Drizzle

We are very happy to announce Drizzle Developer Day is coming again this year. Similar to last year it will be the Friday after the MySQL users conference, April 16 - 9:30 a.m. to 5:00 p.m. We are still working on the exact location and will let you know as soon as possible, most likely it will be at or very close to the Santa Clara convention center. We invite anyone interested in providing feedback, implementing new features, helping to fix bugs, or just wanting to learn more about Drizzle to attend.

Please sign up at http://drizzle.org/wiki/Drizzle_Developer_Day_2010_signup. Space will be limited.

Also, to make this day a success, please provide your input on discussion topics at http://drizzle.org/wiki/Drizzle_Developer_Day_2010

Hope to see you there!
-Lee

We recently added support for running the dbt2 benchmark as part of our drizzle automation suite. dbt2 is an OLTP transactional performance test. It simulates a wholesale parts supplier where several workers access a database, update customer information and check on parts inventories. We currently are using the defaults (10 warehouses, 5 minute test runs and running various number of connections for each run up to 1024. The initial runs exposed a race condition in our TemporalFormat::match() code at 1024 connections.

We now have a very nice suite of tools to help us with tracking performance and scalability of Drizzle, besides dbt2 we also are running sysbench, sqlbench, crash-me and randgen for all of our builds in the staging branch of Drizzle. Of course we are always looking for more, so if you have any suggestions on other benchmarks or tools to add, please let us know.

If you want to receive all of the various benchmark results you can subscribe to the mailing list.

Note also we are in the process of getting our changes to dbt2 merged to the dbt2 trunk, it should be there very soon.

Happy New Year to all!

-Lee

Drizzle source tarball based on build 1223 has been released. This release contains some changes from Eric for the start of changes to the Drizzle protocol. You must now use --drizzle-protocol-port=N when starting drizzled instead of --port=N. See his emails for more details

The download file and change log can be found here.

Lots of great updates from Jay on his recent Replication work, make sure and take a look and provide feedback.

Previous posts on Replication work including Padraig's plugin work:

Just a reminder that OpenSQL Camp is coming up quickly. For details see Eric's Blog and also this latest entry.

We now have the Random Query Generator running as part of our automated testing framework! Thanks to Stewart for kick starting the port for Drizzle and Philip Stoev and Bernt Johnsen of the MySQL team for merging in our changes. I'm still trying to figure out all the bells and whistles and fine tuning but we have several canned configurations running and even found a bug in Drizzle as a result of some of the tests. You can see the results on the Drizzle dashboard. I welcome any feedback or suggestions for improving the setup.

-Lee

See Brian's summary here

Thanks to Patrick Galbraith we now have sqlbench running and thanks to Jay's guidance also have it integrated into the Drizzle Automation framework. We have added sqlbench as a job in our Hudson framework so that it will now automatically run every time there are updates to the staging branch. If you want to be on the email list for receiving any of the benchmark results, please register here.

Next up, crash-me which is currently running but needs some polish before we declare success.

-Lee

Just a heads up that Brian will giving a Drizzle presentation at OpenSource World in San Francisco tomorrow. While I don't see a detailed speaker schedule on the website, I know that Brian will be presenting at 9:00 a.m. tomorrow.

See Eric's blog for the details.