Skip to main content
Version: I2023.2.0

Performing Full Rebuilds of Incremental Aggregates

Incremental aggregates process only new windows of fact data rows, rather than processing all of the data. In some cases, you may want to reprocess all of the data to ensure that aggregates are accurate. For example, if the data of a dimension dataset has changed significantly or if you have older data that missed a processing window.

Before you begin

Your user should have the Login, View Aggregates, and Publish Projects permissions. For more information, see Creating and Editing Roles.

Procedure

To trigger a full rebuild of all of the aggregates for a cube:

  1. At the top of the Design Center, choose Projects.
  2. In the list of published projects, find the project that you want to specify a trigger file for and select the project to navigate to the project overview.
  3. Select the published version of the cube associated with the project, and select the Build tab to reveal the aggregate rebuild drop-down menu.
  4. From the Build drop-down menu, select Full to trigger a full rebuild (including incremental aggregates) for the aggregates within the cube.

To deactivate an aggregate, so that the AtScale engine will perform a full rebuild of it when the aggregates for the corresponding cube are rebuilt:

  1. At the top of the Design Center, click Aggregates.
  2. At the left, click Instances.
  3. Filter the list of aggregates by the Cube name.
  4. Locate the current instance of the aggregate that you want fully rebuilt and click Deactivate.

The next time AtScale builds aggregates for the cube, it will do a full rebuild of the selected aggregate. Then, incremental builds will continue after that.

note

Note: It might be the case that, after you deactivate the instance, the AtScale engine decides that it will not rebuild the instance. The engine might instead create a new aggregate definition that is based on the current aggregate definition, or the engine might create an entirely new aggregate definition that better serves the query workload against the cube.