The slow query log provides exact information about queries that take a long time to execute. However, sometimes there are a large number of queries that each take a very short amount of time to execute. This feature provides a tool for analyzing that information by counting and displaying the number of queries according to the the length of time they took to execute. The user can define time intervals that divide the range 0 to positive infinity into smaller intervals and then collect the number of commands whose execution times fall into each of those intervals.

http://www.percona.com/doc/percona-server/5.5/diagnostics/response_time_distribution.html

This feature is almost perfect to be ported as audit plugin. There are a few constraints though:

• FLUSH QUERY_RESPONSE_TIME. We will substitute it with system variable like "query_response_time_flush". Update of this variable will trigger flush.
• we won't port old-style "have_response_time_distribution" variable.
• we probably won't port old-style "SHOW QUERY_RESPONSE_TIME".