Since assuming the mantle of leadership of MySQL AB in 2001, Mï¿½rten Mickos has steered the database software company from open source darling to open source powerhouse, multiplying revenue some twentyfold in a very short time. On the eve of the release of MySQL 5.0, Linux Magazine spoke with Mickos about the unique challenges of running the decidedly unique — and very successful — software company.
Since taking over as CEO in 2001, Mï¿½rten Mickos has been the business face of MySQL AB(http://www.mysql.com/),the company behind the very popular open source database, MySQL. It’s a job that, judging by the companies that use MySQL, should have put Mickos’s face on the cover of every major business magazine in the world. With more than six million deployments in total(according to the company),MySQL has become the go-to data management tool for giants Google, Yahoo, Nortel, the New York Stock Exchange, and a host of other high-profile clients.
Yet MySQL’s commodity status is both a reason for the database’s popularity and a reason that Mickos’s company, a privately-held operation with 200 employees worldwide, keeps such a relatively low profile. Although the company has had an enviable growth rate — 2004’s $20 million in revenues are supposedly a twentyfold increase over what Mickos encountered his first year — such growth relies heavily on forces outside the control of MySQL AB’s leadership. Like most thriving open source companies, MySQL AB depends on the good work of outside developers and the good faith of young companies eager to exploit the freely-distributed GPL version of MySQL, yet equally willing to pay for the commercially-license non-GPL version once business success compels an upgrade. However, deftly mixing the machinations of open source and a mandate to generate and grow revenue is a tricky balance that MySQL AB has mastered.
Linux Magazine senior contributor Sam Williams interviewed Mickos in between keynote speeches and sales conferences and on the eve of the release of MySQL 5, a significant new release that moves the database and the company closer to rivals Oracle, IBM DB2, and Microsoft SQL Server.
LINUX MAGAZINE: Since MySQL is a Scandinavian company, here’s a Scandinavian-themed question: Between the 8th and 10th centuries, the Vikings dominated Europe by mastering the forces of wind, tide, and current. To dominate the software marketplace of the 21st century, MySQL has to master the forces of open source software development. What have the last few years taught you when it comes to relying on third-party ideas, talent, and innovation?
M?RTEN MICKOS: First, let me say that the Vikings weren’t as brutal as some make them out to be. It turns out that the Vikings were actually friendly, except for a few whose actions created the myth.
LM: My apologies to our Viking readers.
MM: As for MySQL, we’ve realized since the start that innovation happens elsewhere. If you want to win today, you must acknowledge that other guys are smarter than you and that you must work with them. You must build a strong ecosystem to support you.
LM: But what about that issue of relying on forces outside your control? It’s sort of like a 19th century approach to business, trusting the wind to move cargo from port to port.
MM: That notion of trust hasn’t changed. When a company grows as fast as we have, it looks like we did it because we’re so smart. But we should be a little bit more humble and realize that there are strong undercurrents taking the software industry in certain directions. If you didn’t have trends such as commodization, standardization of software, and Moore’s Law — it wouldn’t matter how smart we were.
LM: Do you think the sailing metaphor explains how businesses make money off open source?
MM: That’s a good way to put it. You put trust in things that are outside the company. Initially, we were crazy to do so, because there was no evidence. But now there is evidence that, even if there’s not a strong wind blowing on all oceans, there’s always wind blowing somewhere. And yes, you need to navigate to the waters where there are good sailing winds, and if you don’t, you’re suddenly not moving.
LM: The company website describes MySQL as the “M” in the “LAMP” stack, “LAMP” being the integration of Linux, Apache, MySQL, and Python, PHP, or Perl. All of these components are freely-available and outside the control of a single company. That’s good if you’re a small company looking to set up shop as a vendor of LAMP-related software and services. As your company grows larger, however, does that lack of control become a constraining factor?
MM: That’s a good question. I don’t know the answer just yet. When I had a keynote at the[ 2004 Australia] Open Source Developers Conference, I argued that a self-regulating stack evolves faster and produces more value than a tightly-regulated stack. With a tightly-regulated stack, where a single company controls the entire stack, the stack can only improve, even by a small degree, if that company makes the improvement. For every problem, a sole vendor must come up with the solution.
