As people who have read my blogs over the last few years will know I have always been a strong proponent of a data centric view of the integration world (covered in more detail here for instance). However when we discuss data integration we can fixate on the high-end problems and forget where the greatest challenges and risks associated with data integration can lie: In the excel spreadsheets we all create and use in our everyday business lives.
The reality is that in financial services and elsewhere organizations rely on spreadsheets as the decision tool of choice across a whole range of critical functions from risk modelling to pricing to monitoring and financial control. Why spreadsheets are used is pretty obvious: they are quick and easy to put together (the original opportunistic application), and do not require heavy IT involvement (typically reducing the time, cost and the sometimes perceived over-head associated with getting IT involved). However, many of the drawbacks are also pretty obvious and are made worse when the spreadsheet is created outside of IT’s governance mechanisms (these are outlined in greater detail by Theresa Clarke in an excellent article in gtnews – a free registration is required):
-They get complex quickly (undermining the original simplicity and cost benefits),
-They are not properly tested (as they are created on the fly and often only tested for the original scenario – they are also hard to test)
-They often evolve and are distributed outside of the usual IT governance structures which means multiple versions with increasing likelihood of further errors or misuse.
-They are prone to mistakes (a survey of spreadsheets used in enterprises quoted by Theresa states 90% are materially incorrect)
Any finally, the data may well be out of date or simply wrong with the obvious implications on decisions (and risk measurement and compliance!!). The reason the data will be out of date is obvious: in many cases there is no easy mechanism to ensure it is up to date and there is no easy mechanism to flag the problem when it is.
While the issue is recognised by many organisations (and auditors), it is far from solved. The choice of stopping people using spreadsheets is not an option and bringing the spreadsheet into an effective data integration strategy is very hard across even moderately complex organizations.
Ronan













I agree totally that data integration is often forgotten - and just in case any of your readers are interested, the Integration Consortium is hosting a free webinar on Thursday 18th September (http://www.integrationconsortium.org/events/event_details.asp?id=28947 for details), entitled "Data - SOA's Last Mile", looking at how data integration is frequently neglected in service oriented archtiecture deployments today, with horrendous results!