Data Slice
A data slice is an optimization feature that helps direct queries to the data of the appropriate partitions. A data slice is not a substitute for or an alternative to a partition's filter. That is, a data slice should not be used to limit the data selected from the partition's fact table and included in the partition. You can specify a data slice when you create a partition using the Partition Wizard.
A data slice of a partition should reflect as closely as possible the data in the partition. For example, if a partition is limited to 1997 data, the partition's data slice should specify the 1997 member of the Time dimension. It is not always possible to specify a data slice that reflects the exact contents of a partition. For example, if a partition contains data for only January and February, but the levels of the Time dimension are Year, Quarter, and Month, there is no way in the Partition Wizard to select both the January and February members. In such cases, select the parent of the members that reflect the partition's contents. In this example, select Quarter 1.
When the partitions of a cube contain data slices, Microsoft® SQL Server™ 2000 Analysis Services can more efficiently determine which partitions must be accessed to resolve queries on the cube.
Although a data slice sometimes modifies the WHERE clause that is used to populate a partition, this does not always occur. For this reason, whenever it is necessary to limit the data selected from a partition's fact table, you must use a filter, not a data slice.
You cannot slice partitions on a ROLAP dimension.