January 16, 2008

Open source DBMS as a business model

Sun’s planned acquisition of MySQL is inspiring a lot of discussion about open source business models. Typical is Michael Arrington’s cheerleading for the idea that you can make a lot of money with open source. More interesting is Gordon Haff’s suggestion that it’s a lot easier to make money with open source when you have other things to actually sell to the same customers (e.g., the rest of Sun’s product line). (A similar view can be found here.)

To analyze this more carefully, it helps to distinguish among three different aspects of open source models:

Here’s what I think about each in the case of database management systems.

Open source product packaging
says “Here. Take our code. Do whatever you like to it.” For specialty DBMS and appliance vendors, that’s terrifically valuable. Netezza, ParAccel, EnterpriseDB and a number of others started with open source code, especially but not only PostgreSQL. But for enterprise users, that kind of flexibility has less value. DBMS are supposed to be fast and stable, and few enterprises have the capability to much around in their internals safely.

Here’s one example: The Netezza Development Network seems to consist mainly of ISVs and classified-agency government users. Or to be even more on-topic, I can’t recall ever having somebody pitch me about an extension to any open source database product, a few entire MySQL storage engines notably excepted.

As another plausibility example, think of all the stories you hear of huge website owners like Yahoo customizing open source products like operating systems. Then try to think of cases of them actually customizing DBMS. Can you? Offhand, I can’t. (EDIT: Oops. See the comment thread below.) Structured extensibility such as user-defined functions can be very valuable, but total-freedom code customization is rarely important to enterprise database customers.

Open source product development says “Please work for free developing our product.” That has worked remarkably well for Linux and a variety of other projects — including WordPress, which powers this blog. Indeed, it’s behind PostgreSQL itself, which is a pretty decent DBMS. More problematic, however, is when large, for-profit companies want the community to do free development for them. Maybe when something grows by agglutination, like Java itself, that can work. But for carefully architected structures like database management or operating systems, it seems more problematic. The rewards – psychic or financial – just aren’t there.

Open source pricing says “It’s free! Sort of! Now please pay us for something related.” Had MySQL remained independent it might have become the first $100 million company with that kind of model – with the possible exception of Sun. I say “possible” because it’s unclear how Sun has directly benefited from giving away Java. Yes, the indirect PR and image marketing benefits seem huge, but somehow I doubt Sun plunked down $1 billion just to add further to those.

Frankly, I don’t think the extreme form of that kind of business model is apt to work. And in fact, MySQL has long sold licenses to enterprises, albeit at much lower prices than traditional DBMS vendors would charge. I think MySQL will rise or fall at Sun primarily on its ability to directly generate revenue from license fees, maintenance, and to some extent professional services. Fortunately for Sun/MySQL, I think there’s indeed a huge market for mid-range DBMS.

Please sign up for our feed!

Technorati Tags: , , , ,


5 Responses to “Open source DBMS as a business model”

  1. Lobotomia on January 17th, 2008 8:02 am

    Open Source is the future in business model

  2. John Sequeira on January 17th, 2008 3:30 pm

    “Then try to think of cases of them actually customizing DBMS. Can you?”

    Google and Skype have hacked the dbms source code of mysql and postgres and made nontrivial source code contributions back to the parent project.


    I’ll bet Yahoo did as well, but it’s not always something folks make public.

    Then of course instead of doing it in-house, enterprises can simply fund the development via bug bounty etc. and control the roadmap indirectly in a way that’s not possible with off-the-shelve dbms.

    Not something that’s often useful but I wouldn’t discount it for web vendors who can save a ton of licensing money by strategic spending to extend an OSS platform.

  3. Curt Monash on January 17th, 2008 4:54 pm

    Thanks for the info, John!


  4. Daniel Weinreb on January 18th, 2008 6:40 am

    Does the third item (“Sort of!”) include dual-licensing models, in which the product is free for you to download and use by yourself, but requires licensing if you use it in a product? This is how SleepyCat’s Berkeley DB is sold. And now they’re owned by Oracle, so Sun may not be the first.

    I think Sun’s “Glassfish” open-source J2EE app server uses something like this, but it’s hard to tell from the web site.

  5. Curt Monash on January 18th, 2008 2:49 pm

    Actually, Dan, I was blurring the distinction between “Free” and “Very cheap”. 🙂

    But I wasn’t thinking of cases where developer versions are cheap/free, but you have to pay big bucks to deploy ANY app. That’s just a common delivery model now, including among companies that make no pretense at being open source. Indeed, I’d guess half or more of the non-appliance vendors I write about in this blog have some sort of a model like that.



Leave a Reply

Feed: DBMS (database management system), DW (data warehousing), BI (business intelligence), and analytics technology Subscribe to the Monash Research feed via RSS or email:


Search our blogs and white papers

Monash Research blogs

User consulting

Building a short list? Refining your strategic plan? We can help.

Vendor advisory

We tell vendors what's happening -- and, more important, what they should do about it.

Monash Research highlights

Learn about white papers, webcasts, and blog highlights, by RSS or email.