An Interview with Marten Mickos of MySQL
MySQL is a superstar among open-source databases. There are more than ten million MySQL installations worldwide, and it is used by top sites like Google, Yahoo, Slashdot and Travelocity. It is particularly strong among the Web 2.0 crowd, counting among its customers Wikipedia, Craigslist, Del.icio.us, Digg, Flickr and Technorati. MySQL was created in 1995 by two Swedes and a Finn: David Axmark, Allan Larsson and Michael “Monty” Widenius. Since 2001, when the MySQL AB company was set up in Sweden, its CEO has been another Finn, Marten Mickos.
GM: I believe you knew one of the founders of MySQL a long time before you joined?
MM: I met Monty, the CTO, in 1981. We enrolled at the Helsinki University of Technology to study Technical Physics. Monty at that time was not an open-source developer, he was just an alpha geek. And I thought he was spoiling his life by not going to the parties and having fun but just working and programming all the time. But, he did build some amazing games and stuff that we played with.
When he started MySQL, I worked for this other small database company, Solid Information Technology. I told Monty that his project was just going to fail, and that it was a stupid thing to do, and that he didn't have a chance because we had a chance.
GM: What was your view of the Free Software world when you were at Solid—were you even aware of it?
MM: I was getting more aware of it, and I was getting excited about it. At Solid, I drove an initiative of not open-sourcing the product, but making it very popular on the Linux platform—and that was why I was an advertiser in Linux Journal, because we were the leading Linux database in the world in 1996. We gave it away free of charge, so we had taken a step in that direction.
Then Solid decided to cancel the project and just focus on high-end customers, and that's when I left the company. So in that sense, when I got to MySQL, I had some unfinished business. By that time, I had completely bought into the notion of code being open.
GM: What attracted you to MySQL?
MM: Now it sounds like we're reading some great wisdom into my decision, and I don't think there was any. MySQL at the time was not administered at all as a company. There was virtually no bookkeeping. There were no offices, no contracts, nothing. So in that sense, it was a garage startup and a big mess. But I knew about the enormous potential of the technology. Monty has saved an e-mail I sent to him in '97 where I said—I'm referring to some URL—“hey guys, you seem to be getting some traction.” And that's the first time I admitted MySQL had a future.
GM: Unusually for an open-source project, all the copyright of the source code is held by the company MySQL AB. Where did that idea come from?
There are some natural reasons for it. One is that the vast majority of the original source code was written by one man—Monty. Now his portion is much, much smaller, but at that time, most of the code was written by him. So it was natural that the copyright was held by the company. But second, Monty and David learned from the Ghostscript project. They were the first implementers of the dual-licensing model where you retain copyright but at the same time you release it under open source.
GM: Why did the company decide to adopt the GNU GPL in 2000?
MM: Initially, they had another dual license that said it's free on Linux but you pay on UNIX and Windows. And at some point, they realised to get included in the Linux distros, you needed a license that people could readily accept. People had nothing against the MySQL license, but it took time for them to read through it and accept it. And they argued that if they would adopt the GPL, there would be no questions asked.
When they made the decision, monthly sales fell to 20% of what it had been. So it was a huge risk financially for them—they had no financial backers, no VCs. There was a half year of slower sales and then they were back on track.
GM: You still have a commercial license alongside the GNU GPL. For what reasons do people choose the commercial license?
MM: The interesting thing is that we are known for the dual-licensing model, and as pioneers of it, but today our main business is not on dual licensing, because we are now becoming a major player in the enterprise market and with Web sites, and they don't buy commercial licenses from us, they buy subscriptions.
GM: You mean they use the GNU GPL license and pay for support?
MM: Yes. So dual licensing was a good starting model for us, and it works well in the OEM space, where people “OEM” the code from us and put it into their own products that they ship to customers. And that's where it works very well. But if you look at our most famous customers, like Google and Yahoo, Travelocity and Craigslist, they do not use our commercial license.
GM: You have some very high-profile customers. What do they use MySQL for?
MM: With ten million installations worldwide, we're used for everything that relates to data. We're used for structured data, unstructured data, transactional data, non-transactional data. We're used in Web applications and business applications.
Take one example, Google. The system for its commercial ads, AdSense and AdWords, those two run on MySQL, so when you get ads popping up on your Google screen, you know we are there.
GM: What about Yahoo?
MM: They started by using it in Yahoo Finance, where they built a publishing system called Jake. All the news items and whatever content they publish came out through Jake and MySQL databases. And from there, it spread out to many of the gaming solutions and hundreds of applications within Yahoo.
GM: And Travelocity?
MM: There, MySQL is used for the airfare searches. So if you make an airline reservation, it still goes into the same HP NonStop [SQL] database that they've had there for some time, but all airfare searches go into our databases. Interestingly enough, it is the airfare searches that grow exponentially. There aren't too many seats being sold, because there aren't too many airplanes being flown today. But to make one reservation consumers can make tens, hundreds or thousands of searches first. So it shows a change in the landscape; it's not just the travel agents and the professionals who make very specific airline reservations and searches, it's everybody.
