MySQL presents a textbook case for executing a winning Open Source business strategy with as much brute energy as the Norse creation myth. Before there was Asgard and the Rainbow Bridge, fire fused with ice to bring forth life in an abyss draped in eternal twilight. In the land of man, the furnace of relational calculus tempered by cold bottom-line logic forged the rising B-school legend, MySQL AB. Start with a painstakingly practical Swedish database company; put a maniacally dedicated Suomen samurai at the helm; and you have the makings of an epic success saga.

Leading that business strategy is Mårten Mickos, a marketing workaholic that knows his technology and  is an outspoken champion of Open Source. He was appointed CEO last year in May to bring business growth to MySQL AB.  For the vast majority of MySQL's staffers, work is done from their virtual office at home: For Mickos, the virtual office suite consists of planes, trains, and hotel rooms. On one of his trips last month, Mickos sat down with Open in an extended conversation spanning the current state of database technology to the future state of Open Source in the IT market.

On the Open side, we began by stating the obvious: Being an Open Source company is not easy. Most companies try to start with  proprietary software business plans and then attempt to reconcile these plans with GPL software. Far more often than not, they fail abysmally. So how does MySQL AB succeed in taking business away  from Oracle and Microsoft? In the the words of Mårten Mickos, "Whatever we do, we do exactly the opposite of what's expected."

“You can kill as many
 Open Source companies
 as you want, but you cannot stop Open Source."

—Mårten Mickos, CEO MySQL

Open: We don't get it.  For many early Open Source companies, marketing was often an overwhelming part of their message. Yet for MySQL, marketing is almost nonexistent. In the dark past, a lack of marketing spelled doom for many companies with superb technology, Ingres being a classic example in the database space. So how is that  MySQL succeeds while apparently repeating the mistakes of the past?

Mickos: "Our marketing is primarily marketing communications. We have one PR company in the Bay Area of San Francisco and that’s all the marketing we do. In the company, we have about 60 people, who are almost all techies: mostly software developers, a few sales people and some admin people. Not one, however, has a marketing title. So 99.9% of all that is written about MySQL comes from outside of the company and we have no control over this.

For example, we don’t write white papers: We let others do that. When we give speeches, we create them by ourselves with such low quality that everyone knows that we wrote them. At Linux World, we had a big success because Scott McNealy said we {Sun} embrace MySQL. He just said that in an interview and suddenly journalists from all over were calling us to ask what we were doing with Sun and what it means.

In our business model, we have an ecosystem around MySQL, which is huge worldwide. As a company, we are very small in the middle of it. This is very different from the traditional proprietary model. In the past, if you had a fantastic product like Lotus did with Lotus Notes, then you owned or controlled as much as 60% of the surrounding ecosystem, including services, books and education, yourself. We don’t mind that our ecosystem is 200 times larger than our company is. That’s why our cost level is ridiculously low. When it looks like we’ve done something, it’s usually someone else who did it.

Of the installed base of relational databases, we now have more than 20% of the installations. In terms of market-share dollars, however, we have only 0.02%. So naturally, you have ask what’s wrong here? Of course, the answer is you can use MySQL free of charge.

The absolute fantastic thing with this business, which I am very thankful for is that Open Source is a rising tide. Nobody can stop it. You can kill as many Open Source companies as you want but you cannot stop Open Source. The underlying force that nobody can stop is the growth of Open Source. When Oracle says MySQL is not a threat, we say, 'That’s fine if you think that way.‘

Look how long it took for Michael Dell to turn that company into an empire. If you look at how many years it took, you realize that big things don’t happen over night. We have patience and patience has virtue. We can wait 10 or 15 years. Just in the past few month we have gone from 20,000 to 27,000 downloads a day on average. Today we are not a big dent in their business, but over time we definitely will be. It’s better to take time and do it right.

If you look at the database market there are really three big players: Oracle, IBM, and Microsoft. They command 86% of the market. No one has been able to break into that inner circle in the last 15 years. Many traditional software vendors have tried, but they have all failed.

Open: Are you breaking into that circle? How does a small company take on a resource-giant like Oracle in product development?

Mickos: We are doing it today by going the other way around. Whatever we do, we do exactly the opposite of what's expected. If you are to enter and change an old mature industry, you have to do it differently. If you go there and say our product is better and it doesn’t cost as much you never win. That’s the sad truth. You never win unless you are a leader.

