"There are primary three ways in which we store information in OLAP:
Multidimensional OLAP (MOLAP)
and fact data in a persistent data
store using compressed indexes. Aggregates are stored to facilitate fast data access.
MOLAP query engines are usually proprietary and optimized for the storage format
by the MOLAP data store. MOLAP offers faster query processing than ROLAP and
usually requires less storage. However, it doesn’t scale as well and requires a
database for storage.
Relational OLAP (ROLAP)
stores aggregates in
relational database tables. ROLAP use
of the relational databases allows it to take advantage of existing database resources,
plus it allows ROLAP applications to scale well. However, ROLAP’s use of tables
store aggregates usually requires more disk storage than MOLAP, and it is generally
As its name suggests,
Hybrid OLAP (HOLAP)
is a cross between MOLAP and ROLAP.
Like ROLAP, HOLAP leaves the primary data stored in the source database. Like MOLAP,
HOLAP stores aggregates in a persistent data store that’s separate from the primary
relational database. This mix allows HOLAP to offer the advantages of both MOLAP
and ROLAP. However, unlike MOLAP and ROLAP, which follow well-defined standards,
HOLAP has no uniform implementation. "