Then I first heard about IBM Corp.'s Linux Technology Center, my ears pricked up because Austin, Texas, was regularly mentioned in conjunction with it. I pictured the LTC as part of IBM's complex of buildings on and around Braker Lane in north Austin. That impression was so firmly entrenched in my mind that I contacted IBM and asked about visiting the center and interviewing some of the company's top Linux geeks. Big Blue gently informed me that I was mistaken; the LTC is a virtual center, not a physical one. Still, my curiosity about the LTC remained high. This week, I'll present what I've learned about the LTC since then.
Many of the LTC's key players are my neighbors: for example, Sheila Harnett, the LTC technical lead; George Kraft of the Linux Standards Base; and Steve Best, who leads the team porting IBM's JFS (journaled filesystem) to Linux. But of course, the LTC doesn't pique my interest solely because of its connection with Austin. After I spoke recently with Dan Frye, director of the LTC, I began to see the LTC not just as a cool thing happening in my neighborhood, but as hard evidence of IBM's commitment to open source and free software in general, and to Linux in particular.
This is not hype. This is not an ad on national TV proclaiming that Microsoft software "plays well with others." This is IBM the behemoth, the legacy megacorporation, the king of punched cards, and once the monarch of monopolies, making a positive contribution to the open source and free software community.
All about LTC
The LTC was founded in August 1999. Frye said the approximately 185 IBM employees who make up the LTC are located in 16 cities and 6 countries. As Frye put it, the LTC is "a real place -- it just resides on the Web." But as the old IBM marketing line used to go, the billion-dollar internal budget for Linux at IBM this year is just the tip of the iceberg.
So what does the LTC actually do for Linux? Frye observed that in the open source community, everyone basically works on what they are interested in. LTC employees are not free to hack whatever Linux-related code they want, but the LTC works on aspects of Linux that are of interest to IBM.
You can visit the LTC homepage to find out exactly what those aspects are. (See Resources for a link.) The LTC page has references and links to many different Linux-related projects. Frye cited scalability, serviceability, reliability, test, systems management, and journaling filesystems as being important to IBM. That list is long, but not all-inclusive.
The LTC team consists of people from nearly every part of IBM. Frye is obviously proud of the quality of his team. He told me, "We have built a group in the LTC that really takes some of the best from a number of different IBM groups." Those groups include Sequent, OS/2, Tivoli, AIX, and S/390. The LTC isn't very old, but has already made important contributions to Linux.
Did you read about the recent joint effort by Oki Data and IBM to provide Linux drivers for Oki Data printers? That falls under the umbrella of the Omni printer project, which provides GPLed drivers for nearly 300 printers; 8 months ago, that number was only 50. Did you read or hear comments about the size of the latest patch for the 2.4 kernel? That's because of all the S/390-specific code that went in -- courtesy of the LTC.
I mentioned earlier that IBM is "making a positive contribution" to the open source community. I'll give an example. IBM has worried about the scheduler at the heart of the Linux kernel. Some of IBM's code for the scheduler was accepted, but the company wanted to make it perform better on high-end machines. Linus Torvalds has rejected much of IBM's work on the scheduler because of concerns about performance at the low end. That creates an opportunity for a head-on collision between a giant corporation and the leader of the Linux hackers.
Is there trouble or conflict brewing as a result? Not a bit. IBM has a project in plain sight on SourceForge to rewrite the scheduler from scratch. According to Frye, the scheduler project is "one of the projects that the kernel development conference next month is invited in to hear about, and I think that the meritocracy will apply to this project as well, in that if we show that we have a good design for the scheduler, that it improves scalability [and] doesn't substantially impact performance in a small number of processes, we'll get accepted." Call it enlightened self-interest or whatever you prefer, but I like IBM's approach. Especially because in the past few months, the trade press has raised the specter of IBM "hijacking" Linux for its own purposes.
What else is on the LTC event horizon? How about the old bugaboo that gives our founding editor, Nicholas Petreley, such frustration: standards for Linux. Frye said that programming standards will arrive this summer. You don't think Linux can prevail without structured testing? The LTC is big into test, and is currently assembling a team that will work exclusively on open source test cases for Linux and will continuously test versions of the kernel. Frye said the team will tackle things that are not currently tested in a structured way, such as mixing application workloads, stress testing, and peak workload testing.
Regardless of IBM's motives -- whether the company sees Linux as an opportunity to steal potential customers from Sun or Microsoft, or free software in general as an inevitability better dealt with sooner than later -- I have to give Big Blue credit for the work it is doing to make Linux run faster and jump higher. And for doing so as a positive member of the community.
What do you think? Do you trust IBM? Do you think it provides a model for large software companies' participation in the free software community? Let me know by posting your thoughts in the Linux forum or writing me a note.