HomeData warehouseData Warehouse Architecture – Kimball and Inmon methodologies

Comments

Data Warehouse Architecture – Kimball and Inmon methodologies — 11 Comments

  1. James, You seem to be conflating Architecture with Methodology. In my experience there’s nothing about an integrated, normalized data warehouse (Inmon CIF architecture) that means it will take longer to deliver results or cost more up-front. It may be argued that *waterfall* methodologies can take longer and cost more up-front, but that is orthogonal to what *architecture* you choose. Agile, iterative approaches are surely very popular with BI projects these days and both Inmon and Kimball architectures are often implemented using an agile approach.

    An important advantage of a normalized data warehouse with dependent marts is that it supports the iterative, agile approach better than coupling each denormalized mart (a’la Kimball) directly to data sources. With a normalized warehouse it is typically easier to add new data sources and evolve the warehouse model because it is less tightly coupled to any one set of reporting requirements and because there are fewer moving parts (transformation layer) on the upstream side of the warehouse. The downstream side (between warehouse and marts) is where decision-support business logic goes and that is simplified too because it only has to consume data already validated and integrated into the data warehouse.

    • I agree with the advantage D points out. Having integrated the data into the normalized data warehouse also leads to much more consistency across the various data marts in terms of their data models and vocabulary. This is certainly the approach I prefer.

      Also, a small correction regarding terminology. In a dimensional approach, data is partitioned into either “measures” or “dimensions“. The “fact” is the measure within the context provided by the dimensions, i.e. each row of a fact table.

  2. Pingback:Why you need Business Intelligence | James Serra's Blog

  3. Pingback:Why You Need a Data Warehouse | James Serra's Blog

  4. Kimball is NOT a bottom up methodology (Inmon calls it that but Kimball disputes). Inmon offers no methodolgy for data marts. If you use Kimballs (atomic) data mart methodology with Inmons CIF you end up with 2 full copies of source transactions. Imon is subject oriented meaning all business processes for each subject (for example client) need to be modelled before the EDW can be a single version of the truth. This takes a LONG time.

  5. Pingback:An Architecture For Data Quality Kimball | Architecture Fan

  6. Pingback:Corporate Information Factory Vs Bus Architecture | Easy Architecture Fan

Leave a Reply

Your email address will not be published. Required fields are marked *

HTML tags allowed in your comment: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>