, , , ,

The problem with Data Warehousing is that at a superficially high level it is very easy to explain, and it’s quite amazing that this superficially high level is all that a lot of people need in order to do data warehousing their own way.

This is one of the reasons why many companies have been convinced that they could do Data Warehousing by simply building lots of “cost effective” independent Data Marts. Of course they were right, right?

Well, no. Companies found that the more independent data marts they had the more costly and complicated it became to maintain them and the more costly and complicated it was to add additional independent data marts into the mix. That and the absence of delivery on other promises – such as data integration and singular views of the organisation – ensured that this approach would rightly acquire bad press.

Why did this happen? Again we can go back to the beginning. People were told that data warehousing was about using a subject oriented, integrated, time variant and non-volatile data store in order to act as the source for all data mart development, and ultimately as the source of record for all strategic reporting. The trouble then came in two forms: people thought they could choose one or more characteristic and it would still deliver all of the benefits of data warehousing, and secondly, people were convinced by consultants that data warehousing could also be a collection of data marts.

As Bill Inmon put in 1998: “You can catch all the minnows in the ocean and stack them together and they still do not make a whale”.

It may be informative to take a step back. In the dark ages of data warehousing not all the technology vendors were equipped to cope with the burgeoning DW market, and some technologies just couldn’t hack it[1] when it came to building real Enterprise Data Warehouses. So, rather than miss out on all of the jolly cash that was finding its way into the fighting funds of DW initiatives, they decide to cash in on DW by muddying the waters and pulling the wool over people’s eyes, whilst effectively fleecing the gullible punters[2].

From the strategy of the small technical vendors came the idea that you could simply have data warehousing by building lots and lots of stovepipe data marts – hence Bill Inmon’s remark about the minnows and the whale. These marketing strategies were also supported by those who saw their own path to data warehouse glory passing through data silos rather than via well engineered end-to-end enterprise data warehousing. It may surprise some of the readers that the Kimball approach and the Inmon architecture were so different at one time, but they were. Yet over the years the Kimball camp has moved closer and closer to the Inmon approach, whilst at the same time seemingly[3] maintaining an aversion for anything that isn’t dimensional and explainable in terms of facts and measures. Although, if we are going to be frank and earnest (Frank in New Jersey, Earnest in Chicago), ‘conformed’ seems to be the most overused Information Management euphemism of recent times.

[1] Not appropriate or adequate.

[2] EDW customers.

[3] It’s my appreciation and I could be wrong.