Manage Cubes

A cube is an Analysis Services object that provides a multidimensional representation of data. When you build a cube in Analysis Services, you define measures and dimensions. Measures are the quantitative values in the database that you want to analyze (e.g., Mean Time Between Failures and Total Costs). Dimensions define exactly what you want to measure (e.g., Location or Equipment Type). Hierarchies define how the dimensions are aggregated.

Measures, dimensions, and hierarchies define a cube. For each intersection of a dimension and measure, a value is calculated. For example, the Mean Time Between Failures for Centrifugal Pumps in FCC Unit during 1999 could equal 152 Days.

The Metrics and Scorecards module provides access through the APM application to the cubes that exist in SQL Server Analysis Services, and allows you to develop Metric Views and KPIs using the data that is stored in those cubes.

An Analysis Services cube is a combination of measures and dimensions that together determine how a set of data can be viewed and analyzed. A cube is a static object and initially represents the data that existed in Analysis Services for the selected measures and dimensions when the cube was created. To keep a cube current, it must be processed regularly, whereby the cube is updated with the most current data in Analysis Services.

To make sure that a cube always provides users with the most current data, you should schedule it for processing regularly, usually on a daily basis. One way to process cubes and shared dimensions successfully is to do so manually on the Analysis Server. Using this method, you can process shared dimensions first, and then process the related cubes. Processing cubes manually, however, is not a viable option if you have many cubes that you want to process on a daily basis.

Instead, a preferable option would be to schedule cubes for processing using Data Transformation Services (DTS). This functionality is available in the SQL Server Business Intelligence Development Studio, which is included in SQL Server Standard Edition. For details on creating a DTS package that can be used to process objects according to a custom schedule, see your SQL Server documentation.

When you process a cube on the Analysis Server, either manually or by using a schedule, you have the option of choosing full processing or incremental processing.

• You should use full processing whenever structural changes have been made to the cube. Structural changes include:
  • Deleting APM records that will delete Members from a Dimension.
  • Updating APM records that will result in changing Member names.
  • Rearranging levels or adding and deleting dimensions.
• You can use incremental processing whenever you want to update the cube based on the most recent data, as long as changes to the data do not have any affect on the structure of the cube.

Note that when full processing is performed, Metric Views will become inaccessible to users who are currently logged in to the APM and have already accessed the Views. Users will need to restart the APM application to access Views after a cube has been processed using full processing, if they are logged in at the time the processing occurred. APM does not need to be restarted when incremental processing is used.

Therefore, if you are scheduling cubes for processing and need to use the full processing option, you should schedule the processing so that it occurs at a time when users are not using the APM system. If you are scheduling cubes that are unlikely to change, you can use incremental processing and schedule the processing to occur at any time.