Neither of these variables tells me what I want to know: The highest number of concurrent connections since the last check (X seconds ago).
In the past, I've used
FLUSH STATUS to reset Max_used_connections some time after the critical level is reached, but this method isn't ideal for a number of reasons, and according to the docs:
This is something you should use only when debugging a query.
Before I head down the path of a complicated analysis of the query log, is there anything I'm overlooking?