Question

I understand the differences in ROLAP, MOLAP, HOLAP i think but does that mean that ROLAP never needs to be process since you always getting data directly from database?

Answer 1

In MOLAP you’ll store the aggregations in the data directories of Analysis Services. Using ROLAP you’ll store the aggregations in indexed views in the relational database instead. ROLAP will give you a real-time slower approach than MOLAP, but is good for historical stuff that you need to look at once in a while.

From MSDN:

If a partition uses the ROLAP storage mode and its source data is stored in SQL Server Database Engine, Analysis Services tries to create indexed views to contain aggregations of the partition. If Analysis Services cannot create indexed views, it does not create aggregation tables. Although Analysis Services handles the session requirements for creating indexed views on SQL Server Database Engine, the following conditions must be met by the ROLAP partition and the tables in its schema in order for Analysis Services to create indexed views for aggregations:

• The partition cannot contain measures that use the Min or Max aggregate functions.

• Each table in the schema of the ROLAP partition must be used only one time. For example, the schema cannot contain [dbo].[address] AS “Customer Address” and [dbo].[address] AS “SalesRep Address”.

• Each table must be a table, not a view.

• All table names in the partition’s schema must be qualified with the owner name, for example, [dbo].[customer].

• All tables in the partition’s schema must have the same owner; for example, you cannot have a FROM clause that references the tables [tk].[customer], [john].[store], and [dave].[sales_fact_2004].

• The source columns of the partition’s measures must not be nullable.

• All tables used in the view must have been created with the following options set to ON:

• • ANSI_NULLS

• • QUOTED_IDENTIFIER

  –

• The total size of the index key, in SQL Server Database Engine, cannot exceed 900 bytes. SQL Server Database Engine will assert this condition based on the fixed length key columns when the CREATE INDEX statement is processed. However, if there are variable length columns in the index key, SQL Server Database Engine will also assert this condition for every update to the base tables. Because different aggregations have different view definitions, ROLAP processing using indexed views can succeed or fail depending on the aggregation design.

• The session creating the indexed view must have the following options set to ON: ARITHABORT, CONCAT_NULL_YEILDS_NULL, QUOTED_IDENTIFIER, ANSI_NULLS, ANSI_PADDING, and ANSI_WARNING. This setting can be made in SQL Server Management Studio.

• The session creating the indexed view must have the following option set to OFF: NUMERIC_ROUNDABORT. This setting can be made in SQL Server Management Studio.

Read the whole thing here.