Register now or log in to join your professional community.
Relational OLAP (ROLAP) stores the cube structure in a multidimensional database. The leaf-level measures are left in the relational data mart that serves as the source of the cube. The preprocessed aggregates are also stored in a relational database table.
Multidimensional OLAP (MOLAP) also stores the cube structure in a multidimensional database. However, both the preprocessed aggregate values and a copy of the leaf-level values are placed in the multidimensional database as well. Because of this, all data requests are answered from the multidimensional database, making MOLAP systems extremely responsive.
the ROLAP system does not add latency to leaf-level data. The disadvantage of a ROLAP system is that the retrieval of the aggregate and leaf-level values is slower than the other OLAP architectures.
A MOLAP system, therefore, does add latency to the leaf-level data. The MOLAP architecture also requires more disk space to store the copy of the leaf-level values in the multidimensional database. However, because MOLAP is extremely efficient at storing values, the additional space required is usually not significant.
reference: Delivering Business Intelligence with Microsoft SQL Server2005
by Brian Larson
