SAND CDBMS supports a form of "Massively Parallel Processing" (MPP), where a table's data is partitioned among multiple computers in a network to take advantage of simultaneous intra-query processing. When intra-query results are received from each node, they are combined, and the full query results are transparently returned to the end user.
Benefits of MPP include the following:
- Query speed is usually improved if the queried data is divided into smaller chunks on independent, connected computers and accessed simultaneously
- Performance improvement is scalable
- Huge tables/databases are not a bottleneck for the system
- Database growth is not really an issue, since more nodes can always be added
Furthermore, dimension tables and domains can be automatically replicated and maintained on all of the partition nodes, so that a partitioned table can be used as a central fact table with fast dimension lookups at each node.