Using InetSoft for a BI Front-end to OLAP and OLTP Databases

Semantic Layer – Data Model

Data models are an easier way to expose business data. Data models can be created against all types of databases. A basic data model includes both a physical definition (tables, fields, and joins) and a logical mapping (business units, and readable names).

Logical Model Methodologies

Even though all data models share these same basic characteristics, different database schemas warrant different modeling methodologies. This holds true regardless of the purpose of your database. In other words, modeling a Data Warehouse in a ‘star schema’ is very similar to modeling a Data Mart in a ‘star schema’. However, a Data Warehouse in a multi-dimensional database is modeled very differently than a Data Warehouse in a ‘star schema’.

ER schema modeling needs to address a range of issues that make data access complex and difficult. The InetSoft data model includes four components:

  • Physical View – Transform
  • Logical Model – Business View
  • Virtual Private Model – Security
  • OLAP Overlay – Multidimensional Analysis
#1 Ranking: Read how InetSoft was rated #1 for user adoption in G2's user survey-based index Read More

A physical view captures database meta-data information and transforms the original schema into a business intelligence friendly schema. A logical model reorganizes a physical view and exposes data in a way that is easier to use and understand.

Virtual Private Models and hierarchical models are both optional. VPMs are the primary data security mechanism. They automatically filter rows and columns so that users only see the data they should. The OLAP overlay hierarchical model is a mapping that allows OLAP operations to be used on non-OLAP databases.

Normalized Schema

A normalized schema is typically used for OLTP databases. OLTP databases are efficient for recording data. Normalization reduces redundancy and maintains data integrity. The downside is that complex, hard to understand queries are required in order to retrieve meaningful data. An InetSoft Data Model, on the other hand, reduces this complexity to a minimum so that data access is widely available.

In the InetSoft environment, the physical view is the first component of data modeling. A physical view captures a set of tables and the relationships among them, and covers all or part of the database schema. A physical view is not created by mechanically copying the design model. To the contrary, schema objects, such as tables and relationships, need to be carefully segmented for simple access. More importantly, the physical view should serve to remove the traps that can cause incorrect queries to be produced by unsuspecting business users. It is common practice to model a single schema with multiple overlapping physical views.

The logical model is then constructed on top of this physical view. A logical model maps columns into attributes and creates new, derived attributes. Entities group related business attributes together regardless of the data item’s location in the physical view. This is not a simple regrouping. The logical model denormalizes or adds additional redundancy, in order to maximize simplicity for data analysts and end users.

view gallery
View live interactive examples in InetSoft's dashboard and visualization gallery.

Multidimensional hierarchies can be optionally created on top of the logical model. This is called an OLAP overlay because it is a logical structure only. An OLAP overlay defines attributes as measures and dimensions. Measures are typically numerical values which are additive, such as sales or inventory. Dimensions are the descriptions of measures such as the location, time, and product information for a sale.

Dimensions are essentially used as the criteria for selecting a subset of, or grouping, measure records. Dimensions often form a natural hierarchy, e.g., a product belongs to a category and a category belongs to a brand. Business users can then very easily use different combinations of dimensions and measures to analyze their performance.