Data integration has multiple aspects

    *  Technical integration — Ensures that the same data formats are used by converting data types, performing basic calculations, and basic data-quality and plausibility checks (such as adding missing information using defaults or derived values).

    *  Semantic integration — Ensures that a common coding of values is used across all objects in the EDW (for example, that the gender is always represented as “M” for male, “F” for female, or “initial” for unknown). While the gender example is a simple one, implementing semantic integration can become very expensive in heterogeneous environments.

    *  Process integration — In many cases, business processes and their respective operational systems have to be enhanced or modified to fully support the requirements for integrated data in the EDW. For example, if a customer group is not maintained on a certain sales system, while it is crucial for analysis purposes, that system and the corresponding business processes must be enhanced to enforce maintenance of customer groups.

    *  Analysis integration — Ideally, the same objects or objects derived from one another are used in all reporting and analysis processes across all systems of the topology to ensure maximum integration.