So we have mid-term evaluations this week and I’ll like to give a summary of whatever work has been done till now and what more is coming.

  • There are two types of constants currently: Engine Specific, which will be different for different engines and Global, which are same across all engines. Currently in the code, I have two Engine specific constants: read_time and scan_time. We also have two Global constants time_for_compare and time_for_compare_rowid.

  • Currently there is one global Cost_factors object, in which the data is populated on server start from the optimizer_cost_factors table present in the mysql database. The case when an engine is loaded after server start has been accounted for.

  • There is one Cost_factors object present in each THD. The stats from queries are collected in a measurement object. This approach is discussed in more detail in MDEV-350. Two approaches have been discussed there, I have started with the first approach. In the first approach we will be solving a system of linear equations to get individual cost factors.

  • We will write the per THD stats to global stats, when the thread disconnects. This has not yet been implemented yet.

  • The code is available at my github repo. The implementation of read_time and scan_time factors have been tested.

Reviews/suggestions on the progress are welcome.



Anshu Avinash

blog comments powered by Disqus