August 17, 1999

MCI Points Finger at Lucent for Recent Network Woes

---

Free Service Is Offered to Make Amends

By SETH SCHIESEL

fter one of the AT&T Corporation's big data networks, based on equipment from Cisco Systems Inc., failed last year, AT&T fully restored the system in about 24 hours.

After a similar network run by MCI Worldcom Inc., based on equipment from Lucent Technologies Inc., began having problems on Aug. 5, the system was not fully restored until Sunday, 10 days later.

Related Articles MCI Outage Could Drive Net Providers Out of Business (August 17, 1999) Network Trouble Enters Eighth Day (August 14, 1999) MCI Network Problems Persist (August 13, 1999) MCI Still Fixing Glitches (August 12, 1999) MCI Worldcom Problems Hit Its Data Network Users (August 10, 1999)

In his first public comments on the breakdown, in a conference call Monday, Bernard J. Ebbers, MCI Worldcom's chief executive, said that MCI Worldcom planned to offer each of the as many as 3,000 customers who were affected by the problems up to 20 days of free service.

He said it would cost his company only a "very, very slight downtick" in revenue.

But the fallout for MCI Worldcom, the nation's No. 2 long-distance communications carrier, may extend far beyond the next quarterly financial report.

Ebbers tried to shift ultimate responsibility for the problem to Lucent, and Lucent -- which sells MCI Worldcom hundreds of millions of dollars worth of products every year -- accepted it. Some of MCI Worldcom's own customers, however, have been scathing in their attacks on the company. And some analysts pointed out that the network debacle raised unflattering questions about MCI Worldcom's policies for handling such situations, especially in light of AT&T's speed in dealing with a similar problem last year.

Calling the problems "unacceptable" and describing as "hollow" MCI Worldcom's assurances last week that the problem was under control, Thomas R. Donovan, president of the Chicago Board of Trade, a big MCI Worldcom customer, wrote in a letter to Ebbers on Friday: "As a result of MCI Worldcom's failure to deliver on their promises to me early last week, the C.B.O.T. is pursuing all available remedies. We have lost all confidence in MCI Worldcom's ability to provide reliable service."

In yesterday's conference call, Ebbers apologized to his customers and said any additional compensation beyond the 20 days of free service would be negotiated privately with individual customers.

But he essentially disavowed responsibility, saying the problem was related to an upgrade of Lucent's software.

"We do not have a definitive cause of the problem," he said. "That is something that Lucent continues to work on. Lucent has acknowledged full responsibility." He added: "Because we do not as a company -- and no one else in the industry does either -- write software for this type of switch and network, that it is not something within our power, to determine the exact cause of the problem."

Ebbers thanked the network experts from Lucent and its Bell Laboratories unit who worked on the problem, but later in the call he took a serious swipe at Lucent, wondering aloud whether the company even employed the right people.

An unusual airing of the long-distance industry's dirty laundry.

"Part of the reason that there is some concern here is that there has been a lot of consolidation in our industry and this software was originally developed by Cascade Communications, who was then acquired by Ascend Communications, who has since been acquired by Lucent," he said. "And so one of the concerns obviously in this cycle of events is what happened to the people and the process that did the development and wrote the software. And was the capability to maintain this software retained through these transactions?"

That might sound somewhat obtuse, but for the telecommunications industry those are very harsh words. Communications carriers generally refuse to discuss their vendors at all, yet Ebbers publicly questioned whether Lucent, which spent $20 billion of its stockholders' money to acquire Ascend earlier this year, has been managing that deal correctly.

Lucent took the high road, accepting responsibility, as indeed it should if its software was at fault.

Still, some analysts said that in the end, MCI Worldcom had to bear responsibility for allowing the problem to continue for as long as it did. After trying for more than a week to fix the problem (or to have Lucent fix it) while the network remained online, MCI Worldcom threw in the towel this past weekend by shutting down the whole system and reloading an older but more stable software version.

But that raises the hard question of why MCI Worldcom waited almost a week and a half to make that decision. In one way, the choice was easier for AT&T because it had an almost complete failure; that allowed AT&T to limit the length of the crisis by reloading the proper software. MCI Worldcom spent all of last week betting that it could fix the problem without shutting down, and it lost that bet.

"When you delay something this long, this is not a vendor problem," said Frank Dzubeck, the president of Communications Network Architects, a networking analysis and consulting firm in Washington. "Somebody in the organization made the decision to let this situation linger. What does the song say? Sometimes you've got to hold 'em, sometimes you've got to fold 'em. This was a great time to fold 'em, cut your losses and move on."