Monday Mar 28, 2011

Oracle BI Server Modeling, Part 3- Conformed Logical Dimensions

One of the primary business benefits of the Common Enterprise Information Model (CEIM) is enabling goal-based visibility and performance management across the enterprise. Business managers need to compare actual performance side-by-side with targets, such as actual sales compared to the sales forecast. The ratio of these two metrics– actual sales as a percent of forecast– can be compared to a threshold to create automatic stoplights and alerts. To measure and improve effectiveness, managers also need to compare measures across processes, such as a ratio of product shipments to inventory levels (inventory turns). This post lays out the basic concept for tying data from these far-flung sources together so they can be interactively analyzed as a whole, automatically manufacturing queries and handling the exception cases for the business users.

In Part 1 of this series, I briefly introduced that basic concept: conformed logical dimensions in the business model layer. In Part 2, I reviewed general dimensional concepts as applied to a single measure. In this post, I will describe the dimensionality of the business model as a whole, and how the conformed dimensions enable business performance visibility.

The concept of grain is a key to understanding conformed dimensions. This post shows how individual measures have differing grain within the overall conformed dimensions of the business model, and how grain is presented in the end user’s analytical experience. Later, this concept will be essential to understanding all the posts covering the logical-physical mapping patterns.

[Read More]

Thursday Mar 24, 2011

Oracle BI Server Modeling 2- Dimensional Schema Shapes

A past marketing tag line for Oracle BI EE was, “An information modeling approach.” In my experience, thinking in terms of information modeling rather than just SQL is one of the keys to developing the Common Enterprise Information Model (CEIM) successfully. When I’m designing a query factory, I need to think in terms of the models provided in the data sources, the simplified model I want to present to the users, and the schema shape transformations from one to the other. The data shape changes over and over throughout the query cycle, from the source schema, to the physical result sets, to the logical table sources, to the business model, to the Logical SQL (LSQL) result set, to the visualization. This is sort of like a*(b + c) = a*b + a*c in algebra, where the different shapes are good for different purposes, but give the same final result. Denormalized, normalized, star, snowflake, multicube, hypercube – the same information can be presented in all of these shapes.

The purpose of this post is to help you become familiar with dimensional shape-changing so you can understand the modeling patterns in the following posts. This will be important for understanding how to design the mappings in the CEIM – being able to recognize the dimensional structures within a larger physical schema, and then map them to the dimensional business model. There is also a bit of review of dimensional concepts and terminology.

[Read More]
About

Best Practices, Tips and Tricks from Oracle BI Development

Search

Archives
« March 2011
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
25
26
27
29
30
31
  
       
Today