GM: What other kind of applications run on your database?
MM: Slashdot runs on MySQL. The Spiderman movie site runs on MySQL. The special effects in Lord of the Rings were built using MySQL. The Mars Rover has an earth-based control program that runs on MySQL.
GM: Do you think that the use of the LAMP stack—GNU/Linux, Apache, MySQL and Perl/PHP/Python—has become almost a given for a Web 2.0 startup?
MM: I think that's correct. In the Internet bubble, many companies had the thinking that they needed tons of VC money, and with that they bought Sun hardware, Oracle databases and BEA Web application servers. And today, you don't do that. You buy inexpensive hardware; you run the LAMP stack on it; you just get going. And then when you start scaling, that's when you need commercial help. So I think the interesting thing today is that you can start small, start on a single Intel-based server, and it costs you virtually nothing. And then, when you get going, you can scale it horizontally, without throwing away the original.
GM: Does that mean MySQL is not really up against Oracle as a competitor—that you tend to go for new companies?
MM: I would put it differently: they are not up against us when it comes to Web 2.0; we are among the pioneers there, the leaders there.
GM: What about in the traditional markets, do you find that you are starting to compete against Oracle?
MM: We do, but it's not a main area of focus for us. This is the major difference between us and the other open-source databases. Most of the others are trying to become a replacement for Oracle, so if you look at PostgreSQL, EnterpriseDB, Ingres and all those guys, they try to mimic the old-style databases so that they one day can claim that space. But my guess is that by that time, the space will be gone.
GM: What effect did Oracle's purchase of Innobase, which supplies InnoDB, one of the main database engines for MySQL, have on you and your customers when it was announced last year?
MM: I think it sent shock waves through the industry. People took it very seriously, especially the financial analysts and journalists. And even our customers saw it as a risk, and they came to us to ask what this meant and whether they were safe or not. And I think what we have shown in the last six months is that open source is such a self-healing ecosystem that if InnoDB truly had been taken out of the equation, there would quickly have been replacements. And there are replacements today.
GM: Where did MySQL's pluggable architecture, which allows different database engines to be used, come from?
MM: It was a smart design decision by Monty back in '95. He had built the first MySQL engine. He realised he needed to revise and upgrade the storage engine. But he was lazy, so he didn't want to move over abruptly from one to the other. So he thought, what if I allow both of those to coexist at the same time? And when he did that, he had to create an API between the upper layer and the lower level. He didn't know at the time what a fantastic design decision it was.
In Web 2.0, the usage of data is much more varied today than it used to be in the old client-server world. If you have a big Web site, you have some data that is transactional, you have other data that is read-only but is needed in milliseconds, and then you have logging and archiving data that you typically don't need [immediately] but which needs to be available somewhere. By using different storage engines, you can cater to those various needs within the same database installation.
GM: What was the thinking behind your decision to work with SCO at a time when it was taking legal action against IBM that was seen as threatening to the Open Source world?
MM: We are not supportive of SCO's legal actions, and when they ask us for advice, we tell them to stop it and just get out of it and ask for forgiveness. We don't share their thinking there, but they have customers who need a database. Why wouldn't we sell our stuff there? With the money we get, we can hire more developers to develop more, cheaper software.
I think it's so easy to be black and white, but if you think twice, you realise this could be the best way to deal with the situation. Because now SCO cannot go out and say open source is bad, because they just bought a database license from us. Of course that won't change the litigation, but every little step counts.
GM: In March 2006 you joined the Eclipse foundation. What took you so long?
MM: That's a very relevant question. We just don't know how we could be asleep at the steering wheel like that. We should have joined a long time ago. It's just when you get too caught up in your own stuff, you don't act fast enough. But it was just wrong, we should have joined earlier.
GM: Moving on to corporate matters, at what points did you take funding?
MM: We did that in 2001, when I joined, and we got a professional board of directors, but it was only 4 million Euros. They said, this is exploding in our hands and we need to grow it, but we also need funding to grow it properly. So I helped them raise the first round. Without even having decided to join, I just said, I'm helping these guys.
And then in 2003, we raised our next round, 13 million Euros, mainly from Benchmark Capital and Index Ventures. And then early this year, we did a series C round, although we hadn't even consumed the previous round—we still had plenty of money left.
GM: Investors obviously expect something back at some point, so are you looking to get bought or to do an IPO?
MM: We're aiming for an IPO. We're actually aiming for an independent existence and to do that you need to do an IPO, but the IPO is not the aim, the IPO is just a step. People ask, “What is your exit plan?”, and we say that we're not going to exit.
We think that when markets mature they tend to go horizontal, so you have players who specialise in certain components of the stack. Intel is a fantastic example on the hardware side. They produce processors for all vendors in the world, and nobody has acquired Intel. And, it makes sense for them to stick to their knitting and focus on what they are good at. We think that the database has a similar role in software—that it makes sense to have a team dedicated to data management: storing data, retrieving data, sorting data.
GM: Would you contemplate broadening your portfolio to include non-database products?
