A data warehouse is designed, built and loaded with data. Fast
forwarding and looking forward to the next phase of the project, it would most
likely be implementing a sophisticated Business Intelligence (BI) reporting,
analytics platform. Enterprises - big,
medium and small, all like to make use of very sophisticated BI tools/software
or services to aid decision making.
In this post, let us explore how a "BI Metadata Model" is framed to enable capabilities for
authoring, viewing and modifying BI reports and interactive visualizations - online
or offline, in Microsoft Office applications or in-process applications, in the
office or on the go. In majority of the organizations, the tool used for BI
modeling is most likely to be provided by one amongst the leading BI
software/services vendors shown below in the Gartner BI Magic Quadrant
image:
 |
| Gartner BI Magic Quadrant - 2014 |
A BI tool supports in sourcing data from heterogeneous data sources (OLAP
cubes, relational database, spreadsheet, flat files etc.)and is not restricted to just a
data warehouse as its data source. An ideal BI metadata model can be defined in
terms of 3 layers: 1) Database layer 2)
Business Layer 3) Presentation Layer. All the layers can be equated to a
folder containing multiple database objects either in their original form or
modified form. The first image shown below is a diagrammatic representation of
an ideal BI Metadata Model. The second image below shows an actual real-world
Oracle BI metadata model with the above said 3 layers. Let us go through each
of the 3 layers to gain basic understanding of the use of building a metadata
model with 3 layers.
 |
| 3 layer BI Metadata Model |
 |
| Real-world Oracle BI Metadata Model |
Physical Data: BI modeler
imports all the required data base objects from the source data warehouse. Generally
tables, views, materialized views, stored procedures are imported to this layer
with very little modifications or changes. In this layer, these objects should
be related to each other to form the star-schema model. If there are multiple
star-schema, conformed dimension joins of many fact tables are also done so
enable drill-through capabilities.
Business Layer: This layer pulls
metadata (objects) from physical layer and supports any logical modifications
required to handle various business and user requirement. The database objects are all named the way a
data modeler understood them and not the way an user can understand. Static
filters can also be built inside the tables or views to implement data level
security. More such modifications arising from business requirements are all
handled here in this layer.
Presentation Layer: From the
name of this layer it is evident that the purpose of this layer is to create a very
presentable form of metadata model to the business users. Combining objects by
user group or by subject areas, adding object level security based on user
groups and roles are some of the main things done at this level.
Following this 3 layer approach a BI metadata model derives the
following advantages:
- Separates the logical behavior of the application from the physical model
- Provides the ability to federate multiple physical sources to the same logical object, enabling aggregate navigation and partitioning
- Provides dimension conformance and isolation from changes in the physical sources
- Enables the creation of portable BI Applications
- Hides the complexity of the source data models from the user
There are many other uses and advantages of having each of the 3 layers in a BI metadata model. It would require a blog post on each of the layer separately to discuss those uses and advantages in depth. However, in its core essence the 3 layer model supports easy of usability and simplicity of maintenance.
If your curiosity has now just gone to another level in terms of knowing what next after BI metadata modeling, stay tuned here for my next post on BI report development.
References:
http://www.informationweek.com/big-data/big-data-analytics/gartner-bi-magic-quadrant-winners-and-losers/d/d-id/1114013
http://docs.oracle.com/cd/E17904_01/bi.1111/e10540/intro.htm
http://www.rittmanmead.com/2011/12/agile-exadata-obiee-puzzle-pieces/
No comments:
Post a Comment