HUSC Network Breaks Down

Approximately 1300 students were unable to access their computer accounts Saturday and Sunday on the Harvard Arts and Sciences Computer Services (HASCS) network due to a hardware failure.

One of four hard disks attached to the HASCS' HUSC 10 computer network failed Saturday around 5:00 p.m. and prevented access to computer accounts, HASCS/Unix System Manager Michael G. Burner said.

The information was restored early yesterday morning from backup tape drivers. However, any work done by students on HUSC 10, one of the system networks, between 6 a.m. Friday and the crash on Saturday afternoon was lost.

"I lost two days of work time," said Jeremy A. Rassen '95, a computer science concentrator from Adams House. Rassen could not access his account and therefore had to postpone his computer science work until yesterday.

Burner said that after the hard disk failed he restored the network with the three remaining hard disks, which allowed other students to continue to use the system.

Students with accounts on the hard disk that failed were shut out, thought.

Eventually, he was able to restore backup files for those accounts from Friday morning, but any modifications made after that were lost.

Burner said no students have yet complained to his office about their data being lost. He said if something "extremely crucial" had been lost, it might be possible to restore it, though this would require "hours and hours" of work.

Students use their computer accounts to access programs and information and often need to use the computers to complete work for science or math classes. The accounts store the work that they do and allow them access to facilities such as e-mail.

"Those of us taking computer courses depend on HUSC 10," said Michael J. Morell '95. Morell is an economics concentrator in Currier House who is taking a computer class. He could not access his work over the weekend.

"It was just annoying," he said. "I've been pretty disappointed in the computer service this year.

Burner said little could have been done to prevent the failure.

"Could you prevent the clutch going out on a car? It's one of those things," he said.

A method called "mirroring" could prevent such catastrophes, but Burner said the process is "simply not cost-effective for this environment."

The process duplicated hard drive activity and therefore requires twice the number of drivers.