Cory on Rail systemsgreenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread
The rail "problem" is a systems problem, not a mechanical switch problem, or a people problem.
We've had lots of good info from rail folk but like the banking, food distribution, air travel, power, etc., it seems to be people who have good expertise at a narrow part of the industry.
Here're my concerns.
The location, contents, ownership, destination, urgency of rail freight boxcar contents are stored in a mainframe computer in Washington DC.
Years ago, I saw a portion of the source code to the system. It belonged to a IT shop owned by a consortium of railroads. As reported by others in c.s.y2k, this system is now owned by an entity calling itself, Railinc.
Railinc was Y2K remediating the software last summer using contractors from GEISCO, and was moving their operations center away from Washington DC. They were also moving the programming work away from the operations center.
There was some suggestion that they were converting to new technologies at the same time. This may or may not be important or true.
The business rules contained in the Railinc system require current, realtime information. The business rules drive the rail switches in realtime.
There has been some discussion on the ability of track operators to switch track and trains. I don't see that as the heart of the problem.
The problem is switching the cars within the trains. To do this, you need accurate information in the central computer. This information has three pieces, where is the car going, where is it in space, where is it in time.
My concern isn't with the ability of brave strong workers to switch the trains. They can do it.
The problem is, how do they know what and when to switch? Especially when.
Certainly the issues of the telcos, power, finance, all apply but I'm ignoring those. The question is, will the Railinc central database process time dependent data correctly?
To know that, we need a first hand report from someone with current experience with the code.
My guesses, and these are only guesses, are based on experience in other industries. In these other industries, I have first hand experience that while the systems are in very bad shape, as I read the status reports bubbling up the management structure, the news gets progressively better.
In one striking case, the organization's CIO reported to the press that they had "solved the Y2K problem" but in reality, they had only recognized that there was a Y2K problem. That organization is still mulling over how big the problem is.
In the last week, they ordered a full court press on remediation for a soon to be replaced system.
A few days later, they canceled the remediation because the contractor building the replacement system scammed the CIO into thinking that the replacement system was "absolutely going to be ready on time".
It's still about money, clueless management, wishing really, really hard, and we're about to pay the price.
cory hamasaki 348 Days, 8,367 Hours
http://www.kiyoinc.com/current.html for WRP 108
-- Justa paster (email@example.com), January 19, 1999
So also the following archived thread. In particular, Vern Moores distinction between the ability to manually handle unit trains (coal and grain) v. single carload shipments at the major freight classification (distribution sorting) yards.
-- Brooks (firstname.lastname@example.org), January 19, 1999.
My company was impacted by the "UP/and somebody" (sorry, I forget who)merger a couple of years ago. Justa hit the nail on the head for me. They're computer systems didn't merge right. Most of our raw materials coming up from the gulf just disappeared, and stayed disappeared for weeks. The only way to move the stuff was for yardmen in every location to walk the tracks and report car numbers. Traceability was lost. We almost shut down. Times this by all rails at once. That's why the supply chain worries me the most. Will our South American veggies show up before they spoil? Doubt it. Will coal make it before we lose power? Doubt it.
-- margie mason (email@example.com), January 19, 1999.
It was the Southern Pacific, and the mess really isn't straightened out yet. At its worst, grain cars were in Houston and tanker cars were in Nebraska. I know there are varying opinions on this, but at the local rail yard here in East Texas, there are NO manual switches. They are gone, sacrificed at the altar of the computer.
-- Vic (firstname.lastname@example.org), January 19, 1999.
Many moons ago I was involved (re:chief designer) of a system used by a major steel company to track the creation and movement of seamless pipe between plants. There were 8 major plants & a couple dozen storage facilities in three states. About 99% of the movement of the product was by rail. The issue of car switching (these are open, flat bed like cars) was important because there were different kinds of pipe and semifinished materials moving about. It took a year to write the system and another year to implemement fully. The complexity of a general railcar, realtime switching system is hard to appreciate by non-geeks. In many ways, its at least as complex as the Air Traffic control systems in place. If this one system screws up badly enough (say a 10% error rate), many industrys will come to a sudden GIT halt.
-- RD. ->H (email@example.com), January 19, 1999.