IT industry has recognized that businesses benefit from a model layer over their data sources. Model layers can deliver high-performance query responses even over extremely large volumes of data, and they can encapsulate business rules and effectively secure access to data. The BI Semantic Model is a single model that serves all of the end-user experiences for Microsoft BI, including reporting analysis and dashboarding. The model can integrate data from a number of data sources, whether they are traditional data sources, such as databases or LOB applications, or nontraditional sources, such as OData feeds, text files, and spreadsheets.
The BI Semantic Model can consume data from traditional data sources, including SQL Server, Access, Oracle, Teradata, and DB2. In addition, tabular models can consume data from a wider variety of data sources, extending to Excel workbooks, data files, OData feeds, and cloud services, including Windows Azure SQL Database. The model then enriches and provides accelerated access to the data, and it exposes an intuitive interface that is surfaced by all tools in the Microsoft BI stack: Reporting Services (including the new interactive report authoring tool named Power View), Excel, PowerPivot, and SharePoint. It then explains the BI Semantic Model architecture, and describes—in broad terms—how tabular models can be developed. Finally it describes considerations to help you choose the best-fit approach for developing a BI Semantic Model.
Note: Some of the text is taken from white paper for better describing the paper.
Reference: Pinal Dave (http://blog.sqlauthority.com)