PostgreSQL 10 Administration Cookbook - Fourth Edition

There's more…

If not enough of the data fits in shared buffers, lots of rereading of the same data happens, causing performance issues. In pg_statio_user_tables, watch the heap_blks_hit and heap_blks_read fields, or the equivalent ones for index and toast relations. They give you a fairly good idea of how much of your data is found in PostgreSQL's shared buffers (heap_blks_hit) and how much had to be fetched from the disk (heap_blks_read). If you see large numbers of blocks being read from the disk continuously, you may want to tune those queries; or if you determine that the disk reads were justified, you can make the configured shared_buffers value bigger.

If your shared_buffers parameter is tuned properly and you can't rewrite the query ...