Hi
I have 2 similar tables from 2 different source systems that have the same core identifiers but one table has some additional ones -
Should I be trying to get these sources in to the same link table or should they be separate? I could put the shared identifiers together in one link and then create separate links for the additional relationships (3 of these look like same-as links) however, I’m conscious that this could likely break the unit of work.
Could I create the link based on the 9 identifiers from source system 1 and then have source system 2 populate zero record keys for the identifiers it doesn’t have?
Would it just be safer to have them as separate link tables even though they are, conceptually, the same thing?
Many Thanks
Hey Martin
I could potentially give a straight forward answer but the core of your issue here is that you’re looking at this from the wrong end.
Look to your business not your source. They will know best if if the relationship they describe includes all of these entities or if it’s better to have different relationships between these entities. Nobody here will be able to answer your question as well as they can!
All the best,
Frankie
1 Like
Do you have replica business processes being automated? Should one be retired?
Ideally in the RV they should be separate but there are those that suggest that a Link can be multi-sourced— but then if you do that, one of the entities is not participating in the same business process, what do you do with those? Use the zero key? What if you have the same relationship being loaded to the same link table, one has a zero key, the other maps to an actual hub record (business key)? Which do you choose? Why* should you choose?
Define what you business case is, one** of those business process automation is the correct one. If you really need to fetch both, what is the rule that chooses one over the other? With two RV links you have the auditable unit of work in RV, a softrule may decide which to choose.
1 Like
They’re not replica business processes. The 2 source systems hold the same kind of data from 2 distinct market units. One source system was built from the core code of the other so they do have a lot of similarities in the core tables although over the years the 2 systems have gone in slightly different directions
