SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Discuss Year 2000 Issues -- Ignore unavailable to you. Want to Upgrade?


To: Bill Ounce who wrote (885)1/14/1998 10:41:00 AM
From: Bill Ounce  Respond to of 9818
 
NY Times FAA article, comp.software.year-2000 review

Below is Tony's very informative posting
======================================================================

Newsgroups: comp.software.year-2000
Subject: Media Sighting: Ny Times on the FAA
Date: Wed, 14 Jan 1998 05:39:15 GMT

nytimes.com

[Confirming what many had stated in this newsgroup. I decide to post
the entire article as it's very interesting. My comments are enclosed
in square brackets.]

January 13, 1998

Year 2000 Raises Safety Risk for Air Traffic Computers

By MATTHEW L. WALD

WASHINGTON -- A set of crucial computers in the nation's air traffic
control system should not be used beyond December 1999, because they
may not operate reliably when the date rolls over to Jan. 1, 2000,
and there is no way to predict the effect on air traffic, according
to IBM, which built the computers.

[No news to this newsgroup]

But the official in charge of that system at the Federal Aviation
Administration said on Monday that "it would be an extraordinary
feat" to replace about 40 mainframe computers by then. Instead, his
agency, with the help of a retired IBM programmer and a team of
software experts, is racing to determine whether the problems can be
anticipated and eliminated before the turn of the century.

[Oh, it'll be extraordinary all right]

Computers all over the world will have difficulty dealing with Jan.
1, 2000 and beyond, because many of them record only the last two
digits of the year, and assume that the first two are "19." For those
machines, the day after Dec. 31, 1999, will be Jan. 1., 1900, not
Jan. 1, 2000.

The extent of problems with the air traffic computers is not certain,
but experts say that the 3083 mainframe model referred to in a letter
from IBM to an FAA contractor, might, for example, refuse to accept
flight plans for planes that take off on December 31, 1999, and land
on Jan. 1. That landing would be 99 years in the past, from the
computer's point of view.

[Ummm, a 3083. Weren't those built in the mid 80's? Surely Cory's
desktop MVS on a card is more powerful, has more memory and MIPS?]

"Who knows, it could do anything," said Michael Fanfalone, the
president of the the Professional Airways System Specialists the union
that represents FAA technicians. There might be no problem, he said,
but "no one knows until it's up and running and there's no way you can
take that kind of risk."

Already, FAA teams have found, deep in the computer code, a monthly
command that enables a computer to switch from one cooling pump to
another; if it is not fixed, experts say, that routine could stop
running, allowing the computers to overheat and fail if the pump
breaks down. In fact, experts say, there could be many such land
mines -- buried in millions of lines of computer code -- that could
cause failures for days, weeks or months after the new year.

[Say, looks to me like the FAA was desperately looking for some small
problem to blame on IBM. Hey you at the FAA, maybe you should get
current on the hardware. At least within this decade.]

"We're kind of worried about it," said Jack Ryan, a former FAA
manager who is now the air traffic control expert at the Air Transport
Association, the trade association of the major airlines. "I think the
FAA has the right sense of urgency, although it's a little bit late."

[Gee, I'd be worried too.]

Monte Belger, the associate administrator for Air Traffic Services,
said in an interview that the FAA should know within 90 days whether
the computers can be de-bugged. The problem is that the date
functions are not in programming languages, like Fortran or Cobol, but
in machine language -- strings of ones and zeros more basic to the
computer than even the operating system.

[90 days? 90 FREAKING DAYS FROM NOW? Why not a bloody year ago!
ASSEMBLER??? <he moaned, head in hands> Cory's right. If they had
any sense at all, which they don't they'll get the brightest and best
of the assembler heads together but no...]

The FAA does not have money in its budget to replace the 3083
computers, but could probably borrow it from other areas, by delaying
modernization of other computers or navigational equipment. The cost
is still being studied but could be as much as $200 million. In the
long run, however, cost is not as important as it might seem, because
the FAA had planned to replace the computers by 2003 and would like to
do so sooner because IBM says spare parts for the mainframes,
installed in the late 1980s, are rapidly disappearing.

