So I have been doing quite a bit of benchmarking and while most of my tests have been amazingly simple, I thought I would share some of my findings. The following is the total time it took to run a series of simple inserts over various threads and engines:

You’ve heard me on the MySQL Podcast at http://www.technocation.org, now come work with me, the “She”-BA!

The company I work for is an online social networking/dating site. Our main product is for men seeking men in 87 countries throughout the world. We’re looking for another MySQL DBA, as designing schemas maintaining data integrity for our 1 million users (and growing fast!). The salary is dependent upon experience of course, but the company I work for pays on the high side of the industry standard for the Boston area.

Application Instructions
Please send cover letter, résumé and sample schema to work@online-buddies.com, with “MySQL DBA” as your subject. The
sample schema should reflect your abilities, so if you send along a schema you would like to see improved, include a description of what you would love to do to that schema to make it better.

…

This may be obvious to some but it wasn't to me so I thought I'd write about it.  Recently I discovered a problem with our implementation of DbConnectionStringBuilder.  Whenever you changed a field that took a string (i.e. character set or server), the connection string property would not update.  I know the changes were applied because I developed a test application that queried the connection string.  It was correct even though the UI of the property grid was not. 

After an embarrassingly long time working on this I emailed a friend of mine who I felt would know the answer.  Turns out he didn't but he knew someone who did and not more than 30 minutes after talking with him I had my answer.  Each property of MySqlConnectionStringBuilder needs to be decorated with the RefreshPropertiesAttribute.  This causes the property grid to refresh itself when they change.

Welcome to the 44th edition of Log Buffer, the weekly compendium of database blogs. This one’s a grab-bag. Let’s start with some Oracle stuff this week. Christian Shay notes that Oracle 10g for Windows Vista, is available for download. That is sure to please Oracle doyen and fervent Vista booster Doug Burns, who also has [...]

Often people ask me how to retrieve the value of an autoincrement column when they are using the MySqlCommandBuilder object.  Generally this is done by appending something like ";SELECT last_insert_id()" onto the generated insert command text.  Command builder objects usually provide some type of syntax where a user can indicate that they want this to be done.  Connector/Net 1.0 had some syntax for this but starting with Connector/Net 5.0 that all changed. 

With the new ADO.Net 2.0 API, Microsoft did not directly provide for any type of API that users could use for selecting this.  With ADO.Net 2.0, there are two things that need to happen.  First, a SELECT statement similar to above must be attached to the INSERT command.  Second, the command object must have its UpdatedRowSource property set to Both or FirstReturnedRecord.  What this does is tell the data adapter update engine to take the …

"

Red Hat formally announced the Red Hat Exchange (RHX) at this week's Red Hat Summit in San Diego.  This was a great conference for Linux sysadmins, developers and those who are using or contributing to open source development. It was also nice to see how well JBoss is doing at Red Hat.  Although Marc Fleury left the company some time back, they've retained a huge strength in the JBoss development organization and customers are now getting the benefit of an integrated stack.  There were a lot of folks rooting for a successful integration of JBoss into Red Hat and I think they've made it happen.

There were many technical sessions as well as keynotes highlighting several new initiatives, the most important of which I believe is RHX.  While RHX is not a technical …

Sometimes you may need to create a set of rspec specifications with pretty similar structure and small differences. I’ve got such situation in my project and decided to try to use Ruby’s dynamic code generation features to make my spec file shorter.

I have some multiplexing helper in my templates which allows me to use the same template for different similar pages. This helper returns URL from the set of params and a type. It could accept 5 different url types and raises an Exception when requested URL type is invalid. Without this dynamic code generation feature I would need to create 5 different specifications (one for each URL type) to be able to see each URL type test as a separate line in test results log. But with this simple technique my code looks like following now:

describe VideoHelper, 'when profile_video_url method called' do

  before do

    @user = mock('user')
…

I released version 0.9.8 of my MySQL UDF code generator yesterday and i finally moved development over to my own php-baustelle.de family of sites.

Continue reading "New MySQL UDF generator release, project moved"

Today I upgraded the blog software at sheeri.com (and sheeri.net and sheeri.org). Please let me know if you find something that doesn’t work as expected — awfief@gmail.com.

At the MySQL Users Conference, my good friend Mark Atwood (creator of the free Amazon S3 Storage Engine) mentioned that any site with a login should have OpenID as an option.

Mark, I upgraded for you! I was using Wordpress 1.5.2, now I’m at the “latest” version. Anyway, this is just to let folks know that if you so choose, you may now use OpenId if you wish to login and make comments.

Of course, I do not require login (and have a great spam filter) so that’s just another choice you have.

This is a hack I've heard about a couple times now:

Paul wrote a script that reads from the logfile the queries that are going to be executed moments later. He parses the queries and constructs new select queries that populate the cache with the data that speeds up the upcoming writes. He claims, if I remember correctly, a three to four times speed-increase.

Here's the problem in a nutshell. The master can write transactions in parallel but slaves can only write them in series. [1]

This means you have a lot of optimizations on the master (TCQ and NCQ being examples) that aren't possible on the slave.

What this patch would do is precache the data so it's already available in memory. Since you're pre-reading the binary log you can run SELECTs in parallel on the SLAVEs so that the cache is hot when …