Skip to main content

Performance: Dataset analysis

Performance problems in an application are often caused by datasets. In addition to following the rules of thumb, you can use dataset performance analysis to identify any specific issues that are causing a dataset to slow down your application.

Analyze dataset performance​

To analyze the performance of a dataset:

  1. Open your application in the appropriate DTAP environment and either:
    • Open Developer Tools, select the Datasets tab, and load the dataset. Then, open the Performance analysis tab and copy the string to the clipboard. For more information, see Developer Tools.
    • Open the view containing the grid or tile widget that uses the dataset and then open Developer Tools and select the Log tab. Reload the dataset (for example, by refreshing the page or showing additional columns) and then select the log row containing the getDataset socket. Then, click the graph icon and copy the string to the clipboard.
  2. Return to the Studio and open the dataset designer for the relevant dataset.
  3. On the toolbar, click the data analysis icon and then paste the string that you copied to the clipboard and click OK.

The results are displayed in the designer. A green check mark indicates that a column or condition can be retrieved from the database directly. Red exclamation marks are displayed for:

  • Conditions (and associated columns) that cannot be retrieved directly from the database. Note any runtime conditions. For OR conditions, move your mouse over the condition to determine which conditions are a problem and which are merely a consequence.
  • Sorts (and associated columns) that cannot be retrieved directly from the database. Note any runtime sorting.
  • Columns that cannot be retrieved directly from the database.

Optimize dataset performance​

Unfortunately, there is no standard step-by-step process for resolving the above issues. Potential solutions include:

  • Ensuring data types are set correctly and used consistently in templates.
  • Simplifying your templates. Sometimes something can be done in more than one way (use rules of thumb).
  • Using cached computed attributes to avoid real-time calculations.
  • An extra (AND) condition that is fast (green check mark and much filtered out).

You can also experiment to see if the dataset is faster when:

  • Certain columns are not queried.
  • A sort order is applied to particular columns.
  • No sort order is applied to any of the columns.

If all check marks are green, apply the rules of thumb for extremely large datasets.