There are several aspects of tuning the performance of PostgreSQL. These aspects are related to hardware configuration, network settings, and PostgreSQL configuration. PostgreSQL is often shipped with a configuration that is not suitable for production. Due to this, one should at least configure the PostgreSQL buffer setting, RAM settings, number of connections, and logging. Note that several PostgreSQL settings are correlated, such as the RAM settings and number of connections. In addition to this, one should take great care with settings that require a server restart because these are difficult to change in the production environment.

Often, PostgreSQL produces a good execution plan if the physical structure of a database is normalized and the query is written properly. However, this is not always the case. To overcome performance issues, PostgreSQL provides the EXPLAIN utility command, which can be used to generate execution plans. The EXPLAIN command has several options, such as ANALYZE and BUFFERS. Also, one should know the limitations of PostgreSQL in order to write good queries, such as cross column statistics, CTE execution boundaries, and PL/pgSQL features. In addition to these, one should know the exact difference between different SQL statements and how they are executed, such as UNION, UNION ALL, DISTINCT, and so on. Furthermore, one should learn how to rewrite the same query in different ways and compare the performance of each.

Finally, PostgreSQL has many features to boost performance, such as constraint exclusion and techniques of table partitioning—mainly horizontal and vertical partitioning. Add to this many external tools to handle caching as well as connection pooling.