In the context of TDP is the logical application of transformation of data a characteristic pattern.. This pattern includes aspects like extract data from the source, transformation of the model or the protocol and selection of the source and the target model.
Transformation exists at two levels in the hourglass. The first is transformation from the source data to the target data in a standardized model. The second is the transformation of the generic model to a specific model in use by the consumer. The last one is not explicitly modeled in the hourglass model because this is considered as an implementation of the consumer.

Description of generic requirements for all the transformation functionalities

Specific requirments for specific application functions like specific transformations

Transformation of a datafile (mostly semi structured) for example an XML file with an internal datamodel that has to be transformed to the standardized model.

Transformation of a XML stream (mostly semi structured) with an internal or specific datamodel that has to be transformed to the standardized model in the data target.

Transformation of a NoSQL datasource to a standardized data target

Transformation of relational data entities (stored in a relational database) to the standardized data target.

Transformation function from the specific source data model to the standardized target data model. This can be a model or a protocol transformation.
Often this transformation is divided in sub steps and these sub steps will be analyzed and modeled in the

--