Differences Between Flow and Dataset Modeling

1. Understanding Datasets as Container Components:

• Datasets are container components that hold one or more source datastores.
• These datastores are related using filters and joins.
• To other components in a mapping, a dataset behaves like any other projector component (e.g., a datastore).
• The results of the filters and joins inside the dataset are shown on its output port.

2. Relation in Datasets:

• Within a dataset, data sources are related using relationships (e.g., joins, filters) rather than using a flow.
• These relationships are visually represented in the entity relationship diagram.

3. Physical Tab in Mapping Editor:

• When switching to the physical tab in the mapping editor, datasets disappear.
• ODI (Oracle Data Integrator) models the physical flow of data in the same way as if a flow diagram had been defined in the logical tab of the mapping editor.

4. Dataset vs Flow Metaphor:

• Datasets follow the ODI 11g way of organizing data sources, which is different from the flow-based metaphor used in ODI 12c mappings.
• When importing projects from ODI 11g, interfaces converted into mappings will contain datasets that group the source datastores.

5. Creating New Mappings:

• When creating a new, empty mapping, you are prompted to decide whether you want to include an empty dataset.
• You can delete the empty dataset if you don’t need it, and it can also be added back to any mapping at any time.
• The option to include an empty dataset is for convenience and is not mandatory.

6. Dataset’s Scope:

• A dataset exists only within a mapping or reusable mapping.
• Datasets cannot be designed independently as separate objects outside of a mapping or reusable mapping.

These steps summarize the key differences between flow and dataset modeling in ODI. Datasets are container components that use relationships (joins and filters) to organize data, whereas flow modeling is based on a continuous flow of data.