To use DRITA reports for performance tuning, review the top five events in a given report, looking for any event that takes a disproportionately large percentage of resources. In a streamlined system, user I/O will probably make up the largest number of waits. Waits should be evaluated in the context of CPU usage and total time; an event may not be significant if it takes 2 minutes out of a total measurement interval of 2 hours, if the rest of the time is consumed by CPU time. The component of response time (CPU "work" time or other "wait" time) that consumes the highest percentage of overall time should be evaluated.
Checkpoint waits may indicate that checkpoint parameters need to be adjusted, (checkpoint_segments and checkpoint_timeout). WAL-related waits may indicate wal_buffers are under-sized. If high, may need to decrease bgwriter_delay.