Harvard Arts and Sciences Computer Services (HASCS) released a statement this week attributing the problems that struck the Faculty of Arts and Sciences (FAS) network last Monday and Tuesday nights to an error in the equipment that directs traffic across the network.
Director of Residential Computing Kevin S. Davis ’98 said that the FAS router that controls communication between the FAS network and the Internet at large encountered problems that Cisco Systems, which sold HASCS the router, had never previously observed.
The network troubles earlier this week disrupted e-mail, Internet and UNIX server access across the FAS network for a total of about 17 hours over the two evenings, according to the HASCS website.
Davis said that University Information Systems (UIS), which provides network services to most of Harvard besides FAS, also uses Cisco routers that experienced trouble at around the same times the HASCS router failed.
He said that both the HASCS and UIS problems involved complications with what is known as a routing table—information stored in each router’s internal memory that is used to determine locations on the Internet.
“The routing table for the Internet has grown larger and larger and we reached a size-related bug,” Davis said. “What it really comes down to was a heretofore unknown, very complicated problem.”
Davis said that HASCS and UIS network experts collaborated with Cisco engineers across the world both Monday and Tuesday nights to resolve the router trouble. The current fix is only temporary and work is in progress to find a permanent solution, according to the HASCS statement.
“[The error] has broader implications for other institutional and corporate networks,” the statement reads.
Davis said that while he did not know if the network trouble hit at night for any particular reason, he recognized its ramifications.
“It certainly had a relatively large impact on the residential community,” he said.
—Staff writer Joshua D. Gottlieb can be reached at firstname.lastname@example.org