DW - Data Warehousing & Modeling

SAP BW Aggregates

Aggregates
are ‘mini-cubes’ created on top of an existing InfoCube, to pre-aggregate data
for faster reporting.

Typically they are created after a SAP BW go-live as part
of the support when performance problems arise.

SAP BW Process Chain

The
sequence of ETL load steps are defined in a Process Chain. A chain can contain steps of different types
(InfoPackages, DTP’s, ABAP programs, event triggers, BPC processes …).

Via the Process
Chain monitor the BW administrator can monitor and schedule BW load processes.

Data Warehouse Automation

As Business Intelligence is becoming mature and more of a commodity, there is pressure to achieve things in a faster & cheaper way. In terms of data warehouses & business intelligence this means automation. For example, on a daily basis, we are building & populating fact tables or adding slowly changing dimensions type II functionality to dimensions over and over again.

ODS - Operational Data Store

The main reason why an ODS or Operational Data Store is often build is to give users the possibility to execute operational reports. An ODS can be seen as a (partial) copy of the source system. It typically is also then the source for ETL related activities. The granularity is the same as the source system.

Slowly changing dimension

Slowly changing dimensions are a set of data warehouse techniques to deal with history within a dimension.

Dimensions are related to facts and while the facts are basically always transaction oriented with an associated date, a dimension is not. Therefore a dimension may require history as well.

3 basic types exist:

Star schemas

Two "dimensional modeling” techniques are generally used for modeling the data structures within a data warehouse context.

These are the "star schema” and "snowflake schema” techniques. Both find their roots within the Kimball dimensional modeling techniques.

A star schema is the preferred technique within the Kimball approach.

Properties of star schemas :

Data Vault

This is one method in which a database or data warehouse can be designed. Data sourced from different operational systems is stored in such a way that it can be completely traced in term of history. In the data vault architecture you can have three basic entities: hub, link and satellite.

Fact table

A fact table contains all numeric information for one single business process in context of multiple related dimensions.eg. a sales fact contains number of sold quantities, revenue, margin, etc. The granularity is ideally kept as low as possible in order to enable a variety of reports and analyses.

Pages

S'abonner à RSS - DW - Data Warehousing & Modeling