[Idiots. You want a system working in less than two years or not.
And why would you wait until 2003 to replace 15 year old main frames?
Or wre you folks thinking of trying yet another client/server system
which is doomed for failure?]

The computers in question are at the 20 Air Route Traffic Control
Centers, which handle all the high-altitude, long-distance traffic in
the country. The 3083 models were once common in business and industry
but few remain in service, experts say. IBM stopped shipping them
about 10 years ago, but some of the software on the FAA models is even
older, dating from the early 1970s.

The FAA has 250 separate computer systems, most of which will require
fixes but the 3083 is the only one that IBM says can't be debugged
before 2000.

The problem, known among information technology experts as Y2K, for
the year 2000, is hardly unique to the FAA. But it is especially acute
there, because the air traffic control system demands an extremely
high level of reliability and because the FAA cannot bypass the
software built into the 3083's that has the date problem by simply
running the air-traffic programs on new computers.

[Hmmm sounds like the assembler is using 3083 unique stuff? Or maybe
the assembler is using only a specific hard drive by hard coding the
sector, track and cylinder numbers for utmost speed? Or ??? But then
I'm not a main framer.]

At IBM, David Cassano, general manager of Year 2000 Global
Initiatives, said that his company had told a few customers that no
software upgrades were available for their older computers but, he
added, "it is not the majority of systems, by a long shot."

The company's Web site includes a database that shows the Y2K status
of every computer system it has built. It can be found at:
wwwyr2k.raleigh.ibm.com.

In the FAA's case, the computers are called the "hosts," and are used
to receive data from radar scattered across thousands of square miles
and integrate the images into a mosaic. Then the hosts divide that
picture into sectors, the subdivisions that controllers use, and pass
the data on to other computers that drive the screens at the
controllers' work stations. The 3083's also receive signals from each
plane stating its identity, type of equipment, altitude and
destination and helps tag each radar blip with the appropriate data.

[Uh oh, this sounds like one of those critical systems, doesn't it?
This ones gotta work.]

In the October letter from IBM to the FAA contractor Lockheed Martin
Air Traffic Management, it said, "IBM remains convinced that the
appropriate skills and tools do not exist to conduct a complete Year
2000 test assessment" of the 3083 computers. "IBM believes it is
imperative that the F.A.A. replace the equipment" before 2000.

Belger of the FAA said that technicians and software engineers had
taken a spare machine in Atlantic City, N.J., inserted a December 1999
date and let it roll over into the new century to look for flaws. They
are not confident that this method will identify all bugs, though, he
said.

[Yup, I'm sure the folks can quickly generate lots of test data to
simulate real life operations. Kinda like debugging the star wars
system. It would've been nice if the Soviet Union would fire a coupla
hundred dummy missiles to test that system out.]

Through the '80s and '90s, his agency has canceled or missed deadlines
on a string of big software projects.

[<sigh> Has any govt bureacracy or large corporation met any software
deadline ever?]

Whether the problem can be solved by analysis and repair or only by
replacing the computers, he said, the FAA will meet the deadline.
"That deadline will not change," he said dryly.

[He sorta gets it but not quite yet.]

Raymond Long, manager of the FAA's Year 2000 Program office, which is
eight months old, said his agency had found former IBM programmers
with relevant expertise. Some of them developed tools for analyzing
the software after leaving IBM, Long said, and the FAA is using those
tools.

[8 months old, 8 MONTHS OLD. This shoulda been 8 years old.]

Long said that his personal plan for Dec. 31, 1999, was to fly an
airplane around the country to show that his office had succeeded. The
flight would be mid-evening on New Year's Eve, he said, because the
air traffic system works on Greenwich mean time, five hours ahead of
Eastern time. And he added that he would check that the software
aboard the plane had been certified as "Y2K compliant."

[And just what is a bureacrat doing with chartering a plane just to
"fly around the country."]

Others are still worried. Fanfalone of the technician's union said,
"There's only two folks at IBM who know the micro-code, and they're
both retired.

[Yikes. So one has a heart attack and the other decides to move to
Fiji.]

"The FAA pulled one of them in, to go through the code, but they
should have brought 10 software people over," he said.

