MySQL Performance Under a Microscope: The Tobias and Jay Show

In this talk we examine the little things that may not be obvious, nor easy to test. It’s not a session about full architectures, or how to scale up or scale out, we examine the microlevel and try to find the optimal way to do the small things. It’s a session about how we found best practices through trial-and-error methods, backed up by benchmarks.

Along with each benchmark, you’ll hear our (sometimes differing) opinions on why the benchmark results are the way they are. If you’ve ever seen a debate between a Brit and a Swede, this can be pretty interesting.

A highlight of the benchmarks this year:

• How much of a difference does specific options make when creating indexes?
• How much faster can I make my searches by making use of partitions?
• When can I speed up my application by using Stored Procedures?
• How much faster are prepared statements?
• Does it make a difference if the columns I retrieve are located at the start of the table definition?
• How much faster are my queries if I use SQL_BIG_RESULT for larger queries? and when should I use it?
• How much faster is an OLTP workload if I disable the query cache?
• How many NULL fields can I have in a table schema before performance is affected?
• Is HAVING faster than WHERE if the HAVING clause filters non-aggregate columns?

If you think knowing the answer to those questions would be interesting, don’t forget to pay us a visit at the User Conference.