Distributed Partitioned Cubes
Distributed partitioned cubes are regular cubes that employ partitions on multiple Analysis servers. Partitions stored on Analysis servers other than the Analysis server which stores the meta data for the partition are referred to as remote partitions and are very useful when dealing with complex cubes based on dimensions having millions of members.
Distributed partitioned cubes enable the processing of queries to be distributed across multiple Analysis servers, because the processing for each partition is performed on the Analysis server storing the partition data. In this way, performance can be increased when querying and processing distributed partitioned cubes. However, the performance increase can be impacted by network speed because information is passed from the remote Analysis server to the Analysis server processing the distributed partitioned cube.
As with other cubes, a distributed partitioned cube must be processed before client applications can browse it. Processing a distributed partitioned cube establishes the links to the remote Analysis servers containing the data for the remote partitions. The meta data for a remote partition can be viewed and updated only on the Analysis server where the partition and its parent distributed partitioned cube are defined. It cannot be viewed or updated on the remote Analysis server. Thus, a cube that contains remote partitions can be administered along with its partitions on a single Analysis server.