[10. Not enough. Maybe 10 very, very good competent people who don't
have to attend meetings and are allowed to concentrate on thier work.
Oh, and you'd better pay them whatever it takes to keep them. And if
it means $200,000 or $400,000 per year then bloody well do it.]

And another deadline is coming on Feb. 29, 2000. Long said that most
computers are not programmed to recognize years ending in "00" as leap
years. But 2000 is a leap year.

[Gee, this is old news to us in this newsgroup.]

Ryan, of the Air Traffic Association, says the larger problem is that
no one knows what else lurks in the 30-year-old software.

"I don't know what I don't know, and that's what's very worrisome
about it," he said.

[Hey, Ryan gets it.]

Tony
----
Message posted to newsgroup and emailed.
Tony Toews, Independent Computer Consultant
The Year 2000 crisis: Will my parents or your grand parents still be receiving
their pension in January, 2000? See www.granite.ab.ca/year2000 for more info.
Microsoft Access Hints, Tips & Accounting Systems at www.granite.ab.ca/accsmstr.htm



To: Bill Ounce who wrote (885)1/14/1998 10:53:00 AM
From: Bill Ounce  Read Replies (1) | Respond to of 9818
 
NY Times FAA article: more comp.software.year-2000 followup

Below is another gem from Cory Hamasaki:
======================================================================

Newsgroups: comp.software.year-2000
From: kiyoinc@ibm.net (cory hamasaki)
Subject: Re: Media Sighting: Ny Times on the FAA
Reply-To: kiyoinc@ibm.XOUT.net (cory hamasaki)
Date: 14 Jan 98 13:04:49 GMT

In <34bc4b05.203797770@news.calgary.telusplanet.net>, ttoews@telusplanet.net (Tony Toews) writes:
>http://www.nytimes.com/library/cyber/week/011398shortage.html
>
>[Confirming what many had stated in this newsgroup. I decide to post
>the entire article as it's very interesting. My comments are enclosed
>in square brackets.]

Thanks Tony, It's much easier for me to see the article in USENET than on
the web, I fire up Netscape for Rick's chatroom.

> The extent of problems with the air traffic computers is not certain,
>but experts say that the 3083 mainframe model referred to in a letter
>from IBM to an FAA contractor, might, for example, refuse to accept
>flight plans for planes that take off on December 31, 1999, and land
>on Jan. 1. That landing would be 99 years in the past, from the
>computer's point of view.
>
>[Ummm, a 3083. Weren't those built in the mid 80's? Surely Cory's
>desktop MVS on a card is more powerful, has more memory and MIPS?]

I didn't know that the FAA had replaced the 360/50's with 3083's. The 3083
is an early 1980's machine, pre-XA, pre-ESCON.

The 360/50 is a quarter MIPS air cooled CPU with selector and byte MUX
channels. It would give an 8088 a run.

The 3083 has a CPU cooled by chilled water and pressurized helium circulating
through cool plates with pistons that press on the integrated circuits. A
3083 is about a powerful as a Pentium-Pro. It has a single 3+ MIPS CPU.

It's a little more powerful than my desktop mainframe. My system is rated at
2+ MIPS but I can increase the clock and get almost 3 MIPS.


> Monte Belger, the associate administrator for Air Traffic Services,
>said in an interview that the FAA should know within 90 days whether
>the computers can be de-bugged. The problem is that the date
>functions are not in programming languages, like Fortran or Cobol, but
>in machine language -- strings of ones and zeros more basic to the
>computer than even the operating system.
>
>[90 days? 90 FREAKING DAYS FROM NOW? Why not a bloody year ago!
>ASSEMBLER??? <he moaned, head in hands> Cory's right. If they had
>any sense at all, which they don't they'll get the brightest and best
>of the assembler heads together but no...]

It's a death march. A commercial assembler programmer, someone who codes
insurance or banking applications has a ramp-up cost when switching to this
kind of work.

>the FAA had planned to replace the computers by 2003 and would like to
>do so sooner because IBM says spare parts for the mainframes,
>installed in the late 1980s, are rapidly disappearing.

There seems to be a rule that after 15 years, the parts disappear, people
forget how to fix it, and the things start to breakdown.


