The press has at last caught a whiff of failure -- potentially a catastrophic failure. The Feb. 4 hearings on the FAA blew the whistle so loudly that the public actually heard a distant sound. This is the first major breakthrough in public awareness in the United States.
The fear will accelerate as more and more reports like this get published in official places.
This long, detailed statement was by Kenneth M. Mead, Inspector General, U.S. Department of Transportation. This is his testimony before the Subcommittee on Technology, Committee on Science, and the Subcommittee on Government Management, Information and Technology, Committee on Government Reform and Oversight, House of Representatives (Feb 4).
He says that the FAA is not finished with its assessment. The
California White Paper says that awareness is 1% of a y2k repair task, inventory is 1%, and assessment is 5%. This gives some indication of just how far behind the FAA is.
As you read this testimony, keep asking yourself, "How can this repair be made on time, given the late beginning and the enormous complexity of the task?"
Ask also: "If IBM says its outdated FAA computers can't be fixed, can they be fixed?"
* * * * * * *
FAA got a very late start on fixing Year-2000 computer problems. It was not until about 6 months ago that FAA began addressing the Year-2000 issue with a sense of urgency. Consequently, FAA is behind schedule on assessing which of its systems have Year-2000 problems, determining what needs to be fixed, and testing and implementing solutions. The good news is that it is not too late. Strong central management and a continuing sense of urgency are the keys to success.
Every piece of computer software and hardware must be assessed for problems, fixed as needed, and tested for Year-2000 compliance. The assessment work--identifying systems with Year-2000 problems--is almost finished, although it is 7 months after the target date specified by the Office of Management and Budget (OMB). Much of the assessment work on the Air Traffic Control System was completed last week under intensive central management and a firm requirement that the assessment work be finished by January 31.
FAA’s most difficult challenges lie ahead. . . .
The Year-2000 problem is not just a major challenge for Government. The entire aviation industry, including aircraft manufacturers, airlines, and airports must be involved. The challenge to FAA is great because its on-time track record for completing computer and software intensive programs has been poor. FAA’s scheduled November 1999 date to have the Year-2000 problem fixed brings us much too close to the "millennium bomb," a term recently used to describe the Year-2000 problem by the Government Executive Magazine. This time there is no room for schedule slippage; the due date is fixed. We urge FAA to move up the implementation date to have all systems Year-2000 compliant, tested, and operational no later than June 1999. . . .
OMB established a five-phase approach for addressing Year-2000 computer problems. According to the OMB schedule, agencies should have finished the second phase, which is to analyze existing systems for the scope of Year-2000 problems, by June 1997. In the next phase, agencies will fix Year-2000 problems by repairing existing software code or acquiring replacement systems. The OMB target for completion of this phase is September 1998. Then, the fix has to be tested to ensure it works as intended, including interfaces with other systems. Testing should be completed by January 1999. After successful testing, agencies will implement Year-2000 compliant systems to support their operations. OMB’s target date for full implementation is March 1999. . . .
Here are examples of three mission-critical systems for air traffic control that have been diagnosed with Year-2000 problems and must be fixed.
•The systems used in the En-route Centers consist of 4,000 pieces of hardware and software, including the Host mainframe computers, that allow air traffic controllers to manage aircraft flying at high altitude.
•The Offshore Flight Data Processing System is used to communicate and display positioning and flight plan information for aircraft over the oceans.
•The Terminal Doppler Weather Radar System is used to detect microbursts, gust fronts, wind shifts, and precipitation. This system alerts aircraft of hazardous weather conditions around airports and provides advanced notice of changing weather conditions. . . .
At a recent event sponsored by FAA, members of the airline industry raised concerns about how ready FAA, and the industry itself, would be for the Year 2000. In general, airlines were not very confident that the Air Traffic Control System would work correctly. Regional airlines thought the awareness level--the first step in addressing Year-2000 problems--among its members was not good. Airport representatives generally agreed that airports were not as aware of Year-2000 issues as they need to be. . . .
An additional concern is that FAA has not yet concluded that any of its Air Traffic Control Systems currently in the acquisition and research phase are Year-2000 compliant. There are 23 major projects under development, each costing more than $100 million. . . .
The Host computer is a key part of the system that enables air traffic controllers to direct high altitude air traffic from the En-route Centers. There are two issues concerning continued service of the Host computer beyond Year 2000: Can FAA make it Year-2000 compliant, and can FAA find replacement parts, which are already scarce?
International Business Machines (IBM), the manufacturer, recommended FAA replace the existing hardware because replacement parts are getting harder to find, and because IBM lacks the talents and tools to assess the Year-2000 problems in the Host computer. However, FAA maintains the Host computer can be fixed, and is considering a parallel effort to both repair the existing computer and replace it with an interim Host before Year 2000. FAA estimates it will cost about $2 million for the repair job, and about $160 million for an interim replacement.
FAA and IBM are at odds regarding the difficulties in evaluating and fixing Year-2000 problems in the Host computer. IBM has claimed the Host Year-2000 problems could not be properly assessed. In a letter dated October 2, 1997, IBM stated "Analysis of 3083 microcode (a machine language) involves reviewing hundreds of thousands of lines of microcode written in several different protocols. . . . IBM does not have the skills employed today that understand the microcode implemented in the 3083 well enough to conduct an appropriate Year-2000 assessment. In addition, the tools required to properly analyze the microcode do not exist." . . .
The Host computer issue illustrates an underlying problem with FAA’s Year-2000 program: a lack of strong leadership. While OMB guidance required agencies to establish Year-2000 program offices by December 1996, FAA did not take action until June 1997. . . .
The Air Traffic Control System is a complex and inter-dependent system. FAA has to analyze millions of lines of code and thousands of pieces of hardware for their individual Year-2000 problems, but also must ensure the systems will continue working together after Year-2000 fixes. . . .
The testing of all Year-2000 fixes and recommissioning the entire Air Traffic Control System is something FAA has never undertaken. . . .
Even with a well structured integrated test environment, FAA has no assurance all Year-2000 fixes will work as intended because field conditions are different from a test environment. The task is further complicated by the fact that local software changes have been made to the Air Traffic Control System. FAA has over 8,000 employees maintaining the Air Traffic Control System. The software developed or changed by local maintenance teams could result in different conditions from site to site. As a result, Year-2000 fixes working in a test environment may not work properly when implemented in the field.
FAA’s November 1999 target completion date for implementing all Year-2000 fixes for all mission-critical systems leaves little cushion for schedule slippage or corrective actions to solve problems unique to individual sites. FAA’s track record for solving hardware and software problems does not instill a high confidence level that the fixes can be made on schedule. . . .
If unexpected problems are identified during testing or implementation, FAA has to repeat much of the entire exercise--changing program code, retesting, and implementing new fixes. The closer it gets to Year 2000, the more it will cost to fix additional problems identified and to repeat such exercises.