Once we are done finding missing indexes, we can inspect the memory and I/O. To figure out what is going on, it makes sense to activate track_io_timing. If it is on, PostgreSQL will collect information about the disk wait time and present it to you.
Often, the main question asked by a customer is: if we add more disks, is it going to be faster? It is possible to guess what will happen, but in general, measuring is the better and more useful strategy. Enabling track_io_timing will help you gather the data to really figure this out.
PostgreSQL exposes disk wait time in various ways. One way to inspect things is to take a look at pg_stat_database:
test=# \d pg_stat_database
View "pg_catalog.pg_stat_database"
Column | Type | Modifiers
----------------+--------------------------+-----------
datid | oid |
datname | name |
...
conflicts | bigint ...