MCI WorldCom CEO Bernard Ebbers says the company will offer double credits - up to 20 days' worth of free frame relay service - as compensation for the week-and-a-half-long frame relay network problem suffered by the company and its customers.
But Ebbers also left a clear indication that the incident may set back one of the company's most cherished but difficult integration projects - the goal of putting all frame relay customers on a single global platform.
Acknowledging that MCI WorldCom has not one, but several, frame relay networks, Ebbers said that over the weekend MCI WorldCom went back to an older software load on the affected Lucent frame relay switches. He added that there's "not a significant financial benefit" to integrating the networks.
That's something that the new, flawed software was apparently meant to help. The software was designed to scale up the Lucent switches to accept new customers. MCI WorldCom has been planning to center a global frame relay network around Lucent fast-packet switches, previously known as Ascend Communications equipment.
Ebbers and MCI WorldCom network-services President Ron Beaumont did not detail the exact makeup of the customer set affected by the congestion problems that began Thursday evening, Aug. 5. Beyond saying that the problem originated in Lucent switches, the pair gave contradictory answers as to which customers' services are based on the Lucent platform - at one point saying the problem affected only international customers, at another point saying it affected international and domestic connections equally.
Network World has reported that the problem appeared to affect legacy WorldCom customers rather than those on the original MCI network based on Nortel Networks' BayStream switches, known within MCI as the HyperStream network. Affected users also included customers with other carriers using network-to-network interconnections with WorldCom, plus customers of many ISPs that use frame relay links to MCI WorldCom's backbone provider UUNET.
Ebbers also laid the blame for the problem - which resulted in excess network congestion, followed by the loss of some frame relay network addresses - squarely at Lucent's feet. "Lucent has accepted the responsibility" for the problem, Ebbers said. "Because we do not write software for this kind of switch . . . it is not within our power" to repair the software, he added.
Following the lead of AT&T's response to its April 1998 frame relay outage, Ebbers said he would more than compensate all affected users for the congestion problems, which he and Beaumont also conceded led to "outages" in actual circuits. MCI WorldCom will provide a credit of two days for every day each customer was affected. For those affected for the duration of the problem - roughly Aug. 5 to Aug. 15 - that potentially means a credit of 20 days.
But some customers have already said that does not compensate them for lost revenue owing to data-network problems.
Ebbers apologized for the problem and said many MCI WorldCom employees have gone with little sleep while trying to fix the problem. Beaumont added that the network is now fully restored, except for a very small fraction of customers for whom engineers are still rebuilding network addresses.