These queries on ASH give an overview of the load on the database over time.
———————————————————————-
Without an overview of activity over time like in OEM 10g’s performance page, it is difficult to get a feel for what database performance looks like. The following two scripts are attempts to provide similar informantion via SQL plus.
Explanation of Average Activity Scripts (PPT)
Average Activity - query output the output column “graph” that shows a similar graph to OEM’s performance page. The “graph” column is composed of “+” and “-” signs. The “+” sign signifies average active sessions (AAS) on CPU and the “-” is sessions (AAS) waiting. The averages are done by default for each hour, but this could be changed to any time width.
Average Activity with top 2 waits - query output , same as aveact.sql but in addition it shows the top two waits for each time bucket
———————————————————————-
Just curious as to why you are counting occurrences of events/CPU instead of using wait_time/time_waited values in ASH.
Anthony, I think your question is “why am I graphing an events and cpu instead of wait time?” wait_time is just a proxy for whether we are waiting or on CPU. THe time_waited actually doesn’t tell us much about system load in ASH data since ASH is sampled we generally get long waits and miss short waits so the data is skewed. On the other hand the sampling of the number of active sessions is a great statistical approximation of load on the system and looking at how much o fhte load is waiting verses working (on CPU) gives us even more information.