|
|
Will year 2000 be ultimate wrong number?
Sunday, August 23, 1998
By SCOTT MORITZ
Staff Writer
The worst telephone failure of the decade was caused by an error in a computer code. At 2:21 on Monday afternoon on Jan. 15, 1990, a software bug -- believed to be some flawed or missing characters in a line of program code -- infected AT&T's call-switching equipment in a lower Manhattan office building.
For nearly nine hours, AT&T's long-distance service was crippled as the faulty code caused one switch after another to falsely report that it was overloaded. As the effects of the code flaw cascaded through the network's 114 switches, callers received busy signals and repeatedly redialed, piling on to an already bad crash.
In the final tally, an estimated 70 million calls nationwide failed to connect.
That episode, say some telecommunications experts, will look like a ripple on a pond compared to the year 2000 tsunami ahead.
And come the end of this century, the problem won't be one bug in one phone switch, but potentially countless millennium bugs in countless programs worldwide.
At the heart of the problem is a characteristic in the programming code that is used in computers and microprocessors that run everything from VCRs to elevators to communications satellites.
This all started in the early Sixties, when programmers decided to cut a corner and abbreviate the date code to show the year in two digits instead of four. This means computer systems won't know the difference between 2000 and 1900. The effects are expected to range from inconvenient to cataclysmic depending on whom you talk to.
In the phone industry, the date confusion could end up being as minor as a backlog of apparently outdated messages, or it could result in a complete communication shutdown as equipment fails to process computer commands as simple as on and off.
AT&T and Bell Atlantic say they are confident that Day One of the new millennium will be just another day in the network.
Since the start of its year 2000 program in 1996, AT&T has spent $463 million inspecting an estimated 500 million lines of programming code, as well as testing, upgrading, and replacing equipment. John Pasqua, director of AT&T's year 2000 project, says 80 percent of the job is complete and the company expects to certify its network for year 2000 compliance before the end of this year.
Bell Atlantic has spent between $200 million and $300 million preparing its network for the new century. The regional Bell operating company, which controls 40 million local-access lines on the Eastern Seaboard, expects to be in year 2000 compliance by July 1, 1999.
Both companies plan to begin testing how well they connect with other domestic and international phone company networks early next year.
One problem.
Phone networks, unlike stock exchanges, don't close on weekends for testing and repairs. AT&T is testing, fixing, and retesting all the various pieces of its network independently, but is unable to test the whole network live by pushing the clock forward and simulating a future date in 2000.
"We are unique," said Dave Johnson, AT&T's network operations spokesman, "it's like trying to overhaul a Boeing 747 in flight. The network is always in flight; you can't shut down segments of the network to work on it."
AT&T is not alone in this. Bell Atlantic and all the other carriers must rely on equipment that is tested in the artificial settings of a lab.
Skip Patterson, director of Bell Atlantic's year 2000 program, says shutting down parts or all of the network to make changes and run tests could end up creating a crisis while trying to prevent one.
A live network test is an appealing notion, Patterson said. "You could get close to a 100 percent guarantee that the network would work, but as a practical matter I just don't see it."
Murray Hill-based Lucent Technologies, the nation's largest telephone equipment maker, is urging phone companies to run live network tests to ensure that all components work together flawlessly.
Morriston-based Bellcore, which is serving as the primary year 2000 consultant to the regional Bells, is also urging phone companies to run tests and recommends that companies form plans of action should one or more networks fail.
Laboratory testing has long been the standard gauge for equipment reliability in the telephone industry, but the true test for these specific code flaws will be Jan. 1, 2000.
Inevitably, some questions will be left to chance, says Lisa Pierce, a telecommunications analyst with Giga Information Group in Cambridge, Mass.
"Because you can't do a live network test, there will be a small potential for problems," Pierce said.
AT&T says it will have SWAT teams ready "to deal with any situation that might occur."
Of greater concern to Pierce and other telecommunications experts are the problems that lie beyond any given network.
AT&T and Bell Atlantic are just two parts of a global communications network that is only as reliable as its weakest connection.
David Isenberg, a telecommunications analyst and founder of isen.com, a consulting company in Westfield, describes four possible vignettes after the clock strikes midnight on Jan. 1, 2000.
One: AT&T and Bell Atlantic's century-old networks hide numerous faulty date codes and will crash.
Two: Code exterminators squash most of the millennium bugs in AT&T and Bell Atlantic's networks, but some or all of the 1,400 or more other telephone networks are still buggy and calls won't go through.
Three: AT&T, Bell Atlantic, and all the connecting carriers are bug-free, but countless office phone networks are in turmoil. Airline call centers shut down; customer service operators for credit card companies, banks, and utilities are silenced; hospital switchboards seize.
Four: overload. Like a stampede, if some unforeseen problem cripples another industry, say, air traffic control for example, callers will flood the network with repeated attempts to check flights. This will cause busy signals and millions upon millions of redials that overload the network.
Phone company officials say forecasts like these are at best educated guesses.
But analysts say the companies are sharing very little information about what their tests have revealed or how prevalent the date code flaw is in their networks, billing systems, or satellite services.
This is especially alarming at a time when cooperation should be a paramount concern, Isenberg said.
The phone companies seem to be circling their wagons in an attempt to either shield themselves from lawsuits or avoid setting off a panic among customers, analysts say.
For its part, the Federal Communications Commission has excused itself from responsibility, saying "no single entity owns or controls" the public phone network. The FCC says it will try to encourage companies to work together and exchange information on preventing or fixing any year 2000 problems.
Pierce says the agency also needs to step in and alert consumers to any specific problems they could prepare for so they won't be caught off-guard.
"The public, in general, doesn't have a source of information on this stuff," Pierce said. "So it has to be the FCC's role."
Coming Monday: How the year 2000 glitch could affect your PC, and a look at how programmers are profiting in the end-of-century rush to fix the world's computer codes. In the Workplace section.