Skip to Content

Funny question to ask but then came across this question in a book that I was reading… This basically talks about the two schools of data warehousing as preached by the messiahs of data warehousing , namely Ralph Kimball and Bill Inmon.

So what is it that I am trying to say ?
The schools of thought are basically whether you have a data driven data warehouse or a metric driven data warehouse.

What are they exactly…?

Data Driven Data Warehouse.

This refers to the Inmon point of view where all the information relating to the organization should flow into the data warehouse and that analysis should be supported on this by the data warehouse….
This would mean that all the transactional data flows into the data warehouse and this would in turn support the corporate information factory view that is advocated by Bill Inmon.
http://www.inmoncif.com/

Metric driven data warehouse.

This is a slightly different tack where you look at your reporting by way of key metrics which you want to monitor and build the data warehouse basis that. This also calls for a detailed dimensional model where the various dimensions by which these metrics need to be analyzed are detailed out and carefully modeled.

Will detail out both schools of data warehousing soon… but then are you a Kimbalite or an Inmonite?

To report this post you need to login first.

6 Comments

You must be Logged on to comment or reply to a post.

  1. Witalij Rudnicki
    … (although obviously we see Inman’s model almost on every second presentation from SAP/BI). Will be interesting to know if they are really two different schools or just different names for the same.

    Keep posting!
    -Vitaliy

    (0) 
    1. Arun Varadarajan Post author
      I share the same opinion too .. but then seemed interesting. Came across this quote in an old research paper written in about 2002 … wanted to know other’s views on the same …
      Arun
      (0) 
  2. Vijay Vijayasankar
    My understanding is that these two approaches only differ ever so sightly – a metric driven system still needs transaction data to be brought into the datawarehouse. The metrics are typically in picture only at the level of cubes.

    SAP’s approach with business content has always been to give extractors for all (or most) transaction data in OLTP, and expect clients to model the rest. Very rarely does SAP provide cross domain analytics as business content.
    Basically, you have to model your own structure to compare say marketing vs sales vs inventory across the same dimensions. I personally like it this way, since every organization has its own way of reporting and we can pick and choose. However, since SAP is emphasising more on SMB sector – I assume they have to provide even more out of the box reports, roles etc – since SMB companies might not have enough budget to build it all ground up.

    (0) 
    1. Arun Varadarajan Post author
      I would rather differentiate the two approaches as how you want to conceptualize your warehouse or how you start building the same but then the end result is predominantly the same give the EDW a few years and then …. it is a mix of both where the EDW is a mix of Information store and Fact based models.
      (0) 
  3. Glen Leslie
    The success of SAP BW and then SAP NetWeaver BI lies, in part, due to the fact that you are not forced into a choice of one approach versus the other.   

    Early implementations of BW were accused of too much focus on dimensional modeling.  SAP expanded its capabilities with SAP BW 2.0B;  Bill Inmon took a closer look at SAP BW starting with that release, he liked what he saw. 

    Take a look at the evolving story in our emphasis on Layered Scalable Architecture (LSA).  Check out Teched 2008 presos for an intro to this approach…

    (0) 
    1. Arun Varadarajan Post author
      Glen,
      I do agree with you .. what I was trying to bring our is that datawarehouses have different perspectives in the non SAP BI world and it is worth looking into as a design philosophy.

      SAP BI of course gives that much wanted break from the world of conformed dimensions and strict dimensional modeling but then the blog was more to ask whether in an SAP BI world – whichever point you start from – do you end up with a mix of both approaches or is it still possible to realize a pure CIF or a dimensional EDW in SAP BI? more of trying to draw a parallel to the non SAP BI design methodologies….

      (0) 

Leave a Reply