The cartesian/cross join should be a last chance option as this could produce huge datasets.
One option would be to only perform this if the user explicitly requests this via a parameter. Additionally some logic should be applied to provide a warning message, allowing an abort, if the join will produce an excessive number of records (testing suggested 100m+ was causing memory issues).
The cartesian/cross join would only be applied for any records not handled by the exact matching process.
The cartesian/cross join should be a last chance option as this could produce huge datasets.
One option would be to only perform this if the user explicitly requests this via a parameter. Additionally some logic should be applied to provide a warning message, allowing an abort, if the join will produce an excessive number of records (testing suggested 100m+ was causing memory issues).
The cartesian/cross join would only be applied for any records not handled by the exact matching process.