Rhetorical question: Kimball vs Inmon DW characteristics

View previous topic View next topic Go down

Rhetorical question: Kimball vs Inmon DW characteristics

Post  andriy.zabavskyy on Thu Jul 12, 2012 9:59 am

Please help me clarify the key differences in classic characteristics of a Data Warehouses defined by so called Inmon an Kimball versions.
In a resource provided by DAMA institute there is mentioned following:
Inmon Version
Distinctive characteristics of data warehouses:

  • Subject Oriented
  • Integrated
  • Time Variant
  • Non-Volatile
  • Summarized and Detailed Data
  • Historical

Kimball Version
The only thing mentioned is "data warehouse is simply a copy of transaction data specifically structured for query and analysis". Different structure (dimensional model) to enable business users to understand and use data more successfully and to address DW query performance.

In my opinion this comparison looks at least not balanced.
Almost all Inmon's charachteristics mentioned above could be applied also to Kimball's version. Maybe it worth only to change subject oriented to be business lines oriented, and explaining the Non-Volatile term to add that in some cases (like Accumulative fact snapshot tables) updates are possible.

What do you think?
Thanks

andriy.zabavskyy

Posts : 18
Join date : 2011-09-12

View user profile

Back to top Go down

Re: Rhetorical question: Kimball vs Inmon DW characteristics

Post  ngalemmo on Thu Jul 12, 2012 2:28 pm

Both Inmon and Kimball can achieve the same results as it applies to characteristics of a data warehouse. Either can store the same information and support the same analytics. Both are Subject Oriented, Integrated, Time Variant, Non-Volatile, Detailed Data (may have summarized), and Historical. Other things I have seen on DAMA sites leads me to think they haven't quite 'got it' yet when it comes to dimensional modeling.

The difference is that Kimball believes you can achieve an enterprise data warehouse through the construction of atomic level facts and dimensions that integrate across conformed dimensions. Inmon argues the only way to achieve integration is to store the information in a somewhat normalized schema, then publish that information into forms more suitable for analysis (such as a star schema). I disagree with Inmon on this point. Properly designed dimensional models can achieve the same end.

It boils down to different methods to achieve a data model. The advantage of Kimball is that the stars are readily useable for analysis. Under the Inmon methodology, the data warehouse is off-limits to end-user queries. The data warehouse serves as a source for the creation of extracts and star schema that are then consumed by end-users.
avatar
ngalemmo

Posts : 3000
Join date : 2009-05-15
Location : Los Angeles

View user profile http://aginity.com

Back to top Go down

View previous topic View next topic Back to top


 
Permissions in this forum:
You cannot reply to topics in this forum