Database Throughput And Concurrency
Predictive Diagnostics makes predictions for the consumption of database resources (CPU, memory, and disk I/O) by generating a model that uses database throughput and database concurrency as inputs. Use the Database Throughput and Concurrency page to specify which database throughput and concurrency metrics are used when creating this model.
Note: Use this page to select database throughput and concurrency metrics for real data only.
Specifically, use the page to:
- Choose as default metrics those that are the best for representing throughput and concurrency for the Oracle database.
- Define a customized concurrency metric if the standard concurrency metrics (connected sessions and active sessions) are not appropriate.
To open the page
- Select the Spotlight on Oracle connection in the Spotlight Browser.
-
Click Predictive Diagnostics | Select database throughput and concurrency metrics.
How to use the Database Throughput and Concurrency page
Set the default throughput and concurrency metrics.
|
Throughput metric A measure of how fast data moves through the database. |
The first action taken when you open the Database Throughput and Concurrency page is an automatic analysis to recommend which throughput metric is the best (broadest) predictor for the utilization for the database resources. The analysis takes a short time, during which Spotlight displays a Calculating recommendation... message. You can choose NOT to accept the recommendation. (For example, you may want to select a different metric if the original analysis was performed on too little real data.) Choose from one of these Throughput Metrics:
| ||||||||||
|
Concurrency metric A measure of the number of users connected to the database at the same time. |
The choice of concurrency metric is usually straightforward. Connected Sessions is a realistic measure of the number of sessions. If you prefer concurrency to be based on specific TYPES of sessions, select Customized SQL and write your own SQL statement to count the sessions of the desired type. Choose from one of these Concurrency Metrics:
|
Specifying a customized concurrency metric
When you choose Customized SQL as the concurrency metric, you can choose an existing metric or set a new metric. Use the custom controls.
Custom controls
|
Control |
Description |
|---|---|
|
Description... |
Type a description used to identify the customized metric. |
|
{SQL statement} |
In the text pane, type the SQL statement that defines the customized metric. |
|
Test button |
Click to execute the SQL statement to verify that it runs without error and returns the expected value. (If the statement fails to execute, Spotlight displays the error encountered.) |
|
Save button |
Click to save the text of the edited SQL statement. |
Revert Changes button |
Revert to the saved version of the metric. |
Set the customized metric
|
To... |
...Do this |
|---|---|
|
Use SQL to define a new concurrency metric |
|
|
Edit an existing custom metric |
|
|
Revert to the saved version of the metric |
Click Revert Changes.
|
Notes on customized metrics: A customized metric is a SQL statement that can be executed on the database to collect data for Predictive Diagnostics. The data that is collected is linked to the NAME of the metric.
Take care when renaming metrics or re-using the names of old metrics:
- When you change the name of a customized metric in the Description... box, you create a new SQL statement; data that was collected under the old name still exists, but is no longer used.
- When you edit the text of the statement WITHOUT changing its name, OR when you re-use an OLD name, any new data collected (however different) is simply appended to the old data.