When you live within a self-regulated stack, on the other hand, it’s more of a Darwinian process. There is no “LAMP consortium.” We don’t have LAMP summit meetings in New York where the “L”, the “A”, the “M”, and the “P” decide what will happen in the next year. There’s no such[ body]. You could ask how much time does Redmond spend in such meetings, defining what .Net should be? How much time does Sun spend deciding what Java should be? And here comes LAMP, where nobody seems to be in charge, which means that everybody is in charge. Everybody takes responsibility.
A year ago we said, “Why are there no business intelligence solutions based on the LAMP stack?” Now there are. You’ve got Jaspersoft (http://www.jaspersoft.com/) and other companies taking care of that problem. It’s a great thing. You just stand up and say, “Hey, here’s something we think the stack needs,” and the stochastics of the community ensure that it’s developed. You never know by who, but you know that somebody will.
LM: So your company is really putting its faith in the “stochastics of the community” — the notion that somewhere out there a developer or development team is working on what you need.
MM: Yes. It’s risky and it takes some give and take, but if you compare it to working within a single company, you see the advantages. Let’s say you send an email to your entire company, asking, “Can somebody help me with this thing?” Typically, if you do that within a company, nobody answers. Everybody thinks, “OK, somebody else will answer the question. I don’t need to do anything.” So you can ask the question, “How the heck does it work in the open source community?”
LM: OK. How the heck does it work?
MM: I don’t know (laughs). I have some ideas. In the open source community, there’s a very strong element of shame and pride. If you do[ something well], you get pride and credit for it. If you do it poorly, you get shame. In closed source, for one thing, you rarely get any credit because you’re hidden somewhere in a big organization. Secondly, if you work on a closed source product and do a bad job, you also get no punishment for it. There’s no penalty for creating bad code.
Inside MySQL AB
LM: Your role at MySQL AB is business and sales. What do you see as the strengths and weaknesses of the engineering culture in the company?
MM: Let me start by saying that I am very strict in making sure MySQL doesn’t become a divided company. I don’t want it to be a company where one part is just engineers and open source and the other part tries to make money. I want the engineers to try to make money, and I want sales and marketing to love open source. And I think we’ve succeeded there. We’ve had lots of controversy, but it never splits into two camps.
LM: Can you give an example of what might cause controversy?
MM: Well, say we discuss the licensing. Should we go from the LGPL to the GPL on the client side? Some of are sales people will say yes and some of our developers will say no. It’s not clear cut, and it never breaks down along those lines.
LM: In 1999, when I first encountered MySQL, I think my reaction was common: “What Linux does in the operating system realm, MySQL is doing in the database realm.” Six years later, that “Linux of databases” tag seems a bit awkward. For one thing, most CTOs I talk to see Linux as a means to an end. They see it as a better platform for the type of scalable, stable, IP-native applications you need to run a business nowadays. Given its position in the “LAMP” stack, MySQL is closer to the Java applications that are drawing customers to choose the stack. Do you think the “Linux of databases” tag is damning MySQL with faint praise?
MM: We haven’t used it. Maybe we used it in 1999, but since I joined in 2000, we haven’t used it. One reason is that we run on Windows and Unix as well as Linux. We love Linux and that may be where our closest affinities are, but[ MySQL database designer] Monty[ Widenius] developed MySQL on Solaris. We would much rather be the “Intel”, “IKEA”, or “Dell” of databases — truly high-quality, commodity software that everyone needs.
LM: How does MySQL 5.0 figure into that? I’m taking particular note of[ MySQL AB cofounder] David Axmark’s quote, in which he told ZDnet that the development team was “fixing 10 years worth of criticism in one release.”
MM: I think it’s a relevant question when people ask us, “Why are you adding the features that you were famous for not having?” I think that software has come to a maturation point, a level of standardization that you can add those features without adding complexity. We think of it as a challenge. We thought, “We’ll show all of you six million devoted users that we can develop these features.” And we’re doing it.
But if you’re a very modern company using us, a company like Google or Yahoo, you won’t look so much at new features such as stored procedures, triggers, and views. You’ll look at the Federated Storage Engine and the Archive Storage Engine, the things that help Internet companies keep up with massive growth in data and distributed architectures. We make less noise about those features, but there are some truly outstanding, modern things that aren’t there for us to catch up with others, but are there to help us serve our modern users.
