It's been almost four years since the OSDL was founded by HP, Intel, IBM, and NEC. According to the group's first press release, it would be "the industry's first independent, non-profit lab for developers who are adding enterprise capabilities to Linux. The four companies plan to provide significant equipment and funding to the lab over the next several years. Additional contributors and sponsors of the lab include Caldera, Dell, Linuxcare, LynuxWorks, Red Hat, SGI, SuSE, Turbolinux and VA Linux." (VA Linux is now VA Software, the parent company of OSDN, which owns NewsForge.com.)
During the past year, Linus Torvalds and Andrew Morton have both been hired by OSDL, allowing them to work full-time on the Linux kernel. Timesys, Unilever, Ulticom, Sun, Turbolinux, NTT Data Intellilink, NTT Comware, Wind River, Novell, and Network Appliance also joined OSDL in 2003.
We recently chatted with a well-known kernel hacker (under the condition of complete anonymity) who is suspicious and distrustful of the motives of the OSDL. His three major issues were money, the kernel development process, and control of Linux kernel development. After our conversations with the anonymous developer, we telephoned the OSDL and spoke with Tim Witham, the OSDL lab director.
It's all about the money
Issue: The OSDL has a budget of dozens of millions of dollars a year, supposedly for open source development, but only seems to employ between six and 12 developers.
Witham pointed out that of the 35 total employees of OSDL, 24 of them "are actively either developing or supporting developers." He also noted that a major portion of the OSDL budget goes to things like supporting the kernel summit and its Linux User Advisory Council.
Witham said, "Most large system companies have a group of big customers, and they get those people together once a quarter and they talk about where they are trying to go." The OSDL LUAC provides that sort of environment for firms that are not used to dealing with the open source community.
"They want to support Linux," Witham said, "Linux is important to them. But you know if your business is completely unrelated to IT, going out into this open forum is a little intimidating because you might cause yourself political problems."
A top-down bazaar?
Issue: Look at the Carrier Grade Linux (CGL) and Data Center Linux (DCL) specifications. Specifying things top-down like that seems quite contrary to the way open source development happens.
Witham admits that the specifications are top down, but insists the development work around them is not. He explains, "We're using the organic development of the community, figuring out which things have already been developed to meet our needs."
After the specifications are in place, as they are for CGL and DCL, Witham says, "Everybody volunteers. We don't get to assign anybody. I have no authority to tell IBM where to put their developers or HP where to put theirs. It is strictly based upon their corporate itch. What's the most important thing to them? They'll assign people to work on it. So in essence it is a top-down specification but it uses a bottom-up development methodology."
Issue: OSDL does not seem active in open source development.
"Every one of our member companies agrees that the way it [open source development of the Linux kernel] is working overall is really, really well. They've done a very good job. The only thing that has been an issue is there hasn't been a lot of 'we need this in two years' sort of planning."
Witham added, "[If] you look at the way that any new projects get fired off of CGL and DCL, it's very much the bazaar. People step forward for it. We are using that process."
Who's in charge here?
Issue: What's their motive? They say right on their Web site that they want to be the center of gravity for Linux.
"That statement is in direct response to certain operating systems companies claiming that there was not a center of gravity in Linux," Witham said. Steve Ballmer, CEO of Microsoft, is reported to have said in a secret internal memo leaked to the press last summer that, "IBM's endorsement of Linux has added credibility and an illusion of support and accountability, although the reality is there is no center of gravity or central body investing in the health and growth of noncommercial software or innovating in critical areas like engineering, manageability, compatibility and security."
Witham made the point during our talk that OSDL is not in any way trying to take over the Linux kernel or trying to direct its development process. Instead, Witham says that the OSDL attempts to bring new players from the enterprise to the game by providing them with a familiar and comfortable environment, whether they be open source-savvy, or even IT firms, or not.
He also noted, insofar as the question of controlling Linux development is concerned, that "Linus's contract with us states that he is the sole arbiter of what goes into the kernel."
As Morton and Torvalds see it -- page 2...As Morton and Torvalds see it
We asked both Andrew Morton and Linus Torvalds by email to comment on the primary issue raised: the distrust of the OSDL motives and fear of it providing the direction for Linux development.
Morton wrote back saying, "There has been no pressure at all. I can see
how there would be concerns however. There _are_ perceptual problems here,
and the only way I can see of minimising such concerns is via openness and
transparency in day-to-day kernel work."
As far as his being aware of unhappiness with OSDL by other kernel hackers, Morton drew a blank, except for one incident. He told us, "I know that when OSDL put out a press release to try to get more corporate testers of 2.6.0-test9, they received considerable flak from slashdotter types, asking 'who are you to go making such an announcement', which was all rather silly and unjustified. Nothing else comes to mind."
Torvalds, after telling us to come back when there are more than one or two kernel hackers complaining, said, "Basic rule of thumb: on the internet, anybody can complain. And people do. Whether it has any basis in fact is another matter. It's easier to come up with rumors about evil-doers than it is to debunk them.
"It's the first I hear of it. That's _especially_ true since I certainly haven't gotten _any_ technical direction from OSDL management at all, and my work agreement specifically states that they can't force me to do
anything and that I have final say on the kernel."
He continued his response on the issue of undue influence from OSDL by writing:
The only technical questions I get from OSDL boiled down to "How can we
help you?", and "Is this useful to you?".
[ Well, to be complete, I should mention the PR stuff, of course: a lot of
people end up contacting OSDL to get me to do speaking etc., since they
have noticed that emailing me about speaking requests doesn't get any
reaction ;). Happily, my contract also says that OSDL can't use me for
public appearances, so I still go to just the conferences I want to go
Of course, maybe I need a tin-foil hat to protect me against those OSDL
mind control rays. They are somewhat weakened by the distance to Oregon,
but I'm told off-the-record that "Big Brother 2004" has improved the ray
focusing and target tracking capabilities significantly.
A view from outside
To gain perspective from an observer outside the open source community, we turned to a well-known analyst, Dan Kusnetzky, vice president, System Software for IDC. Dan observed:
Stuart Cohen, the new CEO of the lab, recognized that the open source community is really a community of communities. He appears to have come to the conclusion that the members of each of these communities focus primarily on the engineering tasks related to their project. They, for the most part, are not really interested in the functions typically offered by a supplier's marketing department, legal department, and sales department.
I tend to agree with Stuart's view that members of these communities don't really value the results of these supplier functions. So, these things either are not part of the project or are included in a rather unsystematic way. This means that many open source projects have comprehensive engineering plans but really don't consider those other issues. Since the open source community members focus more on the functional capabilities of their project and not the overall business environment, the community can be ill-prepared for an outside attack, such as that coming from the SCO Group, and respond slowly or in a fragmentary way.
Executives in medium and small organizations, on the other hand, expect certain things of their suppliers. These expectations require a supplier to have comprehensive plans and processes to back up those plans. If open source software is going to find its way into the software portfolio of these organizations, someone must take up the tasks of market analysis, gathering user requirements, determining what functions are needed to fulfill those requirements, and encouraging the community to work on those functions.
It appears to me that OSDL is moving to the center of those activities in the hope of filling the void that is there now.
As Andrew Morton pointed out, the problem with the OSDL seems to be one of perception, not reality.
The OSDL is doing exactly what it is intended to do: bringing new resources and new users to Linux. On the other hand, as Thomas Jefferson once said, "The price of freedom is eternal vigilance." So it never hurts to take a closer look to ensure free software remains free.
David Graham and Robin "Roblimo" Miller also contributed to this report.