Again, some more charts of database usage. It shows some other types of the regular line graphs, and some other metrics than wait/io and queries.
Above are the cpu and wait metrics shown, but in a bar graph. The first one shows the complete time frame, in the second one only the data of the office hours are shown.
Chart shows trouble; enqueue waits in an OLTP system used by 50 to 100 users. Yes, the numbers on the y-axis are seconds ! So this means 12000+ seconds of production loss in one hour. Massive blocking locks, and a few deadlocks now and then.
This shows a lot of graphs of the same database of the same time span. Graphs show trouble.
Yes, again, y-axis shows seconds ! Massive ‘buffer busy waits’ and ‘latch: cache buffer chains’.
Abnormal ‘db file sequential read’ peaks correspondent with the peaks found in ‘buffer busy waits’ and ‘latch: cache buffer chains’.
Very high activity on temp-files. Graph shows ‘reads per second’ and ‘writes per second’ averaged over one hour.
What is Oracle doing with it’s sga_target in the meantime ?