(January 1999) As 1999 hurtles recklessly towards the year 2000, possibly the most spectacular question facing American consumers is this: To fly, or not to fly? Everybody knows that air traffic controllers are helpless without their computers up and running; everybody
also knows that when January 1, 2000 arrives, computers across the country could start going down like rosebuds in a hailstorm. In congressional hearings early last year, the Office of Management and Budget roundly criticized the Federal Aviation Administration for its lack of progress towards a Y2K solution,. To address the problem, the FAA and the OMB worked together to lay out an aggressive timetable. A year has now passed since those dramatic he
arings, but very little has been heard about what’s really going on with the FAA’s efforts to become Y2K-compliant.
Who knows? Maybe no news is good news. In a recent telephone interview, the FAA’s public affairs office in Washington, D.C., stated that although the FAA is lagging behind the official deadlines set out by the OMB, i
t has completed the "renovation" of its systems and plans to have everything tested and implemented long before the year 2000 arrives. In fact, if you visit the FAA’s special Y2K web site, you might forget that there e
ver was a problem. It’s full of cozy reassurances and an ambitious timetable, the upshot being that everything is going smoothly and that all "mission critical" systems will be compliant by June 30, 1999.
We investigated the FAA progress, and though it’s true that they’ve made many positive gains since last February, don’t be fooled. The FAA is far from out of the woods, and the flip side of its shiny Krugerrand of reassurance is a dull and gritty kopeck o
f uncertainty and unanswered questions.
First off, the biggest, nastiest problem in the FAA's system is the computer at its core, the IBM model 3083 mainframe. Even though they were built back in the '70s -- ancient history, in computer time -- the 3083's still form the backbone of the FAA’s
long-distance control system. They drive the information displays at the 20 FAA "enroute centers" throughout the country; these are the centers that take over the control of air traffic at 60 miles from takeoff and handle all longer-range air traffic in
the U.S. In October of 1997, IBM stated bluntly that "the appropriate skills and tools do not exist to conduct a complete Year 2000 test assessment of the IBM model 3083."
In the face of this serious problem, the FAA has decided to replace or "rehost" the 3083 by the year 2000 using a system called Display System Replacement (DSR). In January of this year the FAA was very happy to report that the first DSR system went online in Auburn, Wash. but there are still 19 enroute centers that must be completely overhauled by the end of December, and the last time the FAA undertook rehosting on this scale it took three years to complete the job. The FAA does acknowledge that rehosting might not get finished in time for the new year, however, and so they offer the following: "...as a contingency to Host (system) replacement, we have already completed renovations of the exis
ting Host as of July 31, two months ahead of OMB’s September 30th renovation deadline. If there is a need for the Host to be operational in the Year 2000, we are assured that it will transition to the new millennium in a routine manner."
But wait. IBM said the problem couldn’t even be assessed, let alone fixed. How has this machine gone from completely unfixable to completely fixed in such a short time?
The key to the mystery lies in the FAA's definition of the word "renovation." You'd think that a system that had been "renovated" had been fixed. Or at least looked at. Not so in this case. What the FAA means by "renovation" is nothing more than simula
ted repairs on a simulated version of the FAA computer system, which are being conducted at an FAA technical center in New Jersey. Most of the testing is still being done on the simulator, and the FAA won't specify how much testing is actually being done
on the real machines.
Using a simulator to fix a network as complicated as the air traffic control system is a valid first step, but it’s literally worlds away from working on a real machine. The gap between simulation and reality is made even wider by the fact that the act
ual FAA machines have been independently maintained, upgraded and altered by separate groups of technicians at nearly 200 separate locations across the country since the 1970s. These systems have been evolving away from each other for decades, and a Y2K f
ix developed on a simulated machine in New Jersey may have a hard time making friends with an actual 25-year-old FAA computer in Tulsa. Hopefully the FAA will move on to the implementation phase of the project soon. That’s when we’ll all find out if the t
heoretical repairs actually work in the big, bad world.
AP Photo/Eddy Palumbo
IN PART TWOA Y2K EXPERT SPEAKS HIS MIND ON THE FAA, AND WE OFFER SOME ADVICE FOR CONCERNED TRAVELERS.
Traffic at airports like Philadelphia International is on the rise.
