UW News

October 8, 2007

‘Google 101’ class at UW inspires first Internet-scale programming courses

UW News

A pilot course taught at the University of Washington has been expanded into a national program that shows students how to program using tens, hundreds or thousands of computers. Google and IBM announced an initiative Monday to promote new programming methods that will help students and researchers address the challenges of Internet-scale applications.

“This is a new style of computing in which the focus is on analyzing massive amounts of data, using massive numbers of computers,” said Ed Lazowska, a UW computer science and engineering professor. “This has come on the scene fairly recently. Universities haven’t been teaching it in part because the software is really complex, and in part because you need a big rack of computers to support it.”

That is about to change. Based on the success of last year’s pilot course at the UW, Google and IBM decided to donate and maintain hundreds of processors that students across the country can use for courses to address large-scale computing on the Web. More information on the program is available in a press release from Google. University partners include the UW, Carnegie Mellon University, the Massachusetts Institute of Technology, Stanford University, the University of California, Berkeley, and the University of Maryland. In future the academic consortium will be expanded to include additional researchers, educators and scientists.

The seeds of the program were planted last year when Christophe Bisciglia, a senior software engineer at Google who graduated from the UW in 2003 with a degree in computer science, was interviewing potential Google employees. Many bright applicants could solve very difficult problems that could be done on a single computer, he said, but when asked what they would do if confronted with 1,000 times as much data, they were stumped.

Bisciglia wondered if he could modernize the curriculum to address these types of challenges. He started sketching out a curriculum in the company’s “20 percent” program, which allows employees to use one day a week to work on their own projects or ideas. Last fall he moved 80 processors to Seattle. The first offering of “Problem Solving on Large-Scale Clusters,” nicknamed “Google 101,” was taught by two UW alumni, Hannah Tang and Albert Wong, who work in Google’s Kirkland office. Fifteen students enrolled in the five-week seminar-style course.

“One of my big intentions was to close the gap between how industry and academia think about computing,” said Bisciglia, who served as course director.

In spring quarter this year, Bisciglia and others refined the curriculum and offered a more formal course taught by UW graduate student Aaron Kimball. One group of students created a program named Geozette that finds news article, scans the text in an attempt to determine where the news took place and then displays the articles on a world map.

“Being able to deal effectively with big datasets is going to be relevant regardless of whether you’re working at Google or Microsoft or Yahoo or Zillow,” said Sierra Michels-Slettvet, a senior UW computer science student who took the class in winter quarter and acted as a teaching assistant in the spring.

“It develops a set of skills that all employers in Internet-scale fields are looking for,” she said, “and it’s not something that’s addressed in the rest of the curriculum.”

Bisciglia said: “Based on the success we had at the UW, people at Google started asking me to spend more and more time on this, to the point that it’s a full-time job.”

Google isn’t the only one that will benefit from the newly trained graduates, Lazowska said. Other companies also use large-scale computing methods. And many scientific fields are confronting colossal amounts of data coming from computerized instruments or electronic sensors placed in roads and bridges, on the seafloor and in living organisms.

“More and more fields are becoming data-rich,” Lazowska said. “And extracting knowledge from massive amounts of data is exactly what this is about.”

###


For more information, contact Lazowska at (206) 543-4755 or lazowska@cs.washington.edu. To schedule an interview with Bisciglia, contact Google public relations officer Andrew Pederson at andrewpederson@google.com or (650) 214-6228.


For more details see the Google/IBM joint press release, http://www.google.com/intl/en/press/pressrel/20071008_ibm_univ.html.