Academic Cluster Computing Initiative

Academic Cluster Computing Initiative


MALE SPEAKER: This program got
started as a partnership between the University of
Washington and Google in an attempt to help students start
thinking about new ways of computation. MALE SPEAKER: This is
not vocational ed. The fact is, the great computer
science programs– Washington, Stanford, Berkley,
MIT, Carnegie Mellon– we’re teaching our students
skills that they’re going to need for a lifetime. CHRISTOPHE BISCIGLIA: This is a
combination of using Google hardware for most of
the CPU horsepower. IBM is also contributing
hardware for management services. And we do this, then, all using
open source technology. ED LAZOWSKA: For many decades
now, people are going to be solving these large scale data
intensive problems. And they’re going to be
solving them on clusters, ever larger clusters. HANNAH TANG: One of the great
things about having the cluster available in the
classroom is that students were able to iterate on their
ideas relatively quickly. DENNIS QUAN: And we want that
same simplicity and elegance to be applicable when they want
to be able to run their jobs across thousands
of machines. ED LAZOWSKA: If you look around
at computing companies, Google, Microsoft, Yahoo, IBM,
they’re all computing in roughly this way. CHRISTOPHE BISCIGLIA: And so
what we’re trying to do with this program is teach students
how to design software, interact with these clusters, so
they can address the needs of users and business
going forward. ED LAZOWSKA: That’s the
motivation from our point of view. It’s produce students who can
tackle today’s problems. CHRISTOPHE BISCIGLIA: During
the process of interviewing hundreds of college students,
you start to realize that even the most talented, brightest
students sometimes have trouble thinking about problems
on the scale of a thousand machines, or
terabytes of data. So I started thinking about
how we could work with universities to better prepare
students to face these challenges when they graduate. ED LAZOWSKA: What we’ve got
today is a whole new approach to problem solving, which is
analyzing massive amounts of data on large scale clusters. It’s no longer a niche. It’s something that’s in every
walk of life, whether you’re doing credit card fraud
detection or web analysis, you’re using that approach. DENNIS QUAN: All across the
industry, I think this is going to be beneficial to the
students in their careers, and also to make sure that this
next generation platform becomes widespread
and adopted. HANNAH TANG: For me, personally,
I think the biggest challenge for teaching
this course was just that it’s never been done before. Distributed computing is
relatively new in the undergraduate curriculum. SLAVA CHERNYAK: My name
is Slava Chernyak. I am a University of Washington
senior in the computer science department. My project was to create a
simulation of the collision between the Andromeda and
Milky Way galaxies. There’s a lot of computation
required to simulate such a collision. BRIAN STEADMAN: I’m
Brian Steadman. I’m a senior at the Department
of Computer Science at the University of Washington. You have a lot more
computational power. You have a lot more data. It’s a lot more real to
what’s currently out there in the industry. ED LAZOWSKA: IBM has developed
a plug-in for the Eclipse platform that all of our
students use that let them have access to a cluster located
anywhere in the world, just as if it were a machine
in this building. JULIETTE SCHWARZ: My name is
Juliette Schwarz, and I am a student at the University of
Washington, senior majoring in computer science. My project, which was called
geozette, was a website which took thousands and thousands of
news articles, figured out the location that these articles
were talking about, and then placed them
on a Google map. HANNAH TANG: Working on a single
computer, it might have taken them days or weeks in
order to find the patterns that they were looking for. But with having the cluster
available, they could try out an idea, experiment with it,
and decide it doesn’t work, and then move on into a
different direction within a matter of minutes. BRIAN STEADMAN: The real
difference this course made was, it allowed us to use
technology that businesses are using right now. MALE SPEAKER: Who’s got ideas? ED LAZOWSKA: More and more of
our students are going to be taking this course. Essentially, everybody wants
to take it, because they realize, it’s something
they’ve got to know. They’ve got to know
for the future. JULIETTE SCHWARZ: Having this
course really gives me an edge on my competition. And it’s absolutely
fascinating. CHRISTOPHE BISCIGLIA: Google’s
expertise is in building and maintaining these large-scale
systems. But we desperately need universities like the
University of Washington, and other institutions, to start
thinking about how to impart these skills on to students
earlier in the curriculum. ED LAZOWSKA: Now we’re looking
at a situation where Google and IBM, teaming together, are
going to deploy data centers that the entire national
university community is going to have available to it for
research and education. What began as one engineer’s 20%
project is now something that’s going to affect
the whole nation. It’s utterly remarkable
what you can do. MALE SPEAKER: Cloud
computing, baby. MALE SPEAKER: Cloud computing.

9 thoughts to “Academic Cluster Computing Initiative”

  1. This is awesome. Heck yeah for open source. I just went Linux yesterday for the first time with Ubuntu and so far so good!

  2. Actually that's a pretty cool initiative, but since that was years ago, I wonder what has come of those students that had this opportunity?

Leave a Reply

Your email address will not be published. Required fields are marked *