Select Data Storage (Storage Design Wizard)
In this step of the wizard, you specify the type of data storage you want to use to store the data and aggregations for a cube or partition.
For more information about data storage choices, see Flexible Data Model.
Options
MOLAP
Select this option to store the data for your cube or partition in a multidimensional structure. The aggregations you design for this storage type will also be stored with the multidimensional data.
Multidimensional OLAP (MOLAP) storage provides the potential for the most rapid query response times, depending only on the percentage and design of the cube's aggregations. In general, MOLAP is more appropriate for cubes with frequent use and the necessity for rapid query response.
ROLAP
Select this option to keep the data for your cube or partition in the existing relational data store. Aggregations designed for relational OLAP (ROLAP) will also be stored in the relational database, rather than in a multidimensional structure.
ROLAP query response is generally slower than that available with MOLAP or HOLAP. A typical use of ROLAP is for large datasets that are infrequently queried, such as less recent historical data.
- Enable real-time updates
Select this to enable the partition to support real-time updates. This option is available only in Analysis Services for SQL Server™ 2000 Enterprise Edition, and only when the data source is SQL Server 2000 Enterprise Edition.
HOLAP
Select this option to keep the data for your cube or partition in the existing relational data store and to keep your aggregations in a multidimensional structure.
For queries that access summary data, hybrid OLAP (HOLAP) is equivalent to MOLAP. Queries that access base data, such as a drilldown to a single fact, must retrieve data from the relational database and will not be as fast as if the base data were stored in the MOLAP structure. Cubes stored as HOLAP are smaller than equivalent MOLAP cubes and respond faster than ROLAP cubes for queries involving summary data. HOLAP storage is generally suitable for cubes that require rapid query response for summaries based on a large amount of base data.