As the European Commission continues toand its open-source MySQL database, that delay will harm MySQL without helping competition.
One who shares this concern is. On Thursday, Mickos sent a letter to Neelie Kroes, the European Union's competition commissioner, urging that the deal be approved for the good of the market and MySQL. He also on Thursday.
Below is the edited full text of the letter.
Helsinki 8 Oct 2009
Mrs. Neelie Kroes
Commissioner for Competition
European Commission, J70
Dear Commissioner Kroes,
I am writing to you regarding your review of Oracle's pending acquisition of Sun Microsystems. As I understand it, the EU Commission is concerned about a risk of undue concentration of power in the database market. Having been the CEO of MySQL from 2001 to 2009, and built a business that was serving a new market unmet by Oracle and others, I can agree with the questions posed, but I do not share the concerns that have been expressed. In the following, I will explain why.
In brief, my reasoning is as follows:
- Oracle has as many compelling business reasons to continue the ramp-up of the MySQL business as Sun Microsystems and MySQL previously did, or even more.
- Even if Oracle, for whatever reason, would have malicious or ignorant intent regarding MySQL (not that I think so), the positive and massive influence MySQL has on the DBMS market cannot be controlled by a single entity--not even by the owner of the MySQL assets. The users of MySQL exert a more powerful influence in the market than the owner does.
Many expected Oracle to harm MySQL as far back as 2005, when they acquired the InnoDB storage engine that plays a crucial role for many MySQL customers. And yet Oracle increased their investment in InnoDB since that time, making MySQL a stronger player in the market.
For further detail on my views on Oracle's intent, please see this interview with me in Forbes Magazine in April 2009.
It may at first blush seem counterintuitive that control of the MySQL assets does not automatically bestow control of the MySQL installed base. But the free installed base of MySQL--enormous on a planetary scale--is voluntarily but not mandatorily coupled to the commercial market of MySQL. It produces huge benefits to the MySQL business, but it is not controlled by it.
The impetus to write this letter comes from my concern with the talented teams of the MySQL business unit and of Sun Microsystems in general. I am also troubled by certain factual distortions about a subject matter that I am intimately familiar with: MySQL and its business model. Open-source business models are complicated and quite different, and it took many years to fully understand and shape the one of MySQL.
A Finnish citizen, I served as chief executive officer of MySQL from early 2001 to February 2008, when Sun acquired MySQL. After that, I served as senior vice president of the database group at Sun until the end of March 2009. Being the only person to have served as the CEO of MySQL and to have attended every board meeting ever held, I believe I have unique insights into these matters.
To be clear, Iin March 2009, and I presently have no commercial or financial interests in the MySQL ecosystem, Sun, or Oracle (or any other vendor in the DBMS market, for that matter), other than my loyalty to Sun employees in general and the MySQL team in particular.
MySQL's Markets and Installed Base
MySQL is the world's most popular open-source relational database, and potentially the most popular relational database of all. It has an enormous influence and impact on the usage and the buying patterns of relational databases (also known as RDBMSs), in particular for Web applications. One might even state that the Internet would not be what it is today, were it not for MySQL. Staffed by a highly talented team of passionate employees, the Swedish company MySQL grew the MySQL business from a small one in 2001 to a massive one in 2008.
"MySQL" refers to two things. On the one hand, there is the huge (community) phenomenon MySQL...On the other hand, there is the business of MySQL...Those two meanings of the term "MySQL" stand in a close mutually beneficial interaction with each other. But most importantly, this interaction is voluntary and cannot be directly controlled by the vendor.
In this discussion, the term "MySQL" refers to two things. On the one hand, there is the huge phenomenon MySQL--an estimated 12 million active installations under a free and open-source software license, millions, if not tens of millions, of skilled users and developers, and tens of thousands of corporations who use MySQL one way or the other.
On the other hand, there is the business of MySQL, which is growing rapidly, thus rewarding the owners of the assets (currently Sun Microsystems).
Those two meanings of the term "MySQL" stand in a close mutually beneficial interaction with each other. But most importantly, this interaction is voluntary and cannot be directly controlled by the vendor.
