Since we upgraded our DB cluster from Aurora 2.X to 3.x (MySQL 8), our DB has been slow on some read ops (on read replicas). When we check AWS performance insights, we see many waits related to wait/io/table/sql/handler, which we never had on Aurora 2.X. Does anyone know how to resolve this? Is it related to a configuration?
The query issues issue seems to be intermittent since we can run the same query reasonably quickly under a second) but sometimes, the same query takes forever and reaches MAX_EXECUTION_TIME.
** I've already read almost any article related to this but couldn't find anything useful.