Andy Konwinski

**Stephanie Gerson** (0:03)
So hello and welcome to The Graduates, a radio show dedicated to graduate student research here at Berkeley. My name is Stephanie Gerson. I'm a graduate student myself, and I'll be your hostess for the show here on KALX Berkeley. So today, I'm talking to Andy Konwinski, a PhD student in computer science. So welcome, Andy.

**Andy Konwinski** (0:25)
Thank you very much.

**Stephanie Gerson** (0:26)
And we're going to be talking about measuring and improving the performance of programs that run on supersized computer clusters. So first, I know that you work in a lab of the computer science. It's called the RAD Lab. So can you first tell us about what that is?

**Andy Konwinski** (0:44)
Sure. Yeah. So the RAD Lab, it's an acronym. It stands for Reliable Adaptive Distributed Systems. And it's kind of the systems lab in the computer science department here. And what that means is that we're working on the kind of the core part of computer programming. And now with... And that's kind of traditional systems. So like kind of building your operating system on your computer. And now we've kind of taken a step back and started focusing on building the operating system for a large number of computers. So like data centers or... This is where we take a large number of small computers and have them all talk to each other. And the RAD Lab then is focused on allowing a very small number of people to manage and write software that runs on a large number of computers.

**Stephanie Gerson** (1:34)
And that's what you refer to as a super-sized computer cluster. I like that. Okay, so in the RAD Lab vision statement, it says that your vision is pretty much what you just described. But to enable one person to invent and run the next revolutionary IT service, operationally expressing a new business idea as a multi-million user service over the course of a long weekend, which you refer to as Internet Fortune 1 million. And I like that too. So what the RAD Lab is doing is enabling the Fortune 1 million to manage these super-sized computer clusters.

**Andy Konwinski** (2:18)
So kind of the origin of that story and that tagline for the RAD Lab is a lot of these stories about one college student, like the guy who made Napster or the guy who made Facebook or the guy who made eBay. And what they do is they sit down for a long weekend and they come up with this new idea that catches on like wildfire. But the first iteration of their program tends to become so popular so fast that they end up rewriting their application like seven or eight times. So eBay is on like it's seven or eight three right now. And that's from scratch just because they hit these kind of plateaus where their servers will crash because the application wasn't written to scale very well. And so the more users they get, the more they have to rethink, well, now the way we were doing this before isn't going to grow fast enough. So what we're trying to do is enable somebody with a new idea to write it once from the very beginning and then it'll scale with them as they grow.

**Stephanie Gerson** (3:13)
And where does the ability to scale come from?

**Andy Konwinski** (3:16)
I guess the difference is that it's actually kind of in the framework that sits underneath the code they write. So they were using kind of more traditional tools, like traditional programming languages, something like PHP. A lot of people use it for web programming. And now we've got all these world famous computer science professors sitting down and putting their brains into the project. And maybe we can still write PHP, but the kind of the underlying frameworks, like the operating system basically interprets that code, is handling it differently. So the same PHP used to have run on one computer. Now without the program even knowing it, it'll run on tens or thousands of computers.

**Stephanie Gerson** (3:54)
Wow. Okay, so who is your target audience for the research that you do at the Rad Lab? Well, someone who's wanting to start a revolution over a long weekend.

**Andy Konwinski** (4:08)
We meet with Facebook and Yahoo! and Google. Our founders, like kind of the people who funded us the most, are the three big sponsors of our lab, our Sun Microsystems, and they do a lot of software, and Microsoft and Google.

**Stephanie Gerson** (4:25)
And you're going to make your work available on your website for the littler people?

**Andy Konwinski** (4:31)
Yep, yep. So it will be available to download everything. I think we'll eventually give the URL for the Rad Lab, and if you go there, you can find links to all the projects that are currently being worked on.