On Wed, 5 Jan 2005 11:52:39 -0800 (PST), David <thump@(protected)> wrote: > I have plans to dunduct that very test now due to the varying responses > and if I'm to recommend such changes then I dman well better be able to > recover. > > Came here first to learn from others first hand experience and avoid > wasting valuable time. > > I'll update with my findings. > Issue does'nbt seem to be as cut and dry as the intial responses would > make out.
It seemed to me that the responses were clear. Yes the *database* is recoverable (subject to the usual caveats of actually having everything you need for a restore), the NOLOGGING changes will be lost - that might mean objects - it could if you allow operations like INSERT /*+ APPEND */ ... mean that data is lost from recovered objects.
If you decide to go the other way (require that all operations are recoverable) then ALTER DATABASE FORCE LOGGING is a nice addition from 9i.
I'd still do the test, not least because I don't regard saying to my boss "I suggest that we do XXXXX, some experts on the Net said it was a good idea" as a career enhancing approach to data management.