LM: So what type of customer is excited by stored procedures, triggers, and views?
MM: Stored procedures and triggers are required by those who build business applications involving multiple queries, but it extends all the way to a C|net, for instance, who has asked us to do stored procedures for a long, long time. Views are needed by anyone trying to make the database easier for non-technical users to use. It’s a very useful, modern feature.
A Very Bright LAMP
LM: What do you see as the big driver of LAMP stack adoption?
MM: I think there are two areas where there’s huge growth. One is new or renewed web properties that are starting to grow rapidly. So the big ones like Google and Yahoo are growing, but even more so are the small ones, the Facebooks, Friendsters, and Flickrs. They all run on LAMP.
The other that’s great to see, though, are the big corporations that are saying, “Hey, we’re building our new departmental applications on LAMP. We’re building our Internet on LAMP.” That’s a new trend that we haven’t seen before. This verifies our assumption that the enterprise market will reach out to the Web market and say, “Hey, you Web guys, how did you do this? Can you show us as well?”
LM: If you look at Circuit City, where they’re merging the in-store sales support system with the online sales system, is that a common story at the enterprise level or are they well ahead of the curve?
MM: Both. It is becoming much more common, but we don’t see every customer engaged in such an effort, especifically the very conservative ones. I’m sure you can travel somewhere in the United States or Europe where you’ll find a whole city with companies that have never touched the LAMP stack. But that’s just the way it goes. Big changes never happen overnight, and it’s not a big change if it does happen overnight. The only exceptions I can think to that rule are Skype and Google.
LM: In the company literature, you’ve mentioned MySQL’s affinity for the so-called “second wave” Internet startups, the Friendster and Facebooks. What is the source of that affinity?
MM: It’s all about lower costs. If you read Joe Krause’s blog[ http://bnoopy.typepad.com/bnoopy/2005/06/its_a_great_tim.html], he points out that when he founded Excite, he needed $3 million. When he founded his latest company, JotSpot, he needed only $100,000. And it’s thanks to LAMP. That’s what’s keeping these companies alive in the bad times.
On the vendor side, you’ve got a handful, maybe not quite ten companies, that have done open source for a long, long time, and know how to stay afloat. Here I’m talking about companies like TrollTech, Sleepycat, db4objects, Funambol, and Digium.
LM: The companies you mention are all very tightly-linked to a particular software offering. Sleepycat is Sleepycat. TrollTech is Qt. You’ve got the main business people and the main developers under one umbrella.
MM: Exactly. But you can go either way. Look at Apache. It’s the most successful open source project ever if you go by installed base and marketshare, but nobody ever made any money on it. Then there are the companies that founded a strong software project, but went too commercial. As a result, they killed the community arm and the business as well.
Building a Better Company
LM: What’s your perspective on Mozilla’s comeback via Firefox? That seems to be the rare case of a development team coming up with market-expanding concepts such as simpler features and heightened usability.
MM: I think that’s a demonstration of what we in Finland call sisu. It means “stamina” or “persistence.” I think there were background forces who were keen to see a new browser and were willing to support those guys, whether quietly or openly. And that’s great. It’s a wonderful combination when the big, influential guys give some money or some publicity or help, and there are a bunch of good developers who make it happen. I think it’s a good story. It’s a great story, because of what happened to Netscape and how they failed in their attempts to open source source their code. With Firefox, a small group of people just never gave up.
To me it’s similar… are you familiar with the “Stockdale Paradox”?
MM: Admiral[ James] Stockdale was a prisoner of war in Vietnam. He spent seven years in prison there, and was tortured and everything. He had a similar attitude. He said, “OK, it will get worse. This is not the worst. Things will get worse tomorrow, but one day I will get out of here.” So on the one hand he was ready to take any punishment and any pain. But on the other hand, he never gave up the long term dream that he would get out.
That’s what I think is so important with entrepreneurialism: that you are stubborn enough not to give up even when it looks really, really bad, but that you’re also naive enough to believe in a positive outcome. If you’re a plain pessimist, you’ll never survive. But also, if you’re a plain optimist, you’ll never survive.
