July 2004
Intermediate to advanced
384 pages
9h 32m
English
The database logoff trigger method for collecting historical performance data comes close to meeting all the requirements for root cause analysis established previously, except it lacks fine-grain data and a SQL statement repository. It looks like you need something between the trace event 10046 and the database logoff trigger. We can live with near fine-grain data, but we must absolutely have low overhead ongoing monitoring capability as well as wait event and SQL statement repositories. To achieve this, sampling may be the best approach. If you can sample every foreground process that connects to the instance at regular intervals and write the data to repository tables, you will have a history ...
Read now
Unlock full access