What I mean is that the vast and free installed base of MySQL is using it of their own free choice, unencumbered by the vendor and under no obligation or restraint. That is the nature of open source. And conversely, the MySQL business is supporting the free installed base of MySQL (by improving the product) voluntarily and in the hope of deriving benefit from the installed base.
This is the paradox of an open-source business, and it took me a long time to truly understand how powerful a force it is. It is unlike any traditional business. The key point is that both the users and the vendors of open source are engaged in a powerful free-market dynamic that cannot be contained by any single entity.
It is in everybody's interest that the two sides of MySQL produce benefit for and derive benefit from each other. But neither group can mandate or control the other one. This is a core philosophy of open-source software and more generally of the "architecture of participation" (as defined by Tim O'Reilly). There is a mutually beneficial voluntary relationship, but there is no control by one group over the other. In more colloquial terms: the owners of MySQL cannot force MySQL users to pay up, and the nonpaying users cannot force the business to subsidize them.
Anyone acquiring the MySQL assets will therefore acquire an ability to control the business aspect, i.e., meaning how MySQL is licensed commercially, but only an opportunity (and no free reign) to derive benefit from the free user base.
This explains how the MySQL business can be valued highly in the market ($1 billion, when acquired by Sun in February 2008) while at the same time providing no way of controlling its installed base. This unusual relationship between market share and installed base is at the core of the topic. The market share is small but controllable, to some degree. The installed base is enormous but not controllable. The installed base is, and can be, hugely beneficial to the owner of MySQL, but only to the extent and for as long as this owner of MySQL enjoys the trust of the installed base.
To put it in numbers, it may be useful to see the usage of MySQL, as divided into three categories:
- Group A: the free installed base, estimated at 12 million.
- Group B: the paying subscription customers of MySQL Enterprise.
- Group C: the customers who use MySQL under a commercial license.
Group A is what MySQL is most famous for and what makes MySQL important to the world. This group derives benefit from the business growth of MySQL (because the product improves), and it produces benefit to the owner of the MySQL assets (in terms of brand recognition, vast groups of software experts trained in MySQL, and so on). But group A is not controlled by the owner of the MySQL assets.
Most of the users operate without commercial support from any vendor (many of them don't even realize that MySQL also is a commercial business owned and operated by Sun). If they need support, there are many alternative vendors. If they so choose, they can run on a different variation (fork) of MySQL. If they feel a need to migrate away from MySQL, there are other databases available (both open and closed source). This group cannot be locked into a specific vendor.
Group B represents the majority of MySQL's business and the majority of the future business potential. These customers use MySQL under the GPL license, but they additionally use the products and services available under the MySQL Enterprise subscription offering. The customers are, to some degree, dependent on MySQL the vendor, but they have options that Oracle's acquisition of Sun will not affect.
Some customers are sophisticated enough to be able to continue running on MySQL without a subscription from anyone, including whoever owns MySQL, and those who lack such sophistication can turn to a third-party support vendor. The control that the MySQL owner has over this group is not consequential.
Group C represents the early business of MySQL. It is a different business model from group B. The model itself (called dual licensing) is highly effective and useful, but the target market for it is comparatively small and limited, especially as the world moves towards Web-based and cloud-based software solutions. The customers in group C are dependent on the owner of MySQL to obtain MySQL under a commercial license. If they are unhappy with the vendor, they will need to switch to another database product, or alternatively they can open-source their own software and comply with the terms of the GPL. MySQL's competitors in group C are different from the ones in group B, and they include both closed- and open-source vendors of embeddable databases.
In summary, the owner of MySQL has a high degree of control over MySQL usage in group C, moderate control over MySQL usage in group B, and no control over group A. In terms of the overall DBMS market and market-influencing phenomena, groups B and C are very small, and group A is very large. The owner of MySQL has control over the product, with respect to two groups that are small in size, and no control over the product, with respect to the group that massively influences the market.
