postgres_dba (https://github.com/NikolayS/postgres_dba) - is a brand new open source DBA tool set, which can be used by any application developer to find database issues possible solutions much faster.
In this talk, we'll discuss why is it important to have proper tools to analyze database health, and why DBA's work very often looks like black magic.
Taking PostgreSQL as an example (including its managed cloud versions like Heroku Postgres or AWS RDS Postgres), we'll discuss what should be done to make DBA's tasks of maintaining and optimizing databases more efficient and automated.
We'll cover these topics:
- controlling sizes of tables and indexes;
- controlling bloat level and autovacuum;
- index set optimization;
- major disadvantages of current approaches to group slowest queries;
- using machine learning to find, analyze slowest and optimize slowest queries iteratively.
More than 15 years of experience with various DBMSes, more than 10 years with PostgreSQL. M.S., MIPT and ISP RAS, speciality "Database systems". Co-founder Postila.ru, #RuPostgres (PostgreSQL user group for Russian-speaking users, the second in size on Meetup.com, 1400+ members), Postgres.tv. Russian press contact, PostgreSQL Global Development Group. Twitter: @postgresmen