Does MySQL really have an open-source business model?
I’ve been thinking about the business of what’s variously come to be called commercial open-source and enterprise open-source. I’m interested in the gestalt — the product, development processes, marketing, licensing and so on.
MySQL has tried many different ways to earn money. These include dual licensing, support subscriptions, a knowledgebase, consulting, an Enterprise/Community split, trying to make parts of the server closed-source, making tools in a split community/premium version, building functionality externally in closed-source software where they can charge for it, and most recently creating three tiers of functionality for the MySQL 5.1 server. And these are only the highlights — MySQL has courageously tried a lot of different models, far more than the blogosphere regularly acknowledges.
But are any of them really open-source business models? I have posited before that the MySQL server is not really open-source. I’ve come to believe that the MySQL business model is not an open-source one, either. I think all of these different approaches are steps towards a more closed-source business model.
Why is this happening? I believe that the sales department is the strongest influence in this direction. I’ve formed this opinion over the last couple of years, in part through many conversations with people on both the inside and outside of MySQL. And more recently, as a Percona employee I have been able to talk with many people who have negotiated with the MySQL sales team and then opted not to buy from them. They sometimes volunteer information on what it’s like and why they’ve come to talk to Percona instead.
I’ve also heard from insiders that the familiar sales tactics don’t work for MySQL. I believe this is because they rely on ways to gain leverage in the sales negotiations (e.g. lock-in, bundling, upselling, withholding). Many of these levers find no fulcrum in truly open-source software. The story is that the sales team came back to the rest of the company and said you have to give us something we can sell. This isn’t working.
As I see it, the changes in MySQL’s business models over the last few years have produced a self-reinforcing cycle, something like this:
- Experienced customers who understand the merits of open-source software recognize that MySQL has trended towards closed-source and isn’t really offering a compelling open-source value. They don’t buy, and…
- … MySQL trends further towards closed-source and tries to sell that; as a result, customers new to open-source software are not educated about its true value.
So instead of making MySQL more open-source and evangelizing that, MySQL might have left themselves few options other than to fall back to closed-source models. And that virtually requires closed-source products to sell. You can call it “feature differentiation,” or “value add,” or even “crippleware” (and Monty Widenius himself has done that) — the fact is it’s not open-source MySQL is trying to sell. Their offerings that are closest to being open-source are not for sale.
At the sales level, it appears to be working. I’ve heard that many more people are buying than before. I don’t doubt that. This fact is touted as proof that MySQL is successfully building an open-source software business, and the pundits love to applaud it, even urging MySQL to take a harder line. I suppose they don’t see the irony of saying that the open-source business model is perfected by becoming closed-source.
I don’t identify with this stance. I would prefer to see the open-source philosophy spread from the company to the customers and community, rather than allowing the customers and sales people to drive closed-source values back through the gate.
Unfortunately, I don’t think MySQL will not do an about-face, because the market for real open-source simply isn’t as big as MySQL wants. If your goal is to make money first and foremost,[1] instead of creating value for customers, you can do that better with closed-source software. And I think that as a result, MySQL might be working against itself, because open-source and Free Software are fundamentally about creating value for the users. If the users and the customers are the same people, there’s a conflict of interest: the software wants to create value, but the business wants to create revenue.
The good news is that MySQL is Free Software, so it does not simply live inside of MySQL or Sun. Free Software is a meritocracy; there’s equal opportunity for all of us, and in that sense Sun/MySQL are offered the same chance as the rest of us to serve this community. No one is in a position of ownership or control (remember, Sun purchased MySQL the company, not MySQL the database). We are all called to be stewards. I believe that if Sun/MySQL accept this role, MySQL the database and MySQL the company will prosper together; if not, MySQL the database will find a new home. The business of MySQL is associated with MySQL the database through freedom and choice, not bound through force.
I believe that a closed-source business model around MySQL will ultimately lose momentum, and those who work to set MySQL free will earn the community’s trust and support. I realize that I’m contradicting all the experienced people who really know a lot about business, and I’m happy to let the passing of time correct me if I’m wrong.
These are interesting times. None of this is without precedent, yet there is so much we don’t know. I think it’s fine to keep trying different approaches and see what works. I’ve chosen my position based on my principles and beliefs, and I’ve tried to align myself with what I see as the Tao of Free Software, if you will. I’ll be continuing to observe and participate, and re-evaluate my position if necessary.
[1]For the record, I have no problem with a company whose goal is first and foremost to make money.
Good writing. One wonders, if Redhat could do, why not MySQL?
While MySQL is a very compelling alternative to “real” closed source DB’s, I think its evident that the lack of good ideas to monetize ( like Redhat ) is what’s dragging MySQL. This creates FUD and FUD is never good. Big fishes like Larry’s and Microshaft’s use that in their scare tactics.
Shankar R
24 Dec 08 at 2:17 am
It’s true. Because of the dual licensing strategy, MySQL is pursuing what amounts to a closed source business strategy with an open source product. This means they have the overhead of a proprietary software development shop, but with relatively lower margins because of their reduced vendor lock-in. And they lose on the collaborative benefits of true open source.
