CHICAGOMichael Tiemann has been at the forefront of open-source software for years, but he wasn't fully prepared for what happened when he became chief technology officer at Red Hat Inc. in January. Tiemann had seen much change since helping to develop the GNU compiler and found Cygnus Solutions, but the rate of renovation was nothing like what he's seen so far this year.
As CTO of Red Hat, the leading Linux provider, he's both watching and guiding the development of Linux. To say the Linux snowball is rolling faster than ever after nearly a decade is an understatement.
"The [Linux] process started in 1991, and the industry was at version 2.2 in 1999, eight years later," Tiemann said. "People say all that happened in eight years. Now we're wondering what will happen in the next eight weeks."
One thing that seems certain to happen is that Linux will become a bigger player in the embedded-systems market. Tiemann said that the acquisition of Cygnus by Red Hat paved the way for that marketing shift. Gaining expertise in the GNU compiler will further enable Red Hat to compete in the embedded-systems arena.
"Cygnus and Red hat play closely together," Tiemann said. "Red Hat could not have legitimately entered embedded systems without Cygnus. Embedded systems is a tools-driven market, and Cygnus had the leading 32-bit tools solution."
However, users didn't need to wait for Red Hat's acquisition of Cygnus to start combining Linux with the GNU compiler. Linux creator Linus Torvalds himself used GNU when he was getting started.
"The GNU compiler is very important for Linux. It sets the basic infrastructure needed for something like Linux," Torvalds said. "The GNU compiler was pretty much there from step one."
Linux has achieved cult status in the past year, with Torvalds becoming so popular that he was recently on the cover of Money magazine. If there's any doubt that Linux has become about as fashionable as an operating system can get, Tiemann defers to the view of those arbiters of what's in: teenagers.
"My wife taught high school at a private high school, and students would ask her what her husband did. When she said I built compilers, she got a blank stare. Now, with a new crop of freshmen, she said her husband built compilers for Linux. A lot of the kids said that was pretty cool."
Earlier this year, the fragmented nature of embedded systems helped to create some concern that the Linux world itself would fragment. In February for example, Tiemann was worried that companies were developing proprietary technology based on Linux 2.2 and taking it in several directions. That's changed, partly because Lineo Inc. (Lindon, Utah) has purchased six companies so far this year.
"We saw that as a bad thing," Tiemann said. "I think the threat has diminished somewhat. Lineo seems to be buying every company with an incompatible version. That puts all the fragmented versions under one roof. Fragmentation is not the general marketplace problem it first appeared to be, but it is still a little early to say we've eliminated the whole problem."
For the most part, Linux developers have created an atmosphere where most do a fair amount of development work for the common good of all, with many engineers spending as much of their free time on Linux as they do on their job. This attitude is necessary, since there isn't any single player like IBM or Microsoft that can drive a market toward its concept of a standard.
"We're the largest company in Linux, but by no means do we have a majority of the market," Tiemann said. "The development of Linux is a little like a coalition government: You need 50 groups to cooperate, or the coalition collapses."
So far, that coalition seems to be doing well in embedded systems. The open-source business model eliminates one of the key concerns for embedded designers: cost. Their other key concern is usually size, which Tiemann said can also be addressed easily.
"Last year we were getting ready to do a demo for an embedded Web server, and we were trying to get our software down to 1.3 Mbytes. One of our engineers rolled up his sleeves and said, 'I maintain this, so I should be able to do something with it.' By Sunday morning, he had reduced it down to 300 kbytes," Tiemann said. "He left a note that said, 'more improvements are possible, but I have to go home.' "
Though Red Hat grew from 40 engineers to 400 last year, those programmers make up only a small portion of the individuals who are driving the operating system's evolution. Tiemann and his staff are shining the spotlight on programmers they think are capable of advancing Linux.
"One way we manage things is that we're trying hard to identify and support the thought leaders in the industry. That's somebody who can do more than prolific development; it's somebody who can influence someone to do things in a way that's constructive for the project. Thought leaders also help set standards so their Linux open-source project plays well with other projects. Linus Torvalds is an example. He's still very involved, particularly in Mobile Linux."
At Red Hat, the capstone for the embedded thrust is EL/IX, an embedded Linux application programming interface that is based on Posix. It makes it possible to use a desktop Linux system for application development, assuring that the applications can be used on the target hardware. This software too has seen a fair amount of development over the past few months.
"We received comments that the applications and run-time software were specified in Posix, so we're bringing EL/IX even more in line with Posix," Tiemann said. "At the same time, we're making it possible that Posix will be less abstract and more applicable to embedded Linux."
It's not only the rapid changes in the software world that provide challenges for executives like Tiemann. Red Hat recently suffered from the whims of Wall St., although it had been a darling of the industry during 1999. Even a quarterly report with increased revenue and decreased losses wasn't good enough for investors, and the stock price took a hit.
"We're beginning to see pressure to make revenues grow," Tiemann said. "One thing that's frustrating is the ease [with which] companies with half-baked business plans get funding. When you compete with noneconomic models, you either hold your breath or become noneconomic. We're not going to change our model, and we will meet our numbers. We've set expectations to do $85 million this fiscal year. We've told Wall St. that we expect to see profits by the end of fiscal 2002."