"I will do everything that is possible to keep MySQL alive," says Michael (Monty) Widenius, the main developer of the well-known open source database. "I just hope it's enough."
Widenius is talking against the background of Oracle Corporation's acquisition of MySQL along with the other assets of Sun Microsystems. According to Widenius, if allowed to stand, the acquisition will mean the eventual loss of MySQL as a first-class database without any adequate open source replacement.
Just as important, he sees the acquisition as bringing the GNU General Public License (GPL) into disrepute and the exclusion of free and open source software from the consideration of anti-trust laws. And as the acquisition gradually gains approval throughout the world, the chances of avoiding these results is appearing increasingly remote.
MySQL, the Takeovers, and the Resistance
Disagreeing with development decisions, Widenius himself left Sun eleven months after the purchase to start Monty Program AB, a non-profit company that develops MariaDB, a community-centered MySQL fork.
Then, in April 2009, Oracle announced its purchase of Sun -- and, with it, MySQL. The deal won U.S. antitrust approval in August 2009, and was approved by the European Union in January 2010. Russian and Chinese anti-trust regulators have yet to approve the acquisition, although Widenius expect them to announce a decision in the next couple of months.
In December 2009, Oracle issued a media release promising to continue developing MySQL as free software for the next five years. But Widenius points out that this promise is not binding in any way.
In fact, Widenius has opposed the acquisition in every possible venue. He has frequently opposed it on his blog, and started two petitions against the acquisition, which he says have collected a total of 40,000 responses. Together with Richard Stallman and others, he has also signed a formal letter to the European Union objecting to the deal.
Although Widenius began by urging that Oracle not be allowed to acquire MySQL, more recently he has concentrated on obtaining legal guarantees from Oracle, either through an appeal of the European Union decision or through the disapproval of Russian or possibly Chinese anti-trust regulators.
Widenius' opposition to the deal has made some people suggest that he and his supporters have delayed the deal, causing Sun Microsystems employees increased stress and uncertainty.
However, Widenius denies this accusation, saying, "We are not the ones who are setting the timetable. It was Oracle that dragged out things by not responding and not providing remedies. The only thing that I have tried to do is get Oracle to accept some proper remedies that are good for everyone. And I don't think it's the wrong thing to do."
Similarly, Widenius denies being motivated by an attempt to receive more money for his former interest in MySQL or to make Monty Program AB profitable by placing himself in the spotlight. He has invested much of the 16.6 million Euros he received from Sun's purchase of MySQL in Monty Program AB, and frankly admits that the company has next to no chance of making a profit. Nor is he interested in selling Monty Program unless he can find a buyer with open source's interests at heart.
Instead, he claims to be opposing Oracle's acquisition out of a sense of personal responsibility: to his employees, to his customers, and to the larger open source community.
Widenius admits, too, that his personal pride is at stake. "I'm trying to save something that I've worked on for a long time. Actually, I'm trying to do something good for the community and open source, just as I have done for the past fifteen years. But people don't understand the concept that some people really believe in open source, and are trying to provide good software for everyone. I have been doing that, and I would like to continue doing that."
The impossibility of a replacement
According to Widenius, everyone in the free software community should question Oracle's acquisition because the corporation is an untrustworthy guardian for MySQL. Oracle, he says, is likely to neglect MySQL, eventually leaving the community scrambling for a replacement.
Widenius points out that, in the past, Oracle has acquired free software databases such as Innobase and Berkeley DB and made promises to keep them as active community projects. Instead, it has brought development in-house, slowing their development.
Now, Widenius predicts a similar fate for MySQL. Already, he says, future versions have been delayed or cancelled, and the MySQL newsletter is increasingly focused on inter-connectivity with the Oracle product line. Although Widenius was unable to compare the number of patches to the code since the acquisition to those of previous years, he did note that many people had left the project, and that Oracle had already reduced the community development team from three to six. All these signs suggest that MySQL will be treated the same as Innobase and Berkeley DB.
Widenius foresees that MySQL will be tied increasingly to other Oracle products with a series of proprietary features. Gradually, "They will market MySQL as an entry database that you can use until you afford something real." And after five years, of course, Oracle will be no longer be bound by its news-release promise to keep MySQL alive as a community project.
Under this scenario, Widenius worries that free and open source software will be left without a replacement for MySQL, especially on the Web. The fact that MySQL competes with proprietary databases today is not enough, he says -- it has to continue to develop to remain a contender.
After all, he notes, "If you're a big company, you don't want to buy a car that you can't buy spare parts for. You would be better off with SQL Server, because Microsoft has a team of more than a hundred people working day and night on it, compared to a product that has no one working on it."