MySQL's Strategy: Pursue a New Opportunity
In the software business, usage patterns of customers are very strong. Someone who grew up on a specific DBMS is likely to stick to it for the rest of her career. For a new entrant such as MySQL, this meant that it made (the) most sense to target a new market segment.
MySQL focused on Web developers when Oracle (and IBM and Microsoft and Sybase) focused on enterprise developers; on so-called scale-out scenarios where the competition was designed for scale-up. This created a greenfield opportunity for MySQL to grow and thrive without having to challenge the usage patterns of the legacy vendors. In the MySQL management team, we decided not to focus on migrations from Oracle and other proprietary databases. We did indeed develop migration tools, but we also knew that such a strategy would not be effective as a main focus.
MySQL was able to derive significant marketing benefit from appearing to challenge Oracle, but we penetrated the markets faster by, in reality, focusing on new, huge opportunities such as Web databases. Key to MySQL's success was its determined focus on markets that Oracle (and others) was ignoring. PostgreSQL, on the other hand, just to mention an example, essentially focused on being as good as Oracle in the areas where Oracle was strong. And hence PostgreSQL, despite being a brilliant product, did not achieve the level of success among users that MySQL did in the same time frame.
Today, anyone studying the markets and the marketing of MySQL will be able to find evidence of both competition with Oracle and avoidance of competition with Oracle. A trained eye will notice the difference between tactical market focus and true, strategic, market focus.
Although I don't believe it would be a viable, likely, or contemplated strategy for Oracle to try to limit MySQL's success in the future, let us for a moment anyhow assume that this, for some reason, could be the case. What would happen in such a scenario?
Group A - the estimated 12 million users of the free and open MySQL who do not rely on the owner of MySQL for support, services, or subscription.
These users don't care who owns MySQL and what the commercial offerings cost because they don't buy them. They only care about the ongoing development and bug fixing of the product. If Oracle stopped developing MySQL, the defined and prescribed response in any open-source setting is forking. If a product is not evolving at the speed reasonably expected by a main portion of the user base, a fork will emerge.
Indeed, in the MySQL ecosystem, there are already a number of forks. Each one of those forks may perhaps currently be individually weak and unpromising. But the reality remains that if the main steward of an open-source product fails to live up to reasonable expectations, the forces of open source will take over.
Group B - the few thousand customers of the MySQL Enterprise subscription offering.
These customers would be alarmed by a slow-down in development of MySQL and/or in the increase of price of the subscription offering. But they would not be alone. If the product did not evolve, these users could turn to the forks that would emerge. And as for commercial subscription services, they could turn to the various firms that provide MySQL services. If there were a sufficient number of such customers, it might turn out to be a market opportunity for a larger services-oriented company.
I know of a concrete case--a major MySQL customer in the SaaS space. They depend nearly completely on MySQL as their database. Not surprisingly, they are (or actually were) customers of the MySQL Enterprise subscription offering. When they learned of Oracle's pending acquisition of Sun, they abandoned their subscription and turned to a small vendor for MySQL support, maintenance, and bug fixing. This behavior demonstrates that the subscription customers have choice.
Group C - the few hundred active customers of MySQL under a commercial license.
MySQL's commercially licensed (dual-license) business has the same control aspects that a closed-source business has. The owner of MySQL has control over these customers, in terms of pricing and further developments, but this business is not large, and there are many options for embeddable databases. Customers who felt that they could not continue as before could switch to alternative embeddable databases (open or closed source) from other vendors.
In closing, I would ask that you, Commissioner Kroes, consider the market dynamics and the facts. Every new day of uncertainty is potentially very harmful to the various businesses of Sun, reducing competition in the market. A delay in the closing of this transaction is therefore only going to work against the respectable goal that you set out to achieve when launching the probe into this acquisition.
I believe that Oracle's acquisition of Sun (and MySQL) will increase competition in the database market. And I also believe that if, on the other hand, it becomes difficult or impossible for large companies to acquire open-source assets, then venture investments in open-source companies will slow down, harming the evolution of and innovation in open source, which would result in decreased competition.
I am at your disposal, if you would like to follow up with me on any of these issues.