Original URL: http://www.theregister.co.uk/2009/04/27/dziuba_sunacle/

Will Oracle kill MySQL? Who cares?

MySQL's already killing itself

By Ted Dziuba

Posted in Applications, 27th April 2009 03:13 GMT

Fail and You When Oracle bought Sun last week, the MySQL community collectively curled its lip into a worried sneer. Is Oracle going to kill MySQL? We'll have to wait and see.

Oracle's acquisition of MySQL isn't like a lottery ticket. There is no chance that MySQL will get significantly better by leveraging some of Oracle's technology. MySQL developers and practitioners are worried because the only outcomes of this uncertainty are either neutral or negative. Is this growth on my pancreas benign? We'll have to wait and see.

Oracle has a clumsy history when it comes to open source. In October of 2005, they acquired Innobase, the company that makes the InnoDB storage engine for MySQL. InnoDB powers any serious installation of MySQL because it supports transactions and generally acts like more of a grown-up than MySQL's default storage engine, MyISAM.

When Oracle bought Innobase, we heard the same kind of worries. Does this mean the end for MySQL in enterprise applications? It turns out that all of these concerns were unfounded. Oracle continued to support and develop InnoDB, without dropping the car keys into its pants and telling MySQL that if she wanted to get home, she's got to fish them out. They managed to keep InnoDB's dual-licensing structure: The product is GPL'ed, but if you want to buy a commercial license for it, you're more than welcome to.

After Innobase, Oracle bought Sleepycat Software in 2006. Sleepycat produces BerkeleyDB, which is the de facto standard for embedded databases. BerkeleyDB was also distributed under this dual license, which made more sense for embedded software than it did for something service-oriented like InnoDB. Oracle continues to develop and support BerkeleyDB, but uses it as a starting point to up-sell customers on its TimesTen embedded database and Oracle 11g.

Thus far, Oracle's open source strategy resembles a sixteen year old boy putting his complete mental capacity to work trying to figure out how to get high off of household chemicals. There's got to be some magical incantation of Drano, bleach, and kitchen spices that'll do the trick. Well, maybe not. Can we market our way into open source profit? Surely if we set a team of graphic designers and ad-men on the task, they can do better than the engineers.

As such, Oracle's next stab at open source was Unbreakable Linux. Unbreakable Linux is a distribution that's 100 per cent binary compatible with Red Hat, except it comes with a heaping helping of Larry Ellison's honey butter. From what I can gather, Unbreakable is nothing more than a marketing campaign around a Linux support program, the main selling points of which are:

So, considering Oracle's open source track record, what's going to happen to MySQL?

MySQL Suicide

It looks like they will continue to develop and support MySQL and use it to push their proprietary databases on larger customers. Oracle isn't going to flat-out kill MySQL, because MySQL is doing a good enough job killing itself. Case in point: the recent release of MySQL 5.4, which finally integrates significant performance enhancements from Google and others. It took so long for MySQL to accept the community's work that the OurDelta project sprung up to aggregate all of the third-party patches considered to be stable and necessary. As an open source project, MySQL is already a dinosaur.

In fact, it is in Oracle's best interest to keep MySQL moving at a glacial pace. MySQL's biggest technical weakness is scalability. Web engineers use MySQL because it's free, and when they hit a scalability wall, they need to engineer around it by building in database parallelism and sharding. MySQL's performance on a multi-core machine is embarrassing, whereas a database like Oracle really shines on big iron.

What software managers recognize - and often need to beat into their engineers - is that it is far easier and cheaper to throw money at a scalability problem than it is to throw engineering resources at it. In the case of web service providers using MySQL, when they hit a scalability wall, Oracle can now step in and sell them a database and Sun hardware that will fix the problem, no re-architecting required.

So, don't expect MySQL to die. Expect it to stagnate. Of course, the free market of open source will eventually develop something else. The Drizzle project, run by MySQL's director of architecture, is a fork of MySQL that aims to support better scalability. It's not complete yet, but it shows that some users and developers are unhappy enough with MySQL's main line to do something about it. If Drizzle gets stable or something else comes along, many web developers will move on to the New Hotness. With the attention span of a fruit fly, your average web programmer is quick to latch on to new technologies. After all, as a web developer, you can't appear publicly to be older than thirty.

However, even with bleeding-edge development, MySQL will continue to be the open source database leader. The Web 2.0, Ruby on Rails hipster crowd may move on to the next flavor of the week, but other business will still want to pay for help with the database they know and love. Larry Ellison understands that a small, vocal minority thinks they control the future of databases, but real-world business pragmatism is slow to change paths.

"I think every generation thinks that their set of innovations will quickly replace everything that came before," he said in a recent interview is Israel. "I think we constantly come up with new technologies. But those new technologies take a very long time to fully displace technologies that came before." Those sound like the words of a man who wants to sell you a high-performance database so you don't get caught up in open source trendyness.

Even so, Oracle has been relatively tight-lipped over the future of MySQL. While they have stated that the project is not going to die, the reassurances have been less than assuring, like a used car salesman telling you that he's going to give you a great deal.

If anything, this uncertainty is a test of the open source community. Years ago, it taught Larry Ellison hate. Now, Ellison is teaching it fear. ®

Ted Dziuba is a co-founder at Milo.com You can read his regular Reg column, Fail and You, every other Monday.