Prior to our certification class, we didn’t know that we could split a satellite based on ‘Type of Data’. We thought that a satellites had to be a direct copy from the source system table. In our training materials, there is a slide showing an example Satellite_Customer split into: Satellite_Customer_Name, Satellite_Customer_Contact and Satellite_Customer_Sales_Region.
Question: If you split a satellite based on type of data, can you hang them from disparate hubs? (Hub_Customer, Hub_Contact and Hub_Sales_Region in this example)
This may solve a problem where we have 2 source systems that define the same entity (Provider) at different grains. The business key is a common id (NPI). One source system has a provider table keyed by NPI but the other has a provider table keyed by NPI + Address. If we can split the satellite and hang them on hub_provider and hub_address with a link between them then I think we can resolve the different grain issue. One consideration is that any brand new attribute in the source system will have to be delegated to the appropriate hub satellite.