ramblings on PHP, SQL, the web, politics, ultimate frisbee and what else is on in my life
back

MySQL AB learns a lesson on release management. Or does it?

At first sight it seems MySQL AB has learned the lesson from its 5.0 release and is not rushing as much anymore with their releases as much. It seems that the 5.1 GA release was pushed back from this fall to Q1 2008. MySQL 6.0 will feature the new Falcon storage engine, but without foreign key and full text indexing. That will have to wait for MySQL 6.1. Now 6.1 is supposed to go into beta 2008/2009 which means a GA release is expected in 2009. So when exactly is MySQL 6.0 supposed to be done then? Last I heard was fall 2008, which would mean GA release for 6.0 and the beta release for 6.1 to be pretty close.

I hope this is not overly ambitious. MySQL 5.0 was to a large extend a feature checklist release. As such it took MySQL AB quite a lot of time to get 5.0 to be reasonably stable. 5.1 adds some features and gets some of the new features to decent performance levels, though it will not be until 6.0 (btw 5.2 has been relabled to 6.0) that I will feel comfortable using sub queries in a performance relevant piece of code. So when was 5.0 released? That was fall 2005. It will have taken them over 2 years to get to MySQL 5.1.

I know from talking to people at the MySQL internal developer conference that QA aka unit tests are getting much more attention now. I guess its high time this is done, especially to not make life needlessly hards for storage engine authors to introduce unwanted minor but critical differences into their engines. However this all sounded mostly like "we will" and not like "we have". So I am wondering if they just feel that confident about their emerging test suite catching all these bugs and performance regressions that they did not catch for 5.0.

If 5.0 seemed rushed because of trying to more closely match feature lists of prospective customers, then 6.0 does not feel any less pressure to be delivered quickly. We do not know all the details on the InnoDB deal. I heard a rumor that actually the deal ends 2008. I would doubt that, but since nobody outside of MySQL AB and Oracle knows the details, we are left to guess what "multi-year extension" means. Either way MySQL AB wants to establish their own storage engine, so that less business goes to Oracle when MySQL AB makes a deal. This entire InnoDB deal must be SAP's worst nightmare. They push their own database to open source in the hands of MySQL AB, hoping to take away business from Oracle providing the database backend for SAP installations. That flops more or less. Then they decide to instead wait for MySQL to get there just to find that now Oracle also owns a piece of the MySQL stack, even trying to buy MySQL AB itself. As SAP is not only an important partner, but also a kay investor, they must be scrambling like crazy over at MySQL AB to fix this situation.

I just hope that they remember that if they want to run stuff like SAP they need to make sure that Falcon is known for reliability from day one. The fact that they pushed foreign key and full text to 6.1 makes me somewhat hopeful that they realized that feature checklist releases will come back to bite you and that in open source the solution is release often and not release "crap". But the seemingly short cycle between 6.0 and 6.1 makes me hesitant to believe that, though a new minor release within less than a year is not heard of in the open source world, MySQL AB has not really had the track record to proof they can pull it off. Looking at the previous steps it seemed more like 18 months for the next major release with 4.1 to 5.0 being the exception of just 12 months (and we know what that got us).

Comments



Re: MySQL AB learns a lesson on release management. Or does it?

Is this news that MySQL always promised a lot but delivered a lot less? Look at PostgreSQL's release history, the developers are capable of putting out one major release a year, they kept this pace for a long time. Now look at MySQL 5.1 that is in beta for a year and tell yourself: "I believe that MySQL 6.0 will be stable in 2008... I believe that MySQL 6.0 will be stable in 2008..." Try not to laugh (too much) in the process.

Re: MySQL AB learns a lesson on release management. Or does it?

I've always thought the SAP deal was a huge mistake on SAP's part (and a serious coup on mysql's); if they wanted to get out of the database business, PostgreSQL was in a far better position technically to build atop of, and it's diversity in corporate support at the project level works well to keep Oracle from putting a bulls-eye on it (like they had done to SAP themselves). Of course pg didn't have anyone to scmooze the SAP brass, and at this point I suspect the amount of investimate they have in MySQL is far to great for them to stomach turning away from.

WRT the Falcon engine, I find it interesting that you see thier current moves as good ones. I can see pushing fulltext support to 6.1 as a good idea, but a lack of foreign key support is just astounding to me; even with DBD and INNODB out for many years now, they still get beat up for a lack of FK support, how many years will it take Falcon to overcome that bias as well?

Before you can post a comment please solve the following captcha.
your name