Compare
Use the Performance Analysis Compare where differences in period activity illustrate underlying performance and monitoring issues. It helps you determine whether a comparison occurrence is an isolated incident or a sign of a potentially significant performance problem. The comparison can be of: whole instances or selected dimension breakdowns (such as user or SQL). Use Compare to address questions such as:
What caused a specific activity?
What were the resource and load demands of today's instance activity compared with that of a previous day's?
Is an instance imposing different load levels now than previously?
How do we explain the difference in a SQL performance compared with a previous period?
What are the differences in program characteristics over two periods that caused different performance?
After identifying the different resource usage, you can use compare to identify what caused this difference: Usage pattern? Different SQL behavior? Environmental problem?
Overall Activity
The upper third Overall
Activity panes graphically displays the Workload and SQL Executions
and itemizes those wait events whose activity is significantly (
) or moderately (
) different between the two
comparisons. Expanding a specific wait event displays graphically its
occurrence over the source (base) and target (observed). (Workloads which
are similar over the source and target are designated with a yellow icon
(
)).
Use this section to understand which wait states are most resource-demanding and likely causing the difference in resource usage or whether the difference in behavior can be explained by the difference in quantity of SQL executions.
Note: The graphical comparisons are on an absolute time basis, with the period commencements aligned.
Activity Highlights
For the wait event in focus, the middle Activity Highlights section displays the dimensions (of up to the top five) which were significant elements of the difference. Expanding the individual lines displays the dimension members which caused the difference and the composing metrics whose differences exceed the specified threshold.
Use the set of performance related metrics (defined by the chosen resource) to help you explain the difference in resource consumption; for example, a rise in I/O Wait might be explained by a rise in the quantity of physical reads.
All the Comparison Results thresholds and metric totaling methods are configurable.
Environment
The Environment section displays key metric values as occurred in both source and target, regardless of their absolute or relative value or difference. The displayed metrics reflect total instance or system activity, which are considered as an environmental factor that might affect the application behavior. Use this section to understand whether system or instance level problems are externally affecting your application so as to be the root cause of a problem (rather than the application itself). For example, an increase in the non-Oracle CPU consumption might explain an increase in an Oracle CPU wait state (as external applications consume more CPU, Oracle must wait more for CPU to become available).