Wednesday, November 4, 2009

Data Model Extensibility

One of the major benefits that creating abstract tiers to a data model provides is that ability to then change the physical representation of the data independently from down-stream applications that are consuming that same data and upstream applications that are delivering the data.

Rules of the road:
  • Users never access base tables directly.  They always go through synonyms or views even if those views are simply "select *" views.
  • ETL always uses staging tables that are source-specific.  Use default column valeues.
  • Normalize base models, but not too excess.  Normalize them to a sound degree that aligns somewhere between the shource system and the downstream data.

Word Count :5,201

No comments:

Post a Comment