MySQL's open-source nature cannot guarantee its future all by itself.

Oracle claims that it doesn't matter who owns MySQL because it's open source and everybody can use it for free and develop it further. They say so to deny that there's a problem. If they really thought so, they wouldn't put the whole Sun deal at risk because of MySQL and Sun would never have paid a billion dollars for MySQL in the first place. Then everyone would just do a "fork". Oracle is already using Open Source like Linux and Apache. If the ownership doesn't matter, why does Oracle want to buy MySQL instead of just forking it.

There are some community leaders, open source advocates and some generally well-meaning people who also believe MySQL is safe under whatever owner because it's open source. Some of them do so as a matter of conviction. Some are closeto Oracle and Sun. And most of the people simply haven't looked at MySQL's success formula. The formula that makes it much more successful than any other open source database.

MySQL's database server has traditionally been a product developed and maintained by a single company, not a community project depending on volunteers or on multiple vendors (like Linux). Every time someone developed an essential enhancement, the company behind MySQL ensured (through a copyright assignment) that it had the right to use such enhancements not only under the GPL but also under any other license.

That approach enabled MySQL to implement its highly successful dual-licensing business model. Those who wanted to incorporate MySQL's code into other GPL'd projects were free to do so ("if you are open source, we are open source"). Those who wanted to incorporate it into or enhance it with closed-source products without having to publish their entire derived work on GPL terms (which would practically preclude them from many commercial opportunities) could optionally purchase a commercial license (identical software but different rights and obligations). In that case, MySQL or Sun said: "If you are commercial and make money with it, we are commercial and want to make money with it, too."

Not only the dual-licensing business depended on that principle but also the fast-growing "MySQL Enterprise" subscription business. That one consists of three types of components: MySQL (as available under the GPL anyway), services (that one could also buy separately, including from third parties) and tools. The most exciting tools, such as the Monitor and the Query Analyzer, were made available on a proprietary (non-GPL) basis even though they are integrated so tightly with MySQL itself that any third party distributing such tools would also be bound to the GPL. Without those tools as a key differentiator (a premium offering to justify a premium price), there would still be a subscription opportunity, but it would be much less profitable.

MySQL AB and later Sun also used their trademark and copyright ownership to steer and increase MySQL's distribution and popularity. There are uniform standards worldwide for official MySQL certification, enabling professionals to prove their knowledge in a way every customer understands. There are authorized MySQL partners around the globe for services and consulting. There are official MySQL books. There is, very importantly, a so-called FOSS exception to simplify MySQL's integration with non-GPL free and open source software, a key success factor for having MySQL bundled with certain operating system distributions, programming languages and tools.

None of this is meant to downplay the importance of the vibrant MySQL community. Many millions of users have spread the word, have through their use of MySQL contributed to quality assurance, have developed software related to MySQL -- but for MySQL's core and for the development of powerful storage engines, there's empirical evidence that it takes real businesses with a serious revenue opportunity to make it fly.

We've been asked many times why MySQL can't just succeed on a GPL-only basis like Linux. There are many reasons. To begin with, applications based on MySQL must be made available under the GPL (unless a commercial license is granted by its owner). That is not the case for Linux, which has a license agreement that comes with an exception (that is labeled a "clarification") that allow you to run any application on top of Linux without them beeing affected by the Linux GPL core. So for an example, if someone wants to embed MySQL into a navigation system, GPL rules apply, but for Linux that would usually be no problem at all. Apart from this fundamental difference in licensing rules, MySQL never had and probably never will have the kind of support from a company like IBM (which has its own database business to protect).

There's always talk about a "fork", meaning an independent, derived version of MySQL. The GPL would allow that to happen but would not allow it to work out successfully. Not only would the "fork" start without the famous MySQL name but also would the "fork" not be usable by everything, for example with closed source applications, plugin or storage engines which are part of the current MySQL ecosystem

Richard Stallman (RMS), the father of the GPL and founder of the software freedom movement, notes that as MySQL, (being a large infrastructure product), cannot just depend on GPL-based community contributions to be successful.

Our petition is flexible. Every supporter can choose to mark one, two or all three of the possible ways forward as acceptable. The GPL is a great license if MySQL is in the hands of a well-meaning owner. The GPL gives maximum control and with a good owner that is desirable while with a bad owner that must be avoided. So a divestiture (spin-off) to such a buyer is the first option and then MySQL could -- it even should -- continue to available under the GPL (maybe GPLv3 instead of GPLv2, but at any rate, GPL) so that a good owner has the power to steer it in the right direction.

It would also be possible to leave MySQL per se under the GPL but to have a "linking exception" in favor of applications, plugins and embedding (of libmysqld). And a change to a permissive license like the Apache Software License 2.0 would give licensing-related flexibility, enabling someone to start a new GPL-based project including it or to develop it under the Apache license or whatever other license, even including the integration of the code into proprietary products.

Liberalizing the license terms would mean that Oracle as the new owner of MySQL could not use its position to lock in some of MySQL's customers and partners. If Oracle doesn't continue MySQL's development in a really good way (which is not just a matter of how much money is spent but of exactly how it is spent), it will lose control because others then have not only a legal possibility but also a real opportunity to take care of MySQL customers in their full diversity and of MySQL-related innovation. For a few years, the market would certainly miss the presence of MySQL the way we've known it so far. Temporarily, Oracle would get rid of its fiercest competitor. But after a while, it could come back similarly strong as before, maybe after several years even be stronger than ever.

One way or the other, nothing can be considered a solution if MySQL's future depends on Oracle's good intentions. "Trust me" type of promises don't help in any way. Every kid learns that fact from popular fairy tales. Regulators should know better.