Regarding what is MySQL AB, what is Sun and what is Oracle responsibility, I think people give these things too much importance. MySQL grew to be the worlds most popular database with an open source business model. They then added some closed stuff after 2006, and continued to grow. (Of course, most people weren’t aware of this. I wasn’t when I joined the company!) Sun didn’t undo that, but did halt any new closed source features, yet business continued to grow and we made the largest deals ever. As far as I know Oracle sales have continued to grow, and I’m sure they will after this day too.

Regarding the releases I think people generally give credit in wrong places: MySQL 5.5 was released by Oracle, but the work was pretty much done during Sun watch. 5.1 was released by Sun but majority of work was done at MySQL AB time. The main reason quality got better was progress in development methods, in particular a continuous integration system – the heroes of that story are people like Kristian Nielsen, Philip Stoev and countless people in many teams who just fixed hundreds of bugs found. The main reason we now see faster feature progress than before is those same reasons enabling it, and the new development process, and Tomas Ulin personally.

When Oracle makes things open-source, or includes features that obsolete those we’ve built for our customers, it benefits us because we get to look at and support the code, and we don’t have to maintain our own fork of that feature anymore, and it benefits users because they get great functionality. It benefits Oracle because they’re the ones maintaining a great database server.

When Oracle makes a closed-source extension with a well-defined API, it benefits them because it’s a powerful upsell for their customers, and a reason for non-paying users to pay. This benefits users because it means money is invested in the whole server, not just the paid-for features, and only small and well-defined portions of their database are closed-source, which is a reasonable thing to trust Oracle to do well (i.e. with good quality, and good intentions). It benefits Percona because those open APIs create opportunities to not just reimplement the same thing, but even do something better or different that serves other uses. This benefits users by providing them with more choice. If the feature weren’t a pluggable extension, but were instead built right into the server, it would be the One Way To Do It, and would be much harder to reimplement or tweak. Plugin APIs are great (one of the things I love about Drizzle). And indeed we have a very, very busy development team that is constantly building special features for customers to suit their exact needs.

I just think this is win-win-win all around and I’m very happy about it. I don’t mind anyone comparing Oracle to the “good old days” at Sun Microsystems, though. I believe in “by their fruits ye shall know them,” and although I saw a lot of good engineering during the Sun days, it wasn’t coming to fruition. Whether Oracle is riding on Sun’s coattails here or not (I suspect not), the thing that matters is that MySQL 5.5 and 5.6 are more advanced and better quality than the releases we’ve seen for a long time before that, and the extensibility of the server is hugely increased and accelerating.

It’s kind of funny you have this opinion – and I should stress I have no doubt it isn’t 100% genuine – given that Percona has been and surely will be a big beneficiary whenever Oracle or someone else introduces proprietary features. Xtrabackup, cacti templates, nagios checks (even pt-query-digest) are all open source software that replace various proprietary offerings, and their development was sponsored by Percona customers. So chances are these announcements will bring you more work from your customers. So you have double reason to be happy for this – both personal beliefs and professional reasons.

Yes, I agree that the licensing is a big hurdle for MySQL’s adoption. It doesn’t help that sometimes sales people combine people’s general fear and ignorance of the GPL with wrong advice or high pressure.

I wish that the GPL had been written as a license, instead of a manifesto. I think that a lot of problems come from it being more rant than explanation. My view on this has changed significantly over time, and if I were to start a new open source project I’d probably use BSD/Apache/MIT or similar instead of the GPL. I’m just sick of discussing what you can and can’t do with a GPLed piece of software.