| Query suddenly slow after a statistics job, patch or upgrade | New plan hash value in AWR plan history; estimated versus actual rows far apart | Correct statistics, then pin the known good plan with a SQL plan baseline |
Screens slow, db file sequential read dominates | Top SQL doing many single-block index reads; nested loops over large row counts | Better index or join method so fewer rows are visited |
Reports slow, db file scattered read or direct path read | Full scans of large tables in top SQL by physical reads | Selective indexes, partition pruning, or accept scans for batch work |
| Commits slow, application waits on saves | log file sync high; redo write latency in AWR; commit rate per second | Batch commits, faster redo storage, right-size redo logs |
| Users block each other | enq: TX - row lock contention; blocking sessions in ASH | Shorter transactions, application change to the locking pattern |
| High CPU with many sessions parsing | High hard parse rate; library cache and cursor: pin S waits | Bind variables in the application; cursor sharing as a stopgap |
| Inserts slow under concurrency | buffer busy waits, enq: SQ or RAC gc waits on the same blocks | Larger sequence cache, hash-partitioned or reverse-key indexes |
| Heavy temp usage, slow sorts and hash joins | Multi-pass work areas; temp space in SQL Monitor | PGA sizing and plan fixes |
| CPU saturated, nothing obvious waiting | DB time mostly CPU; top SQL with very high logical reads per execution | Tune the top SQL before considering more cores |