Properly utilizing the PostgreSQL shared buffer cache is vital for optimizing database performance, but most recommendations don't go beyond just giving a rough guidelines for how big it should be. Starting with an introduction to how the cache is structured and operates, this talk goes deep into using the pg_buffercache tool to monitor what in your database your applications are using the cache for. It finishes off with suggestions for using this information for tuning how big the cache is and the related background writer parameters. This talk is mainly aimed at UNIX-like platforms.
Greg has been using PostgreSQL to write critical business applications since V7.0 in 2000, graduating to patch contributer during development of 8.3. He currently works as a software engineer for Truviso, focused on applying the company's PostgreSQL-based streaming database technology to applications in the financial markets.