Nor is PostgreSQL, the other leading free software database, a contender. "I think PostgreSQL is doing an amazing job, and it's a good database, but PostgeSQL was never designed for the Web environment. It doesn't have replication and a lot of other features that are needed for the Web."
Even counting on a fork (an alternate development of the code) of MySQL is a risky alternative -- regardless of whether it is Maria DB. "To do a fork, you need a huge coordinated effort. You need something like five million dollars a year to pay for developer salaries. And in the whole history of MySQL, we never got a community. We were open and we tried to get help, but a community doesn't just appear and start to do something" -- especially in a field that requires such expert knowledge as a database.
Moreover, assuming that the recent discovery that 75% of all Linux kernel contributions are made by paid developers, a fork would need to convince several companies that it was worth their investment.
Still, Widenius considers that the community's best hope is "some big companies stepping in to say, 'We don't want Oracle being dominant in the database. So let's fund some fork.' It can be mine [Maria DB] or someone else's -- that's the only thing I can see in the long term. But it's not very likely. But I'm still hoping that some people who don't want to depend on Oracle will be willing to do that. I mean, that happened with Linux; a lot of people started to fund Linux. But in Linux's case, it was easier because they can get direct benefits from doing that. With MySQL, it's much harder, but it's still a remote possibility."
The Future of the GNU General Public License and Anti-Trust Laws
However, even if replacement software is found or created, Widenius suggests that the Oracle acquisition could have even larger implications for free and open source software than the removal of what many consider an essential application.
Some commenters, like Eben Moglen, have argued that, because the GNU General Public License guarantees the availability of the MySQL code, Oracle's acquisition is of no concern.
But Widenius argues that this view is short-sighted. Being released under the GPL was not enough to save ReiserFS, a popular partition format, when Hans Reiser, its main developer, was found guilty of murdering his wife. Instead, development on it faltered and gradually ceased. "The GPL," Widenius concludes, "doesn't assure that someone will step up and be a leader and take over a project."
Even more importantly, Widenius worries that, if Oracle's acquisition is allowed to stand unchallenged, then people will conclude that the GPL is not a license to use. The fact is that, unlike the more than 65% of free and open source software that uses the GPL, MySQL is essentially a programming language. That makes MySQL harder to develop than a desktop application like OpenOffice.org (which Oracle also acquired with Sun), but most people will probably not make the distinction.
"They will just think about MySQL being under the GPL, and the GPL not being enough to save MySQL. I that case, it may mean the end of the GPL -- not because the GPL is a bad license but because people [will not remember] that MySQL is a library."
Yet the most ominous implication of leaving the acquisition unchallenged is that, by ignoring the arguments brought by Widenius and his supporters, the United States, European Union and other jurisdictions will have concluded that free and open source software is not to be considered in anti-trust cases.
"That means the laws that you have for commercial products, that you can't buy a competing commercial product to kill it, doesn't apply to open source," Widenius warns. "In effect, that means it's easy to kill an open source product. It means that all the protection that people get from anti-competition laws don't apply to open source. And that's a bad thing, and that's what scares me the most."
In effect, Widenius suggests, Oracle will have provided "a blueprint for legislators to say that, if anyone buys an open source project, they don't have to care." Where copyright and patent have failed to undermine free and open source software, the laws governing corporate acquisitions just may succeed.
Hoping for the best, preparing for the worst
Widenius is continuing to fight, but a hint of desperation was in his comments when we spoke. Although he expects to be involved in an appeal of the European Union decision, he is already making plans for the worst.
Asked how people can support his efforts, Widenius replies, "They can contact Oracle, and try to get some binding promises for the future of MySQL, saying that these public ones are not enough.
"If they don't trust Oracle, then I'd say that MariaDB is their best bet, and I would recommend that they should test MariaDB and do what limited things they can to ensure that MariaDB will get the features they need in the future, either by being a part of it or paying someone to develop the features that they need. If we can create a commercial ecosystem around MariaDB, we have a small chance to save MySQL. But I would still say that it's against all odds."
Should those efforts fail, Widenius hopes that MariaDB can at least be a resource for the community. "It doesn't work for everyone; I mean, commercial applications can't use it," he admits. "But at least we can ensure that we can keep something available for five to ten years. If we can get some sponsors, we can do more, but our ambition is to make MariaDB available for everyone else. For KDE and everything else [that currently relies on MySQL], we'll put some resources into helping them develop for MariaDB.
"The question is just: will the open source community be enough? I mean, we are spending 100,000 Euros a month just to keep our organization alive and do a limited [feature] set for MySQL. But someone has to pay, and it can't always be the open source community."