UUNET goes back up

WorldCom announced late Thursday that its IP network has been restored to normal ending a service outage that the carrier said affected 20% of its IP customers in the U.S.

“A preliminary investigation indicates there was a route table issue,” WorldCom said in a statement. “We will continue to monitor the network, and we apologize to our customers who were affected.”

Between 7 a.m. and 8 a.m. EST Thursday, the UUNET network owned by WorldCom suffered from a sharp spike in packet loss on many areas of its network. WorldCom had conducted network maintenance the day before, installing new router software at peering and hosting locations. Outside network monitors said when the software went live Thursday morning, it caused many of those routers to shut town as tables that determine where IP traffic is routed malfunctioned. Because the problems were located at peering points, traffic from outside WorldCom’s network had trouble getting into the network and vice versa, creating problems not only for WorldCom customers, but carriers that have peering relationships with UUNET such as AT&T and Sprint.

According to Matrix NetSystems, an IP network monitoring firm, packet loss at problem spots on WorldCom’s network climbed as high as 22% and latency went as high as 1000 milliseconds -- far beyond the watermarks that render the network inoperable.

WorldCom said service was restored to normal Thursday afternoon.