From: Norm Dennis
I'm not disagreeing Buck, just pointing out the obvious.
Data Warehousing was implemented to retain duplicate (redundant) data.
No, duplicate data may be a byproduct of a data warehouse, but not much of a reason for it, if at all. The primary reason for data warehouses is to support business analytics, and sometimes to house an aggregate of data from otherwise disparate sources (multiple business entities) or disparate platforms.
Buck made some extremely relevant points about the huge cost of managing duplicate data across disparate platforms. He deserves more credit. If the primary need is business analytics, then as others have said, you can write queries against an IBM i database as easily as writing queries against disparate databases.
Data replication has a place for business continuity purposes, however. Mirror data to a remote site in order to recover from disasters. Some IBM i organizations kill two birds with one stone. They mirror a production server to an offsite location, and perform business analytics and daily backups off the remote server.
Referential integrity constraints are not needed on a replicated server. I'd suggest using SQL Views to simplify and streamline interfaces between related tables.