We're using NOT LOGGED INITIALLY to create a table which act as a workspace for a an app. Since we're perfectly fine in re-creating the table in case of disaster, but we would like the table to persist in case of an application error, this seems like a good solution to avoid excessive transaction logs when populating the table.
However, we're getting some push back from our DBA team to the effect that, if there is a failure during the NIL transaction, it could impact (very badly) the entire tablespace. I haven't gotten a more precise explanation, and I can't seem to find anything in the IBM docs or on various forums that corroborate this risk.
Is there any circumstance where the use of NOT LOGGED INITIALLY can corrupt the entire tablespace? Anyone have any pointers to help me understand the risk, or refute the danger?
Thanks so much,
My understanding is that the risks associated with using NIL are that, if there's a problem with the transaction that isn't being logged, you'll lose your work, and if it's done on create we may need to drop the table on our DR site. That's fine.