Introduction

Prior to the introduction of tabular modeling, which is now commonly known as Business Intelligence Semantic Models (BISM), Microsoft relied on its multidimensional storage model (MOLAP) for Analysis Services (analytical) database. In fact, multidimensional refers to a method of storage, which is still a viable option for enterprise business intelligence through SQL Server Analysis Services (SSAS). The term BISM is not unique to tabular modeling—it also relates to the semantic abstraction of a data model within the MOLAP engine. However, whenever BISM is discussed, it usually relates to tabular modeling (whether that be in PowerPivot or SSAS with a tabular storage mode). The storage engine for tabular modeling is also referred to as xVelocity.

Unlike the xVelocity engine of tabular models, the multidimensional model was basically designed to use a relational source (preferably a SQL Server) as its data source. Additionally, the multidimensional engine assumed that data would be provided through a relational and conformed data structure.

In contrast to this requirement for a single source of data, tabular modeling is designed to support many different data sources which can then be combined in the model as a part of the modeling process. This offers the modeler the ability to combine multiple forms of data within a model, and therefore provides a richer modeling experience. By including additional data sources, model development time can also be greatly reduced, because there is no requirement to source the data from a traditional data mart (or data warehouse).

This chapter examines importing of data from various sources, and managing that data once it has been imported.

Although there are many data sources available for import (including many different relational database engines), this chapter focuses on some of the more ad hoc ones used by analysts, including text files, reports from reporting services, and data feeds. We also examine how the connection to the database can be managed once it has been created in the tabular model.