I wanted to write a quick shout-out to congratulate the PostgreSQL development team, the folks at Sun who work with Josh Berkus, and the folks at EnterpriseDB, all of whom contributed to the excellent benchmark results for this quarter's SPECjAppServer2004 benchmark suite. I'm looking forward to seeing Josh at OSCON in a couple weeks and meeting a few more of the PostgreSQL developers than I did last year.

I know that the PostgreSQL developer team has spent a considerable amount of time and effort improving performance bottlenecks and streamlining code for the PostgreSQL 8.2 release, and the benchmarks show the results of that hard work. It's great to see the pressure put on Oracle and the "big guys" from …

Some recent blog posts by Markus and Frank got me thinking about some important things when considering an upgrade to MySQL 5. Did adding much-needed features — such as stored procedures, views, and triggers — to the MySQL server lead to poorer performance than previous server version without these additional features? This is a question that, over the next few months leading up to the MySQL User's Conference, I plan on writing quite a bit about, and detailing in a number of webinars and blog entries about performance tuning, index selection strategies, and coding techniques for optimal performance. The first of these webinars is tentatively scheduled for March 1st; check back …

Roland Bouman sent me an email yesterday alerting me to an idea presented by Giuseppe Maxia on the MySQL Stored Procedures forum a couple days ago. The idea, in a nutshell, is to create a repository or library of tuned and tested MySQL stored procedures, similar to something like CPAN for Perl modules.

Frankly, I think it's a great idea. A project like this, given a good head of steam, would propel the movement of MySQL 5 into the mainstream and encourage hosting companies and others to adopt the newer technology. If there is enough demand for this kind of thing (which I think there would be, considering both the hype about MySQL 5 and the communities on the web alrady set up for other RDBMS vendors who support stored procedures) it will force …

I found it strange that MySQL was notably missing from an InformationWeek DatabasePipeline article today about growth in market share of both Oracle and Microsoft. At first, I though perhaps IDC had some strict criteria on which they decided what was "a relational database system" and what wasn't, but when I saw no mention of MySQL and yet a mention that NCR's database server held down a 3% market share, I was a bit perturbed.

When is MySQL going to get the recognition it deserves in the database industry? Why hasn't it already, with large companies already relying on it for critical applications? Has the fact that MySQL is open source been a limiting factor to its industry acceptance?

Let me know your thoughts

It's been a long time since I worked on Microsoft SQL Server. Well, relatively long, in IT terms. Since I've been doing all my recent work on MySQL, there are a few things that I've sorely missed from my SQL Server days. It will be quite some time in the future before I see some of those things incorporated into the MySQL environment -- SQL Server Profiler, anyone?

But, to my delight, I found out from Mike Kruckenberg the other day that MySQL, starting in version 5.0.3, was including the SQL standard INFORMATION_SCHEMA meta tables into the source distribution.

This was some welcome news. I'd always valued the ability I had in Microsoft SQL Server to query the sys_ tables for statistical information about the database schemas, users, roles, views, etc. It gave you an insider's view into …

I've been working on outputting the code documentation for MySQL 5.0.2 using doxygen, and have some advice to anyone attempting this feat. Don't configure doxygen to generate call graphs!. At least not for the MySQL core distribution. My laptop ran for 18 hours yesterday and last night and this morning, I checked on its progress and a flurry of emails had been sent to the local root@localdomain.localhost about memory errors running dot, which is the graph visualization toolkit from AT&T labs needed to produce the UML diagramming doxygen is capable of producing.

I knew that the call depth for a lot of the major functions in MySQL was long -- hundreds to thousands of sub-calls -- and was expecting a long generation time. But, this was crazy.

So, I stripped the CALL_GRAPH settings from the configuration file I used for doxygen and restarted the process after rm -rf ing the html directory …

First announcement on this site about my upcoming book, entitled Pro MySQL, published through Apress. Really excited so far. I'm co-authoring the book with Michael Kruckenberg, out at Tufts University.

Been a bit of a roller-coaster ride so far! Between clients and all the other projects we've got going in, finding time to write hundreds of pages on MySQL internals and SQL development is tough. But, I'm very happy with the work so far. I've been concentrating mostly on the development end of things, and Mike's been digging in hard for the adminstrative end of MySQL and many of the newest and most exciting features of MySQL 5.x.

It should be published in mid 2005, and covers a broad range of topics, from the internals of MySQL to clustering and replication for high-volume applications.

You can see the "preview" page …