Dear MySQL users, MaxDB users and friends,
we have two news for you. The bad one: no MaxDB series posting this week. The good one: german language MaxDB web seminar tomorrow!
My last week has been dominated by preparing the upcoming MaxDB web seminar on MaxDB performance tuning. The web seminar will be held in german tomorrow Thursday, 21.04.2006 at 10 CEST (MEZ). You can still register for the seminar at http://www.mysql.de/news-and-events/web-seminars/maxdb-performance.php. Don’t expect too much from the 45-minute talk. In 45-minutes one can’t do much more than talk about the very basics and try to give you an overview on the topic. Due to the web seminar and other duties we did not make it to write the next MaxDB series posting on transactions.
We have more good news for you. My co-worker C.J. Collier is progressing well in his work to integrate MySQL as a source in the MaxDB Synchronization Manager. Scripts have been developed to create the so-called Shadow and Version tables. These tables contain the meta-data required by the Synch Service to push changes to other participants in the synchronization scenario. C.J. is currently developing the MySQL 5.x triggers which will be used to fill these tables.
No new MaxDB version is on the short-term horizon. But the Changelog of the next MaxDB 7.6 version might show some long-awaited changes, originally planned for MaxDB 7.7. MaxDB will implement data clustering features. Data clustering means that you can decide if all individual database pages of a table get striped over all configured volumes or if chunks of pages are striped. Striping of all pages is good if you have many concurrent high-qualified record accesses, because you are automatically striping the I/O load over all volumes and you are avoiding hot-spots. Grouping database pages in ordered chunks is an advantage if you have few concurrent sessions accessing the same data but you want to profit from sequential read performance and a better support of hardware-based prefetching. Having the freedom to choice on a per-table basis what kind of storage you prefer is a fantastic feature. Applications with few concurrent users and frequent scan operations should see notable performance improvements.
This is still not all… It looks as if some 7.7 optimizer improvements found their way into 7.6. Stay tuned, and join the MaxDB talks at the MySQL User Conference in the next week.
Ulf Wendel for the MySQL MaxDB team