>
>The computers in question are at the 20 Air Route Traffic Control
>Centers, which handle all the high-altitude, long-distance traffic in
>the country. The 3083 models were once common in business and industry
>but few remain in service, experts say. IBM stopped shipping them
>about 10 years ago, but some of the software on the FAA models is even
>older, dating from the early 1970s.

Yoo-hoo, any denial-heads here? What about quickee solution peddlers? Let's
chip in and send Timmy or Dash to tell the FAA and IBM how to fix the 3083.
..that's a little unfair to Timmy and Dash but I haven't had my morning
coffee yet, I'm sitting here with morning-breath, what hair I have left
sticking out in funny angles, my gut hanging over my underpants and for some
reason, the socks I wore yesterday still on my feet.

>
>The FAA has 250 separate computer systems, most of which will require
>fixes but the 3083 is the only one that IBM says can't be debugged
>before 2000.

Denial heads, please read the above line slowly and move your lips. " . IBM
. says . can't . be . debugged . before . 2000 . "

>
>[Hmmm sounds like the assembler is using 3083 unique stuff? Or maybe
>the assembler is using only a specific hard drive by hard coding the
>sector, track and cylinder numbers for utmost speed? Or ??? But then
>I'm not a main framer.]

It's not that, Tony, the 360/50 had some special modifications to the
electronics to help with driving the FAA's displays and communications. I'm
guessing that similar modifications were made to the 3083.

>
>In the FAA's case, the computers are called the "hosts," and are used
>to receive data from radar scattered across thousands of square miles
>and integrate the images into a mosaic. Then the hosts divide that
>picture into sectors, the subdivisions that controllers use, and pass
>the data on to other computers that drive the screens at the
>controllers' work stations. The 3083's also receive signals from each
>plane stating its identity, type of equipment, altitude and
>destination and helps tag each radar blip with the appropriate data.
>
>[Uh oh, this sounds like one of those critical systems, doesn't it?
>This ones gotta work.]

Where's that Ya-hoo who was argueing with us about 6 months ago about how the
FAA didn't have a problem.

>
> In the October letter from IBM to the FAA contractor Lockheed Martin
>Air Traffic Management, it said, "IBM remains convinced that the
>appropriate skills and tools do not exist to conduct a complete Year
> 2000 test assessment" of the 3083 computers. "IBM believes it is
>imperative that the F.A.A. replace the equipment" before 2000.
>

But did Low Morale understand IBM's letter? Denial-heads, IBM is sticking
it's neck out on this one. There are no weasle-words in the quote. It's
simple, shock-em, don't let them wish for skills and tools that don't exist,
language. There's too much ..I'm wishing for a solution so one must exist,
going on. IBM is working hard to stomp out that fantasy.

>
> Others are still worried. Fanfalone of the technician's union said,
>"There's only two folks at IBM who know the micro-code, and they're
>both retired.
>
>[Yikes. So one has a heart attack and the other decides to move to
>Fiji.]

Certainly both are dottering old fools, they didn't retire... I'd bet they
were given the golden heave-ho in the RIFs of the late 1980s and early 1990s.
Get the H*** outa here, you useless old natterer, and don't let the door hit
your behind on the way out, we want cheap young EE's who know micro processors
not you expensive geezers who invented most of the technology that runs the
civilized world.

>
> "The FAA pulled one of them in, to go through the code, but they
>should have brought 10 software people over," he said.
>
>[10. Not enough. Maybe 10 very, very good competent people who don't
>have to attend meetings and are allowed to concentrate on thier work.
>Oh, and you'd better pay them whatever it takes to keep them. And if
>it means $200,000 or $400,000 per year then bloody well do it.]

100 or 200/hour? To save the FAA's hairy behind? If the FAA had half a brain
in the 1980s and 1990s, they would have paid IBM and Low Morale to keep the
skills around, kind of the way every city and town pays for firemen to sleep
and watch TV, policemen to drive around, eat donuts, and be available.

The Y2K fire is about to break out. 200/hour is cheap, you're not paying
firemen to sit around any more, you're paying Red Adaire to put out burning
oilwells. Try 2,000/hour, try 20,000/hour.

>
>Tony

Cory Hamasaki