Deploying and Processing Cubes - Getting Started with SQL Server 2012 Cube Development - page 107

Configuring storage modes in Analysis
Services
As explained in the introduction to Processing objects section in this chapter, process-
ing reads data from the relational sources and stores it in Analysis Services. What
gets stored is defined through storage modes. There are three main storage modes
in Analysis Services and these are Multidimensional OLAP ( MOLAP ), Relational
OLAP ( ROLAP ), and Hybrid OLAP ( HOLAP ).
MOLAP
In MOLAP, everything is stored in Analysis Services, both the detail data as well as
all the aggregations. The data that you will see will be from the time of the latest pro-
cessing, this means that you need to schedule a refresh from time to time through
processing of the cube or database.
ROLAP
In ROLAP, nothing is stored in Analysis Services. Whenever you query the cube, a
query will be sent down to the relational source. This means that you will always get
the latest data and no processing needs to be done to fetch data from the source, as
this is done at query time. As no data is stored in Analysis Services, your performance
is due to the performance of the relational database that you use as the source for
your cube. ROLAP will put the burden on the relational database and if you are using
a non-dedicated source, this can cause issues with any other applications using the
same server.
TherearewaysbywhichyoucanincreaseperformanceforyourROLAPcube,acom-
mon method is to use the indexed view feature in SQL Server.
Tip
More about indexed views can be read at http://msdn.microsoft.com/en-us/library/
ms191432.aspx .