Top of the News
THE PHONE SYSTEM OF TOMORROW IS CAUSING MIGRAINES TODAY
On June 26, more than 5 million customers in the District of Columbia, Maryland, Virginia, and West Virginia lost most of their local phone service for seven hours. The problem: A computer network that sets up calls went haywire, possibly because of a bug in its software. The same type of network was implicated in a major outage in Los Angeles on the same day. And in smaller outages in Pittsburgh and San Francisco a few days later. And in a memorable snafu in January, 1990, that crippled American Telephone & Telegraph Co.'s long-distance network for nine hours.
Beginning to notice a pattern? So are the telephone companies. But rather than reappraising the networks involved, all based on a design known as Common Channel Signaling System No. 7, the companies are actually planning to interconnect them in coming years--knitting together local, long-distance, and cellular phone companies more tightly than ever before. Some outside experts fear that doing so could allow problems to hop from one company's phone network to another, potentially bringing the entire national phone system to a standstill.
'BORDER GUARDS.' An outage that cascaded across networks would dwarf all the failures to date--and would be much harder to solve, given the likelihood of finger-pointing among multiple carriers with many brands of equipment. Michael Hills, president of HTL Telemanagement Ltd., a Burtonsville (Md.) consultant, calls Signaling System 7 "a disaster waiting to happen." He accuses the phone companies of overreliance on centralized computer systems. Says Hills, whose own local service was knocked out on June 26: "This is a disease telephone engineers have, of trying to centralize things too much."
That's an extreme view. The phone companies claim that they've designed filters that will allow only those messages that pass a slew of tests to go from one network to another, cutting the chance of a failure spreading between networks. Says AT&T spokesman James Messenger: "Each point of interconnect will have border guards that are very efficient and very fast."
AT&T and other phone companies have a strong incentive to mend SS7: It allows them to use their networks more efficiently, saving hundreds of millions of dollars a year. It also provides a highway for new services such as caller identification and advanced call-forwarding. In essence, it's a high-speed data network of signal-transfer points that overlays the traditional phone network. Electronic messengers race ahead to see if someone's on the phone before a valuable circuit is wasted in sending the call through. The couriers can validate a credit-card number stored in a data base or look up how to route a toll-free call.
Such complexity breeds vulnerability. Investigators say the latest outages all involved signal-transfer-point computers made by DSC Communications Corp. in Plano, Tex. Bell Atlantic Corp., whose mid-Atlantic network was creamed on June 26, says it's too soon to lay blame. Nevertheless, Bell Atlantic President Anton J. Campanella tweaked DSC on July 1, saying: "My tummy gets upset when a manufacturer delivers a product that doesn't work."
Perhaps optimistically, Campanella says he still hopes Bell Atlantic can proceed with its industry-leading plan to tie its SS7 networks in three regions into those of a dozen long-distance carriers by yearend. Meanwhile, BellSouth Corp. is testing SS7 links with McCaw Cellular Communications Inc. in South Florida. As it happens, cellular phones were spared in the recent outages--precisely because they weren't connected into SS7.
ROAMING EMPIRE. Carriers expect to rely on SS7 even more in coming years. The Advanced Intelligent Network of the future is to be built on a relative handful of computers that will know, for example, where customers are at any given time. SS7 will consult the computers and route calls accordingly. Phone customers will have one phone number for life and need never be out of touch. But if SS7 or the computers fail, the network will be brain-dead.
Raymond F. Albers, Bell Atlantic's assistant vice-president for technology planning, notes that phone failures were more frequent in the old days. Only then, they were far more localized--the result of a severed cable, say, or a fire in a switching center. With today's technology, network failures are rarer, but when they happen they can be doozies. While engineers believe that they can fix whatever ails SS7, they aren't making firm promises. "Notice a lot of us are using words like `should' and `ought to' instead of `will,' " Albers says. As networks become more dependent on computers, a lot of faith is being placed in fallible technology.Peter Coy in New York