In this scenario, we skip the aggregation step. We serialize the relational tables "as-is" as XML during the export process without remodeling the data.

Maybe, in theory, this scenario does not make too much sense, but is quite often applied in case of a data archive. Simply because the archive application provides the thin app out of the box, and also the data management workflow.

Then for fetching the relevant info, we have to do two things on demand:

So we need to write almost the same aggregator XQuery what we use in the previous scenario, but this is run on-demand by the end-user in real-time and not in advance as part of the data migration.