That said, MySQL support is nothing short of fantastic and the (non-open-source) Query Analyzer product is compelling for larger development organizations. I suspect that Sun will develop more closed-source plugins to continue their current course, while also funding Drizzle which works on a “truer” open source level. As you say, these are interesting times.
Ryan Thiessen
24 Dec 08 at 3:34 am
Nice piece. I think we should note that the traditionally proposed support-based model for open source just doesn’t work all that well, at least in terms of long-term growth. This piece talks about it a bit.
I don’t know that you were criticizing MySQL exactly, but I know for me as a free-software supporter I’m tempted to look askance at companies doing stuff like that (EnterpriseDB with Postgres is another example). But we’ve got to make a living somehow, and I didn’t think giving things away was a viable business model for Netscape in 1995, and I still don’t. =)
Chris D
24 Dec 08 at 12:30 pm
We had discussed this closed source nature of MySQL at work a while back. With it seeming inevitable that Drizzle will eventually supplant MySQL (the software), there seems few growth options for MySQL (the company). We’ve never found MySQL Enterprise compelling as a product (although Query Analyzer seems vaguely tantalizing) and our few encounters with MySQL support have turned out to be pretty disappointing (far from “nothing short of fantastic” at any rate).
John T
24 Dec 08 at 12:44 pm
Chris D, does it not work all that well? Or is it just, as the article says, “failing to meet the expectations of investors”? What if there are no investors? Is it the nature of companies who are funded by investors that isn’t compatible with support and services?
MySQL AB (the company) was acquired for one billion dollars. Is there a successful open-source business model that doesn’t have to meet billion-dollar expectations? I believe there is.
Xaprb
24 Dec 08 at 12:52 pm
True enough. Maybe there’s a distinction to be made between a “viable business”, which can happen at any scale and maybe it only provides a living to a small numebr of people, and a “growable business” that can be a Cygnus or Red Hat. My impression has always been that large FOSS businesses are outliers, and that “no one ever got fired for buying
IBMMicrosoft” is a defect in the DNA of management, rather than a simple product of FUD. But that could be uninformed bad intuition: I’m sure you know more about it than I do.Chris D
24 Dec 08 at 1:14 pm
Hi Baron
Excellent post and I would agree with you on everything I think. To give a bit more insight…
The state of affairs as you describe it is correct. However, some of it is purely unintentional. For instance, I know MySQL management would love to have the “free workforce” benefit of Open Source, who wouldn’t! But maybe at some point in time this was less of a priority, and maybe all managers failed to prioritize it due to not understanding it fully. (Understanding it sure, but not *fully*.)
For instance, historically MySQL had an active external community, developing complete (and important) modules like the JDBC driver. At MySQL AB, the strategy seems to have been to hire as many community developers as possible, to have them work full time on MySQL. (I have no idea what the strategy actually was, I’m a rather recent employee.) Unfortunately this strategy and MySQL’s business in general was so successful, that the company hired basically most of its own community and then the community was mainly a bunch of bloggers left, not contributing any code. Right now the tide is changing, and some of the brightest minds in the MySQL community external to Sun are former employees – like yourself. Unfortunately however, in the meantime we’ve lost the capability of actually working with the community. Now there are people doing great work out there, but we are not really good at working with them. A lot of information (simply like mailing lists, say) isn’t public, so a non-employee is an outsider. Even if you come up with a great patch, there isn’t somebody at Sun assigned that *must* review it and help you with it. (Someone might do it, but it is always not his first priority.)
The good news is that this is likely going to change for the better. I don’t know details yet, but early 2009 you might actually hear about some initiatives. (And mind you, some steps have already been taken, as noted earlier in Kaj’s and Giuseppes blogs: We stopped using bitkeeper and use bzr/launchpad instead, also the Sun Contributor Agreement is better liked than the old MySQL one, etc… All these steps serve the ultimate goal of getting to a more open development process and it is very intentional, only slow.)
Personally I believe that once our development resembles more a healthy Open Source ecosystem than today, the business may start to gravitate towards something like that too. (Whereas currently, our open-in-license-only development resembles closed source practices and sales thinking easily slips down that route too.) It is kind of like the European Union approach to making peace: Once you have enough mutual trade arrangements, you are so dependent on each other that you have to work together and can never go to war again.
Henrik Ingo
29 Dec 08 at 5:27 am
Henrik, thanks for your thoughtful comments. One clarification: I was never a MySQL employee myself.
Xaprb
29 Dec 08 at 9:45 am
@Baron: Ah yes, I confuse you with another blogger :-)
Henrik Ingo
29 Dec 08 at 10:50 am
[...] said before that I think one of the reasons MySQL was unable to create an open-source business model is that their sales folks pushed the company in the direction of closed source. I’m honestly [...]
What does an open source sales model look like? at Xaprb
29 Apr 09 at 11:28 pm