The current economic climate plays to our advantage. Companies can’t and won’t spend $3 million on an Oracle installation. MySQL on the other hand does 80% of the work and costs nothing. There are various specialized data types and management features that we don’t integrate into MySQL. Oracle has some XML integration that we don’t do. We think XML integration is mostly hype at the present and not worthwhile because it’s not mature enough.

So our client companies take part of the $3 million that they save and develop the extras that they really need themselves. The money they spend on doing part of the work is only a small fraction of the total that they save.

We follow the principle of the 14th century philosopher monk Occam. We follow his rule known as Occam’s razor: No complexity beyond what is necessary. It advises that the best choice to explain or model a system is always the simplest. Occam's razor helps us "shave off" concepts or constructs that are not really needed. So at MySQL, we do nothing beyond what is absolutely necessary to provide a very fast database. We add no bells, whistles, or extra surprises that might introduce ambiguities or inconsistencies in our data model. For the user, that would only add unnecessary complexity.

With that philosophy and because we are an Open Source company, we have no one who is tasked to sit around and think about secret future developments. Everything we are working on is open to our customers and on our web site. We just concentrate on making our small piece of the MySQL ecosystem really, really, work well.

Open: How does that ecosystem actually work?

Mickos: We do our product development by letting people in the ecosphere develop add-ons. When evolution has done the work of making sure that they are good products, we can then license the add-on and sell it as part of MySQL. This is what happened with the table handler InnoDB.  This allows us to offer full transaction support as a feature in the new MySQL Pro version 4.0. With InnoDB, MySQL Pro provides row-level locking, full support for transactions, and foreign key constraints.

We’ve also enhanced security for database traffic between the client and the server in the new version of MySQL. Web developers have been able to use SSL to secure the traffic between the end user browser and web application written in PHP, Perl, or any other web development tool. In MySQL 4.0, the mysqld server daemon process can itself use Secure Sockets Layer (SSL), thus enabling secure traffic to MySQL databases from applications residing outside a physical firewall.

We’ve also added a query cache that is quite unique for boosting performance. Today most databases are accessed over the web. Often over the web, a user will get anxious and repeat a query. In MySQL, the query cache checks if the database has been changed and if it has not been written to, MySQL will not repeat the entire query process on the database engine but rather satisfy the query from cache. This also aids scenarios where there is a large number of users making the same queries over and over again.

Following speed, our second most important feature is stability. Even when we release alpha code, there aren’t any reproducible bugs that we know of. In fact only after we go through alpha, beta and gamma stages do we declare new code as stable."

Open: How  do you go up against proprietary databases like Oracle, DB2, or SQL Server?

Mickos: When we sell commercial licenses for application specific databases, a company will simply build MySQL into their product. For example, a backup program may use MySQL to hold all of the metadata associated with all the backup tapes resident in a tape library. Here we’re like an industrial OEM for the car industry.

This market is very fragmented and there are over 30 database vendors from which to choose. Nonetheless, here we have traditionally seen Oracle and now Microsoft with SQL Server is becoming a bigger player. The problem for SQL Server is that it runs on only one platform. If you are an ISV, building a product that is limited by its underlying database is a dangerous gamble. You will want your product to run on Windows, on Linux, on Unix, on Mac.

Among the very high-end corporate or government IT buyers like NASA, The Census Bureau, we constantly encounter Oracle, IBM and Microsoft. I need to be careful when I say we compete and take business away from against Oracle every day. We are not taking their core business but the business they never deserved to get. Remember, only 30% of Oracle’s revenues come from new database licenses.

Typically we are dealing with companies who have believed in standardizing on just one database and are now beginning to understand that it’s costing them millions to use one proprietary database in all applications. They now realize that by migrating just one third of their basic applications to an open source database, they can save millions of dollars. That’s where we take business from Oracle.

Another interesting part of the ecosystem where we have no part of the revenues is the growing presence of database ISPs. ISPs are making huge profits providing MySQL databases on an outsource basis and our license does not require them to pay anything to us. If there is anyplace that we are leaving money on the table it is with ISPs. The Free Software foundation is looking at this problem for the next version of the GPL, but they’ve been looking at the commercial use of Free Software for some time.

Open: How important is the revenue stream from commercial licenses to embed MySQL into other software products?

Mickos: The fact that we’re profitable is very important. If we were not, then people would be saying that it is too good to be true and they would not trust us. When we say we’re making good money, they trust us more.

Open: What characterizes your best prospective clients?

Mickos: Anyone who listens to their developers and economists.