The Physical Data Model - Microsoft SQL Server 2008 Analysis Services - page 363

Listing 20.1 shows an example of the definition for a partition that represents sales in
December of 1998.

LISTING 20.1

Definition of the Partition for December 1998

<Partition>
<ID>Sales Fact Dec 1998</ID>
<Name>Sales Fact Dec 1998</Name>
<CreatedTimestamp>0001-01-01T00:00:00Z</CreatedTimestamp>
<LastSchemaUpdate>0001-01-01T00:00:00Z</LastSchemaUpdate>
<LastProcessed>0001-01-01T00:00:00Z</LastProcessed>
<State>Unprocessed</State>
<Source xsi:type=”DsvTableBinding”>
<DataSourceViewID>Food Mart2000</DataSourceViewID>
<TableID>dbo_sales_fact_dec_1998</TableID>
</Source>
<StorageMode>Molap</StorageMode>
<CurrentStorageMode>Molap</CurrentStorageMode>
<ProcessingMode>Regular</ProcessingMode>
</Partition>

Physical Model of the Partition

Analysis Services uses several data structures to store partition data and to enable fast
access to this data during query execution:

. Slices, which enable fast detection of partitions that contain requested data

. Map stores, which contains the data loaded from the relational source

. Indexes built by the server for fast access to the data

. Aggregations, which enable fast access to the aggregated data (data that contains

totals)

. Aggregation indexes, which enable fast access to the data stored in aggregations