The UNICOM Intelligence Data Model is not a data structure. Rather it is a common and open way of accessing any data structure (both metadata and case data). "Data Access Model" would actually be a more descriptive name.

The benefit of the UNICOM Intelligence Data Model is that it makes it easier for applications to be compatible with each other, even when they have been developed by different companies. Moreover, applications using the UNICOM Intelligence Data Model are not tied to any one specific data structure. In fact, they can communicate with any data structure (provided there is a suitable DSC for that data structure).

When developing an application, dealing with the data can account for as much as 40 or 50 percent of the development time. Because the UNICOM Intelligence Data Model handles the communication with the data, it can reduce application development time dramatically. Furthermore, the UNICOM Intelligence Data Model has built-in functions, expression evaluation, and scripting capabilities that make it easy to work with market research and other survey data.

The ability to return the virtual tables and support the SQL syntax is actually built into the UNICOM Intelligence OLE DB Provider and is totally independent of any database software. This means that the UNICOM Intelligence Data Model does not require SQL Serverto be installed.

However, you would need to have access to a SQL Serverinstallation if you want to read or store data in a SQL Server database using the RDB DSC or your own DSC, but this is not a requirement for using the UNICOM Intelligence Data Model to read and store data in other formats.

Modern databases tend to be normalized, which means that the data is separated into multiple, related tables rather than a single table. One reason for this is that databases have limits on the number of bytes and columns that a table can contain. However, the tables returned by the UNICOM Intelligence OLE DB Provider are not database tables. They are virtual tables, which means that they are held temporarily in memory and do not actually exist in a database or on disk. This means that they are not subject to the limitations placed on database tables. The virtual tables provide a standard view of the underlying data, regardless of its format (which could be a Quanvert database, an IBM SPSS Statistics .sav file, an ASCII file, a normalized database, and so on).

Depending on the CDSC being used, UNICOM Intelligence Data Model can return two views of the data: a flat view and a hierarchical view. The flat view consists of a single virtual table (called VDATA), which, as you say, is not normalized. However, the hierarchical view consists of a set of hierarchical tables (called HDATA), each one of which has a primary key. This means that the hierarchical tables are normalized (but they are not a relational database).

Note that RDB DSC uses a normalized relational database schema when it stores data in a SQL Server database.