MM: I don't think so. I think we are fairly certain that we would not go into applications—that's for our partners to do. I don't think we would go down the stack into operating systems. But, I can see us being fairly innovative when it comes to dealing with data. Traditionally, a database was just a database. Then you had databases with replication. Now we have databases with different storage engines, and maybe you'll have databases with backup solutions and databases with storage solutions. So there's a world of expansion opportunities without having to go into applications.
GM: Looking at the broader open-source sector, do you expect there to be more consolidations in the wake of Red Hat's acquisition of JBoss?
MM: A few years ago, the common discussion was that open source is capable of competing with Microsoft and the closed source vendors specifically, because it isn't concentrated in one company, but it's a best of breed of a group—who's the enemy when there are so many? So, it was seen as a strength of open source. Now the winds are slightly different. People say Red Hat has grown so strong and look, they have acquired JBoss, but I think this discussion will go from side to side. One occurrence of an acquisition doesn't mean that there has to be more of them.
GM: Moving on to threats to open-source software, I notice there's a “No software patents” sticker and link throughout your Web site. How dangerous do you think software patents could be?
MM: We think software patents are the biggest threat not only to open-source vendors but also to closed source vendors. And not only to vendors, but also to users, because software is being developed in bigger volumes by users than by vendors. The patents that are now being granted are so silly, so detailed, on such a low level, it is just inevitable that there will be enormous conflicts once the owners start thinking they must get a payback for the money they spent on acquiring them. We don't think it's specifically an open-source problem; we think the open-source companies and open-source people are the first to see the problems. It will harm the whole industry.
GM: Are you actively talking to people within the European Union on this subject?
MM: All the time. We have been surprisingly successful so far. We have had campaigns with poor funding but great results; whereas the pro-software patents camp has had great funding and poor results. But, it's a very difficult time because they come back every year with new proposals. So I'm very proud of what we've achieved so far, but I'm actually fairly pessimistic about the situation.
GM: What about in the US? Are you working to fight software patents there too?
MM: Not as much as in Europe, because in Europe, the legislation is still being written, whereas in the US it already exists. But it works both ways. In the US, they already see so much of the trouble with software patents, so there's a stronger movement against them, and Europe is still sleeping.
GM: Do you see any other major threats to open source?
MM: No. I think it's just a superior production model. Whatever happens with legislation or licenses, nothing can stop it when it's just inherently a superior model. People ask about GPL 3—when will it come out and will it be good and will people use it? It's an interesting question, but it doesn't affect the future of open source that much. There will be open source no matter what.
GM: Against this background, what do think will happen to Microsoft?
MM: They'll ultimately become an open-source company. I've never met Bill Gates, I don't know whether he's conservative or not, but I would actually expect him not to be. When he started Microsoft, he did the best thing he could, so why wouldn't he do it again? Open source wasn't available as an opportunity back then, so he couldn't choose it. If he started a company today, I bet it would be open source.
Glyn Moody writes about free software and open source at opendotdotdot.blogspot.com/.
|Designing Electronics with Linux||May 22, 2013|
|Dynamic DNS—an Object Lesson in Problem Solving||May 21, 2013|
|Using Salt Stack and Vagrant for Drupal Development||May 20, 2013|
|Making Linux and Android Get Along (It's Not as Hard as It Sounds)||May 16, 2013|
|Drupal Is a Framework: Why Everyone Needs to Understand This||May 15, 2013|
|Home, My Backup Data Center||May 13, 2013|
- RSS Feeds
- Dynamic DNS—an Object Lesson in Problem Solving
- Making Linux and Android Get Along (It's Not as Hard as It Sounds)
- Using Salt Stack and Vagrant for Drupal Development
- New Products
- A Topic for Discussion - Open Source Feature-Richness?
- Drupal Is a Framework: Why Everyone Needs to Understand This
- Validate an E-Mail Address with PHP, the Right Way
- What's the tweeting protocol?
- Designing Electronics with Linux
- Kernel Problem
2 hours 32 sec ago
- BASH script to log IPs on public web server
6 hours 27 min ago
10 hours 3 min ago
- Reply to comment | Linux Journal
10 hours 35 min ago
- All the articles you talked
12 hours 59 min ago
- All the articles you talked
13 hours 2 min ago
- All the articles you talked
13 hours 3 min ago
17 hours 28 min ago
- Keeping track of IP address
19 hours 19 min ago
- Roll your own dynamic dns
1 day 32 min ago
Enter to Win an Adafruit Pi Cobbler Breakout Kit for Raspberry Pi
It's Raspberry Pi month at Linux Journal. Each week in May, Adafruit will be giving away a Pi-related prize to a lucky, randomly drawn LJ reader. Winners will be announced weekly.
Fill out the fields below to enter to win this week's prize-- a Pi Cobbler Breakout Kit for Raspberry Pi.
Congratulations to our winners so far:
- 5-8-13, Pi Starter Pack: Jack Davis
- 5-15-13, Pi Model B 512MB RAM: Patrick Dunn
- 5-21-13, Prototyping Pi Plate Kit: Philip Kirby
- Next winner announced on 5-27-13!
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?