand built out of assembly language and
DBAM or EXCP. I was there, I used these.
In addition to IBM, RCA's Spectra, GE, Honeywell, Amdahl, Magnusson,
Itel, Control Data Corporation and others built mainframes.
In the 1970's, the second leg of enterprise systems forked out,
Minicomputers. Data General, Digital, Apollo, Wang, Series/1, System 3,
there were others.
Every year, the inventory of code increased. I've never done COBOL.
I've held the hands of COBOL programmers and helped them debug their
applications. In some circumstances, machine language mechanics can
debug and fix code faster than COBOL'ers can. I did a year of Fortran
IV, followed by PL/I. I've written fairly large systems in PL/I and even
after 15 years on C (and recently C++) can still code PL/I faster and
safer than C.
C was a step back and even C++ does not have the expressiveness and power
of PL/I, but I digress.
***************** end skim zone **************
1960's things start. 1970's systems get larger but some mini's appear.
In the 1980's a bad thing happens. Mainframes get very large and
inexpensive. The mainframe applications get so large that they
essentially define the organization. Sometime in the 1980's, IBM finally
makes mainframes reliable.
The result is dangerous. The things are cheap, reliable, powerful enough
to run entire organizations, and already host the aggregate data and
software logic of 20 years of corporate history.
Add to the mix the 1980's PeeCees and Un*x; the new talent is siphoned
off. I took the last mainframe computer science graduate level class
offered at GWU. This was about 1985. Except for a few schools, there
have been no CS graduates who understand mainframes, they all want to
click on VB or write the next Quake.
***************** skim zone **************
I was an early adopter of PeeCees. In 1975 my friend, Peter, bought an
original Altair, we put an 11 slot processor technology backplane in it,
Northstar 5.25 floppy, ran a Zilog 4.0 mHz Z80A; he designed and
wirewrapped special purpose cards (which is how I know about 74LSxxx
devices, 1x8s, decoders, NAND gates, etc.), I wrote code in Z80 assembler
which we burned into 4K EPROMs and sold with his boards. In 1983, we
connected his much modified Altair to my brand new IBM PC, wrote an async
port driver program in IBM BASICA and Northstar Basic and transferred his
files to 180KB IBM 5.25 floppy from his 80KB Northstar. We could have
named it Laplink but we didn't think it was worth anything.
(This is why my understanding of microelectronics ends about 1980. Note:
the IBM PC is really a 1970's type machine. Everything in it is off the
shelf, well tested.)
He junked the system shortly afterward and bought a Leading Edge PC.
***************** end skim zone **************
What do these mainframes do? They do the heavy lifting of commerce.
Utterly reliable, fast enough to drive all the corporate data through a
single point, dirt cheap to run and maintain. Mainframes are the secret
weapon of the Fortune 5,000. There are 50,000 IBM style mainframes.
These are supported by 400,000 AS/400s and millions of other
midicomputers.
Banking, insurance, manufacturing, reservations, chemicals, electronics,
communications, any business that was big in the 1960's and 1970's and
has a large records problem depends on its mainframes. Given that,
direct satellite TV and perhaps Starbucks are safe.
I recall at one SHARE, bumping into a fellow at SCIDS, he was a mainframe
systems internal expert from Playtex, the bra people. If mainframes die,
Anna Nicole Smith may be left without support.
Single point of failure. The ideal target. At the beginning, banks and
other firms used to proudly display their mainframes in glass rooms at
street level. I don't think there has been a hit on a mainframe but the
risk is so great that now you will never see a mainframe, unless you work
on them.
Mainframes are hidden away behind locked doors. Usually there is no
external indication that there is a mainframe at the location. There
might be a sign reading XYZ Corp Annex-2 but more likely it's just a
street number.
Most people working at firms with mainframes do not know that the system
exists. It's a secret.
So now, with an inventory of 50,000 mainframes and 263 Days, 6,335 Hours,
we don't have time to find and fix all the source and data.
I believe that there are Fortune 5,000 firms that have not yet started.
Perhaps the CIO plans to retire, perhaps he doesn't know, perhaps a
consulting firm as promised that they will implement SAP or run the code
through a factory. For whatever reason, they have not done the work and
will fail on roll over.
I believe that there are Fortune 500 firms that are done. I expected
BankBoston to announce completion a few months ago, perhaps their
acquistion by Fleet muddied the water. I know of two other Fortune
500 firms that are done, remediated, tested, independently certified. I
know when they started and how much effort they applied. I don't know
why they haven't announced the "Great News!" Perhaps they are retesting
or perhaps my info isn't as good as I thought.
Whatever. Here are the facts. Some large percentage of the systems
running on the 50,000 mainframes and several millions of mini's (or
midi's) will fail. The nature of the failures will be such that they
cannot be fixed, patched, restarted in 3-6 hours. In some cases, these
systems will never be fixed, the owning organization will fail and there
will not be the resources to execute the repairs.
Experts in embeddeds, LANs, PeeCees, Telcos, power generation, whatever,
if your area is fine, good for you. I'm telling you that the enterprise
systems are not OK. I expect a major bank to fall over, same with a
major insurance company, brokerage house, chemical manufacturer, etc. I
am not optimistic about the prospects of any sector. Something bad is
about to happen.
Perhaps there is enough parallelism that even when we see multiple
failures, as happened with S&L's, we will be able to jump to the
surviving firms.
Perhaps when the economy tanks as a result of the widespread failures
that we will be able to restart it, as we weren't able to do in
1929-1939.
Who knows, I don't. I'm glad that there are big-brains out there who are
certain that everything will be fine. Please, denialists, keep flooding
c.s.y2k with your "Great News!", I need the laughs.
Something bad is about to happen.
cory hamasaki 263 Days, http://www.kiyoinc.com/current.html
------------------------------------------------------------------------------
You can view this message and the related discussion by following this link:
http://www.dejanews.com/=zzz_maf/dnquery.xp?search=thread&svcclass=dnserver&
recnum=%3c7kepWhCNP4qd-pn2-5fIrGHBgFdqd@localhost%3e%231/2
We hope to see
you soon at Deja News, the discussion network.
http://www.dejanews.com/=zzz_maf/
Date: Tue, 27 Apr 1999 16:33:13 -0700 (PDT)
From: cabeal@efn.org
Subject: 1 2 3 4 5 6 7 8 9 10
This is from Cho Qosh, who spoke at the Oakland Y2k conference in March,
forwarded on by Gary Gach.
It is a synopsis of one message that the Hopi Elders have for us now as we
face the uncertainties of these times. I find the guidelines extremely
useful.
Best,
Cynthia
1. Where do you live (not just geographically)?
2. What is it that you do?
3. How are your relationships?
4. Are you in right relation with the Earth?
5. Where is your water?
6. Know your garden (and nature around you).
7. Speak your truth; it is time now.
8. Be good to each other.
9. Don't look outside yourself for the leader.
10. This could be a good time.