ASH: Introduction (Part 1 of 3)

Here, Italian version

There are tools that allow you a super fast sampling (10-100 times per second) using C programs. The goal of those tools is record information about a specific session. Really, in most case (the 99,9%) the monitoring can be done sampling 1 time per second, showing you both system load and system activity. This information could be difficult or too expansive to collect (there is also the problem that the tool impact the application itself) 

Most of the performance could be sampled, storing just the most important once and in a particular way: instead of to collect the exact statistics (information obtained instrumenting the C-code), ASH sample each second the status of all ACTIVE sessions (the session in "ACTIVE" state is a session executing a database call [1]) as well as the SQL that the session is running.

ASH stands for "Active Session History" and was introduced in Oracle 10g. The view of data dictionary is V$ACTIVE_SESSION_HISTORY. Over time the information collected from ASH have increased so release after release the columns of the view are grown in number. Following picture shows the change during the release

https://sites.google.com/site/embtdbo/wait-event-documentation/ash---active-session-history/ash_across_versions.PNG?attredirects=0

References

• [1] Active Session = session currently running (spending time) in a database call (sessions with status=ACTIVE in v$session view)
• https://sites.google.com/site/embtdbo/wait-event-documentation/ash---active-session-history