FAS IT Undergoes Needed Renovation

Last Saturday, FAS IT completely replaced a piece of hardware that has caught fire two times within the past year.

In February, a short circuit generated extreme heat and smoke inside one of two Uninterruptible Power Supply (UPS) systems in the Science Center data center, leading to a power failure that shut down much of the computing services for the Faculty of Arts and Sciences (FAS).

The UPS system itself is the “safety net” intended to provide power during the few minutes the electrical power grid may suddenly fail, according to Harvard Computer Society president Joshua A. Kroll ’09. The system also provides a “blip in power” if officials briefly shut down the data center for electrical work, he said.

A similar fire had occurred in the same UPS system in March 2007. After that breakdown, the machinery was repaired but not replaced, Kroll said, and FAS IT continued its partnership with the vendor, MGE, that had delivered the faulty technology.

But the reoccurrence of the same problem in February 2008 damaged FAS IT’s trust in MGE, said FAS IT spokesman Noah S. Selsby ’94.

Selsby said MGE claimed to have never seen such cases of an “extremely rare malfunction”—Harvard’s two instances of power failures were the only ones found in MGE’s records.

The first failure “made us question our confidence in the vendor,” Selsby said. “The second one pretty much put them down.”

So when FAS IT finally replaced the faulty UPS system last week, they chose a different vendor.

The two power failures within the past year have only increased the need to investigate and improve the FAS IT system.

FAS Chief Information Officer Larry Levine sent an e-mail announcement last week to all faculty, staff, and students at the College about a system-wide electrical maintenance this month to address the way power is provided in the Science Center machine rooms.

To ensure that future power failures will not be of similar magnitude, FAS IT will be increasing what Selsby termed “redundancy,” or backup systems that ensure that not all services will be affected if one part of the IT system fails.

FAS IT has split its three data centers so that if any one of them suffers a power failure, the rest will remain available, Kroll said.

As a result of the outage in February, services such as the my.harvard portal, FAS e-mail accounts, and FAS Web sites were disrupted—freshmen were not able to section for Expository Writing 20, the mandatory freshman writing class, and many students lost access to the Office of Career Services Web site in the middle of internship recruiting season.

“By doing this work and creating these redundancies, we’re aiming at making these maintenances obsolete,” Selsby said in reference to the planned shut-down of electrical services to the data center.

Selsby said that these maintenances are also part of an ongoing, larger effort funded by FAS known as the “Core project,” which seeks to upgrade FAS networks. Over the past few years, Selsby said that FAS IT has been upgrading “all sorts of things,” including actual servers, machine rooms, and the electrical system itself.

Though FAS will not provide “infinite” funding for this project, Selsby said that FAS IT will continue to improve areas in need of upgrade, recently focusing on the student and faculty e-mail system.

“Technology doesn’t plateau,” Selsby said. “The average life span of a computer is between three to four years at best, so factor that in to see what the infrastructure needs.”

—Staff Writer June Q. Wu can be reached at junewu@fas.harvard.edu.

—Staff Writer Esther I. Yi can be reached at estheryi@fas.harvard.edu.