LM: We’re drifting back to Darwin. You just killed off all the optimists and pessimists, leaving only a few optimistic realists to repopulate.
MM: Yeah, or at least those who carry this paradox within them, the ability to be both optimistic and pessimistic at the same time.
LM: How does this “second wave” phenomenon affect the amount of money flowing through the industry? If companies can make do with $100,000 in the first year or two, I’m guessing MySQL doesn’t make too much money from new companies.
MM: No, we don’t. So we have to have the faith that there is good business down the line. We’re looking and saying, “Wow, this is the fastest developing segment of the industry, so let’s make sure people are using MySQL.” What do we care if half of them die after a year? Those that survive will become paying customers eventually.
LM: I recently wrote a story that quoted Microsoft’s former in house open source evangelist.[ You can find that story online at http://www.linux-mag.com/2005-10/microsoft_01.html.]
MM: Steve Walli.
LM: Yes, and he mentioned SAP’s upgrade of Adabase and how SAP eventually licensed that upgrade[ SAP DB] under the GPL and handed management duties over to MySQL. He says that while he was working inside Microsoft, executives following SAP’s progress couldn’t make sense of the move. First of all, they couldn’t understand why SAP would embrace the GPL. Secondly, they couldn’t understand why SAP would give the resulting product away. Walli says he eventually saw it as a way to shield SAP’s proprietary offerings from the GPL and that your company, MySQL, was used to dealing with the GPL and was basically offering an escrow service for the software. Is that a fair characterization? If so, it suggests that MySQL’s core competency might be project management, not software services.
MM: It doesn’t completely resonate with me. Usually I agree with Stephen, but not[ here]. There is some nervousness around the licensing in the industry, but I think it will go away. But I do think the model is otherwise relevant.
There’s a need for a commercial channel for open source products. You can look at Southwest Airlines and Jet Blue, who are very profitable and good companies, but when big companies try the same, when United created Ted and Delta created Song, it didn’t work. You have to have a very specific corporate DNA, one geared towards low cost, agile, fast decisions, and working within your chosen ecosystem. That’s why I think it made sense for SAP to hand over SAP DB and then Max DB to us, not because of the licensing issues.
LM: Walli definitely framed it as a strategic move. Given Oracle’s dominant position in the database market, it would have taken years to come up with a rival relational database. By putting SAP DB under the GPL, the company gave its customers something they could use, along with the promise of speedy improvement. They were basically innoculating their most loyal customers, boosting their immunity to the Oracle sales pitch.
MM: That’s right. That does make sense. But it was not because they were afraid of the GPL.
LM: Unlike Microsoft.
MM: Don’t assume anything about Microsoft. They have open source software and will have more. If Bill Gates started his company today, he would have started an open source company. That’s my belief.
LM: Why do you believe that?
MM: Because we are all children of our times. When Bill Gates started, he picked the business model he could see around him. When we started, we picked the best business model we could see.
LM: What lessons does the more free-wheeling database/business intelligence marketplace offer to the operating system marketplace? If, as you say, Microsoft comes to embrace open source more and more, are we going to see a more dynamic marketplace? The enterprise software marketplace is definitely consolidating, but it still has enough growth to support multiple players.
MM: I don’t think it’s a market unless there are multiple players. That’s why monopolies are so bad. I don’t mind so much that the monopolistic player gets all the money. I just mind that the development and the evolution of the market slows down, and nothing happens. And you can see that happening today. How has the Windows desktop improved over the last 10 years? Not much.
LM: It’s got a lot more stuff in it.
MM: (Laughing.) Moore’s Law has made sure those machines are a thousand times more powerful, but they’re just a thousand times more complex for the users to use. They’re not productive. The most productive use of a PC is for email and “skyping.”
LM: Speaking of complexity, how do the added features of MySQL balance with the simple- and easy-to-use reputation that MySQL has cultivated so far?
MM: Very well, because it’s such a modular design. If you don’t use stored procedures, it’s just one #ifdef within the source code. You can even compile MySQL without stored procedures if you really want to get rid of them.
LM: So it’s sort of choose what you need and dump the rest?
MM: Very much so. But you’re right. Every time we add features we have to compromise between ease of use and richness. For instance, a user complained that because MySQL installs without a superuser password, a default MySQL installation is open for attacks. That’s a vulnerability. So then we said, “Should we require everybody to set the password?” Well, that would make it more complex and difficult to use, so we had to balance between security and ease of use.
