||Craig Shallahamer is a longtime Oracle DBA (1989) who specializes in Oracle Database performance. He is also an Oracle ACE Director, performance researcher and blogger, consultant, author of two books ("Oracle Performance Firefighting" and "Forecasting Oracle Performance"), an enthusiastic conference speaker, and a passionate teacher to thousands of Oracle professionals. He is the founder and president of OraPub and creator of the performance analysis tool Stori, and he clearly pushes the teaching envelope with his online seminars. His blog, A Wider View, is where to find his most recent public performance research. He can be contacted at firstname.lastname@example.org, @CShallahamer, and LinkedIn.
This is a true story. Out of nowhere, the OEM Activity Monitor shows an ominous 200-session five-minute spike, but the phone doesn't ring. The users are silent, but management is freaking out. They want to know why the incident occurred, will it happen again, what can be done so it doesn't happen again, when did the issue really begin and end, and how many users were affected.
An Oracle time-based analysis (DB Time) did not work. The five-minute incident mixed with the 60-minute instance level summarization was undetectable. Active session history (ASH) data was available, but the volume of data would make sifting through the data tedious and overwhelming, the proverbial needle in the haystack.
However, there was a way to figure this out. In this session, using ASH data with a couple creative but straightforward scripts, you will learn how to vary the level of ASH data summarization, converge on the problem area, and create a session-level timeline of the incident. Master this, and management will be satisfied and you'll know you've earned your wage that day.