Shared cubes, such as those built by an Essbase/DB2 OLAP Services or Microsoft OLAP Services are not found in every...
data warehouse program. The benefits of using cubes are numerous and usually it's the performance (of specific queries) that is cited as the main benefit. However, one of the lesser-cited benefits of using shared cubes is one that I've found to be among its biggest benefits.
Cubes are not just the storage format. These physical cubes are accessed through specialized access layers that can make numerous assumptions about the way in which the cube data will be accessed. This access layer is one of the biggest benefits of cubes. You can get your users up and running with fairly robust slice-and-dice and drill-down capabilities just by building a cube and pointing an OLAP Services cube interface tool such as those by Essbase (for Essbase/DB2 OLAP Services) or netExs, Cognos, Crystal Decisions or ProClarity (for Microsoft OLAP Services) at the cube.
I call cubes "hyperdimensional" since they are modeled dimensionally, yet stored with every possible combination of fact-dimension values precalculated such that random access anywhere within the cube is fast.
Consider cubes when access requirements are known to be of a single fact table and its dimensions. But don't abuse cubes, building them over 40 GB and with so many dimensions that the rebuild, which is necessary to run often, can take too long. It's easy to get carried away adding dimensions to cubes, but remember the consequences of adding dimensions -- more space requirements and more time for the rebuilds.
For more information, check out searchCRM's Best Web Links on Data Warehousing.