Making MySQL pay: A question of core and complements

Jeff Gould has written an excellent piece on the big question arising from Sun's acquisition of MySQL: how will Sun make enough money on the deal to justify the $1 billion valuation? Gould's analysis is generally solid, but he misses a few key points.

First off:

Only time will tell. But in my humble opinion, MySQL's open source business model will make Sun's road to payback a lot steeper than if it had bought a software company with conventional revenues and profits.

Ah, the good old days! Just one problem: those days are gone. Pining for an acquisition of the old way of selling and distributing software is like pining for Mayberry: you can want it, but those days are never coming back. VCs aren't investing in proprietary Mayberry anymore. Except from the consolidators of 20th-century software (Oracle, IBM, SAP, Microsoft), customers aren't buying into the false Mayberry that left them destitute of innovation and options.

Open source is the way forward. But that doesn't make it an easy road, as Gould suggests. Here's where his analysis becomes relevant.

MySQL attempts a Red Hat Fedora/RHEL model, but it actually has this backwards as its Enterprise version is on a faster release cycle/development path than its Community product.

This is a moment-in-time problem but not a long-term problem, in my view. MySQL has the benefit (and problem) of having relied on a support and dual-license model for too long. The dual-license model has worked--roughly 60 percent of its revenue has come from those that want a proprietary license to the MySQL code (which gives them the benefits of open source without the requirement of contributing back modifications).

But the fact that MySQL didn't distinguish an Enterprise product from its Community product for so many years means that MySQL will need to continuously discover new ways to differentiate the two. This need not be with proprietary software, but will almost certainly involve "proprietary" services. Like the JBoss Operational Network or Red Hat Network, such services allow MySQL to distribute its core product freely (i.e., open source and free of cost) while providing add-on value for which customers will pay. This calls to mind Gould's next point:

"It appears though that the additional features of the Enterprise version are not enough to compensate for the revenue-destroying effects of the free Community alternative."

This is emphatically wrong. Gould neglects the tremendous value that giving away Community provides to MySQL. From where does he think the company's sales leads come? From Community, of course. MySQL only recently in the last year started to distinguish between the two versions. Over time, its ability to convince customers to pay for Enterprise will improve. Adding Sun to the mix will accelerate this.

But it needs to be said that MySQL's way forward is not, as Gould suggests, to add proprietary bits to its software. MySQL has not done this. Instead it has added proprietary services to the periphery of its product line (like Workbench) which add value to the core software without inhibiting free and open-source use of the core software. It is critical for MySQL's success that it not lock off its core. Doing so would trample on its primary lead generation tool.

Equally importantly, letting people evaluate the full product before buying also ensures that they are well-qualified leads and that the sales cycle is much shorter. I should know--my own company routinely sells six-figure deals in under 90 days, start to finish, completely over the phone/e-mail. You can't do that with a proprietary product.

Some will think my distinction between proprietary services and proprietary software is sophistic and ultimately impotent. It's not. Customers don't want to be locked in on their database choice, their content management system choice, etc. When a vendor locks up this core product, they lock in the customer.

If, however, services around the core are made proprietary (services like support, like networks that administer patch updates which can also be implemented by a more manual process, etc.), this is a way to ensure dollars more easily flow in but doesn't prevent customers from using the core product for free. And it certainly doesn't prevent them from switching to a competing product. They lose the peripheral value, but that's a small price to pay. Importantly for MySQL with its Workbench product, in particular, MySQL is not locking anyone out of its products. The same things that a developer would use Workbench to achieve can be done directly without Workbench. It's a time-saver but not even one that a good developer necessarily wants. Peripheral value, not core.

Reading Gould's informed but ultimately misguided commentary on MySQL makes me very glad that it didn't decide to sell to another buyer. It tells me that MySQL's management is firmly committed to open source, not merely cash. Like Google, MySQL demonstrates that it's possible to be financially responsible and socially responsible. The way forward is with open source, not proprietary lock-in. MySQL and Sun recognize this.