Weblog

Friday September 16, 2005

Curve Conspiracy You've probably noticed the stylish "Sun curve" in our new share campaign. I was in some of the internal focus groups that bantered about Sun's identity, mission and vision. Along the way, I sent a note to Ingrid VanDenHoogen, our VP of Brands and Stuff, including a profile shot that showed me in, well, a less than complimentary posture. I took creative license to claim inspiration, if not motivation and perspiration, for the S-curve. With all due respect to David St. Hubbins, there is a fine line between corporate and corpulent.

I'm waiting for Ingrid to up the ante by suggesting I appear on a Dove billboard.

| |

Live, from New York, it's Tuesday morning With a variety of people in the Big Apple for the Big Launch, we did a remote taping of Sun's NetTalk program. If you haven't checked out the NetTalks, you'll find them full of technical detail, practical hints, and frequently unexpurgated commentary. They are podcasts in larval form.

Our mobile studio included microphones, mixer, tape deck, producer, host, high quality headsets, and enough cable to fill a small conference room in Sun's midtown New York office. We were lucky enough to avoid sirens, walk-ins, paging announcements, and fire alarms for the hour we were putting audio bits on tape. It should be up on the web within two weeks, after editing and production. Overall, a great experience, highlighted by the opportunity to let Bryan Cantrill go off-script for a bit.

Our topic was application scalability, and touched on vertical versus horizontal scale, algorithm design, and threading models. Bryan was holding up the kernel memory allocator in Solaris as an example of an algorithm that scales with demand by adjusting its cache and resource pool allocations. "Holding up" is purely figurative in the audiocast; six months ago Bryan's description would be accompanied by one of my stick figure diagrams that make Shalom of Safed look like Chuck Close.

That was then, this is the open source now. As I pointed out in the NetTalk, if you want to see how the kernel memory allocator works, use the source, Luke. It's all there on the OpenSolaris website, ready to be downloaded and browsed. And perhaps copied.

This is a rather long preamble to answer one of the questions that came up in press and analyst discussions the day before: aren't we afraid that someone will copy ideas in our open source projects? That's one of the goals -- sharing code that scales means that developers are more likely to build more code that scales.

As Simon Phipps points out in a variety of open source discourses, "free" shouldn't stop at zero cost but imply zero encumbrance. Use it and learn from it and build something new and different (and scalable). We certainly hope that scalable code demands a scalable operating platform, and a scalable Java environment, and a scalable hardware system. That is, after all, why we had a packed house in midtown for our product launch. ( Sep 16 2005, 10:15:18 AM EDT ) Permalink Comments [0]

| |