We are building the test framework described in the book “The Data Vault Guru” and wondering about one specific test: the orphan check in the satellite (SAT_SKEY_ORPH): why does this query not use DV_TENANTID in the join just like orphan check in the link?
I wrote that a while ago… but aren’t raw-vault satellite tables single sourced and therefore will only ever be of one tenant?
Hey @frank_boekamp,
adding a bit in general, not speaking about @patrickcuba’s book.
Single source satellites are my favorite, too but it’s not “the” way, it’s best practice.
If we absorb a multi-tenant system I’d say, that’s one source and I bsorb that one customer table where tenant ID is a part of the identifier. I wouldn’t split it different satellites.
Why? A new tenant is quickly set-up by configuration in the source system and this done, be keep absorbing everything but not with a sat split by tenant (we have to implement a new sat upfront).
If we’re having some different system instances, we should stick to different satellites. Maybe they’re in synch now but improbable patched and upgraded exactly in synch.
Some thoughts abut the topic in general …
Warm regards
Volker
Tenant-id is a separate concept and column built into every hub, link and satellite. Absorbing a multi-tenant system is not the topic of this conversation.
1 Like
That’s the answer, thanks!
1 Like