LM: Does MySQL 5.0 require any change in marketing strategy? Are you looking at customers that, a year ago at this time, you just couldn’t reach?
MM: Certainly, there are new customers that we couldn’t reach previously. In terms of marketing, though, we don’t want to change[ the message] too often. When you sell a commodity product, it’s very important to have a clear focus that you maintain for many, many years, because if you start flitting from side to side and trying all kinds of different buzzwords and positions, you play yourself out of the market.
GPL: Benefit or Baggage?
LM: As the company grows, does the GPL become a challenge? I mean, if the bulk of your growth is coming mainly from the younger, smaller, more agile companies, do you have to devote more resources to ensuring compliance?
MM: I don’t think we have a license compliance issue. I think we have a good relationship with the community. People always narrow down on the gray zone and say “Why do you interpret the license this way or that way?” In reality, though, the 95 or 99 percent of all use cases are absolutely clear and the customers or users have no questions whatsoever. It’s an interesting scientific experiment to figure out where the borderline is between this or that interpretation, but it doesn’t really have any business significance.
LM: Are you looking primarily at increasing your marketshare within the open source database market or in the general database market?
MM: Our view is that open source is not a business model and it’s not a market. Open source is a production method. Therefore, it doesn’t make sense to compete in the open source database market. There’s a market for commodity databases, and there we are absolute winners, irrespective of whatever kind of source code the others may have. That’s how we view it. We have grown rapidly and we are a proud representative and sometimes an icon of the open source community, but we have a long way to go to be really big. I mean, the market is $15 billion.
LM: Going back to Darwin, the one thing that stands out when you look at evolution is the equilibrium. Things are constantly changing, but globally or historically, the ecological niches stay the same. At what point does this stability become a trap? Young companies rely on growth to reward both investors and employees. Is there a point where the open source ecosystem impedes that growth?
MM: Complacency can hit any human being. I don’t think there’s a glass ceiling in the market, and, yes, we need to be active and remember to get outside our comfort zones and question what we are doing. But on the contrary, I think open source will provide business opportunities for the next 50 years or 100 years.
LM: What about the companies that aren’t in the open source community?
MM: Well, I don’t own them, and I don’t run them. So I don’t worry.
LM: Is MySQL looking to join Google and other “second wave” Internet companies on the Nasdaq?
MM: There are no immediate plans. We think we have a good business model. We think we have what it takes to be an independent company. We have venture capitalists on board, who, at some point, will need an exit. Does that then lead us to an IPO at some point? Of course. But you shouldn’t write that we’re focusing on an IPO. We are growing the business, and an IPO is just one of many steps when you grow your business.
LM: What are the company’s immediate strategic goals for 2006?
MM: The strategic goal is to serve the customer very, very well.
The next question is,” How do we continue to serve our customers?” We have to continue to deliver 5.0 and its follow-up releases. We have to continue developing strong partnerships. People will take notice when many independent software vendors support us and that it’s becoming easier and easier to integrate MySQL with existing software.
Sicily for the Taking
LM: Before wrapping things up, I’d like to going back to my original Viking analogy, because it seems like a flipside to your Darwinian analogy. As you noted, the Vikings were mainly a peaceful people. Periodically, however, they would get on the ships and go out looking for new places to conquer.
MM: I think it was the nasty guys who went out on the ships and the nice guys who stayed home.
LM: True, but even the nice guys seized the opportunity when it was there. I just wonder, when do you go from farming your plot of land to saying, “Oh my God, Sicily is just sitting there for the taking.”
MM: (Laughs) That’s an excellent question. I think we are becoming bolder, if that’s what you’re asking. We have been carefully planting our seeds and growing our gardens and growing our crops and making sure we have a strong society. We’ve built our buildings so they can withstand storms. And we will be building ships for a long, long time. And you’re right. We do smell fresh, new winds that will take us out on new escapades.
Sam Williams covers business and software technology for a number of publications and is a regular contributor to Linux Magazine. He is the author of two books, Arguing A.I. and Free as in Freedom: Richard Stallman’s Free Software Crusade.