Unit 8, Elms industrial Estate
Church Road, Harold Wood, Essex, RM3 0JU, UK
Tel: +44 (0)20 8817 5296
7100 Six forks Road, Suite 135, Raleigh, NC 27615, USA
Tel: +1 919 747 9400
2002 - 2014 Copyright Triumph Consultancy Services. All rights reserved.
Data modeling - often overlooked, always essential
8th Aug 2014
Duncan Hall, CEO Triumph
I never fail to be amazed at the number of companies that seem to jump right into integration and data warehousing projects without actually having defined their operating model, data model and business processes first. So why is this?
Is it the desire to start developing on day one - pressure from sponsors to show some tangible IP as soon as possible?
Is it that the people developing integrations, reports an warehouses aren't the ones that have to maintain them and so don't understand the need to build traceability?
Is it a capability gap - in an environment where clinical systems are often poorly or not documented at all, how do you even start developing that data model?
I suspect it is a bit of all of these, but if we can't get developers, BAs and sponsors to understand the importance of data modelling, we're never going to drive this critical change.
So let me paint you a vision......you are a report developer and you've been asked to add a new field to a report to meet a specific regulatory need. In our ideal environment you'd go straight to the data model. It would describe the data entity you have been asked to add, when it is mastered, by who and in which system. The model would then show you which data warehouse / data mart it resides in. Bingo! You now know you have the master data, the format it should be in and where to get it. All report development should be like this, but with the lack of data modelling so prevalent in our industry the reality is often far removed.
Back to Blog page