>>>"After further review the call is reversed"
>>>
>>>Went into PowerDesigner and pulled up an old diagram and looked at it. You were correct, sir. I'm gonna think up a better one and post it tomorrow.
>>>
>>
>>If it helps, I have yet to EVER see a data design situation that required denormalization for anything other than performance.
>
>What about fault tolerance ? Hereby I mean redundant data that can be used for recovering purposes when a database crashes.
>

Replication is replication, and has nothing to do with data design, IMO.