Stan Graham of Tech-Beamers, a code-repair organization, testified to Congress that the FAA will not be fully compliant by Jan. 1, 2000. His presentation was on Feb. 4, 1998.
One comment was especially notable for all y2k repair projects: "Finding the cause and fixing the last 10% of the problems can take 20-30% of the time."
This suggests that a lot of repair jobs -- possibly all of them -- will experience slowdowns that project managers have not taken into account. Yet in the case of the U.S. government, deadlines for testing are sometimes as late as November, 1999. And testing takes up to 70% of the total job!
* * * * * * * *
Thank you. Mr. Chairman, I hope to be of help to your subcommittees on the FAA Year 2000 challenge. Based on information supplied by the GAO and your staff, I concluded that many of the FAA’s 414 critical applications will not be "Year 2000 ready" by the end of 1999. I agree with the GAO that the problem at the FAA is indeed a project management one, however with technical complications. Furthermore, it appears that the FAA does not have an objective methodology for planning and tracking their Year 2000 project schedules. . . .
Most projects start off slowly, the FAA applications are no exception. In order to become Year 2000 ready, the FAA must significantly accelerate their progress during the next few months. However even when they do, experience shows that this higher rate will not be sustained. Finding the cause and fixing the last 10% of the problems can take 20-30% of the time. . . .
. . . Even if the FAA stops slipping their schedule, they would miss their deadline by more than 7 months. If they continue to slip at the same rate, they would finish almost 9 ˝ years late. The FAA schedules are "open," and therefore not reliable. As a result, I believe that there is little chance that the FAA’s critical applications will be Year 2000 ready.
According to the GAO report, the number of application systems is large; more than 875 systems, 18,000 subsystems, and 65 million lines of code. Furthermore, through November 15, 1997 the Department of Transportation reported that the FAA’s large complex systems were only 3% through program renovation. Although this number should rise with later data, we will not have a reliable number until all 65 million lines of code and their data have been analyzed or accounted for.
Furthermore, testing is projected to take more than 50-70% of a project’s schedule. A strong testing discipline is required. . . .
Integral to executing each of the FAA’s applications are extensive information processing and communication technologies. These systems depend on 80 inter-related computers just for the en-route system to safely guide and direct aircraft. What makes the FAA problem particularly serious is that the FAA is running computers that are about 20 years old.
What makes the situation even more difficult is that special instructions have been added to the FAA computers. The Agency also has it’s own proprietary computer operating system software that is used no where else in the world. . . .
In general, project planning and tracking methodologies most organizations use are antiquated. From the data I have seen, the FAA project is no different. These methods are undisciplined, rely largely on intuition and hunch, and are constrained by edict. Industry experts report that 80% or more of computer projects fail to meet their dates with the function committed. This is unacceptable for the Year 2000. . . .
Tech-Beamers was founded in 1993. It is staffed by former IBM employees, mainframe and personal computer system experts who average more than 25 years in the computer industry. Tech-Beamers, with headquarters in Poughkeepsie, N.Y., provides consulting, training, and S/390 micro and mainframe sales management services world-wide. The company, through its network of business partners, serves countries on the North and South American Continents, and Europe. They are considered leaders in system migration, testing, and project management tools for the Year 2000. They are a leading system technology company that develops and integrates migration tools, application enabling software, and client- server solutions for the IBM S/390 mainframes on RISC and Intel based servers and workstations. Tech-Beamers utilizes the best of breed Internet and remote networking tools to demonstrate their products and service their customers world-wide. They have developed unique methodologies and authored white papers on the Year 2000 that focus on Testing, Testing Schedule Management, and Project Management which have appeared in the leading Year 2000 publications. Several have been selected as the "Best Methodologies" for the Year 2000.