An interesting tale, demonstrating the importance and longevity of data modelling, was recounted recently by an ex-colleague:
A few weeks ago I attended the retirement party of a long-serving employee of a company I helped develop software for thirty years ago. On arrival at the company's offices I could not help notice significant changes in the office environment (the trusty kettle had long been consigned to the skip) and, of course, in technology. As you might expect, the people had also changed, almost out of all recognition. In fact, none of the smartly dressed reception staff were actually alive when we implemented the original software solution.
I had a long chat with the former managing director. He told me that four generations of management had passed through the company since my previous visit three decades ago. But as he described the current operation, most of it seemed very familiar, very reminiscent of the data model that I developed for that software solution all those years ago. Vessels still undertook voyages, voyages still had consignments of timber, bound for receivers and customers in the United Kingdom. The scope and technology of the software had changed radically, with customers now accessing and using the system through the internet, but the underlying data structures remained very similar and very familiar.
All of which made me recall the assertion made many years ago that underlying data structures remain far more stable than the processes and people that use that data. Managers and users come and go, but the fundamental structures of the data that the company deals with often remains very much the same. Indeed the former managing director amused himself for a while telling me the story of his successor who specified a system based on 'some wacky ideas he had been fed by a consultancy'. The system, whilst genuinely reflecting the main user's requirements, was a complete disaster, and the ex-managing director was asked to take up the reins again during an interim period until a more suitable successor was found. The developers used a 'user-centric' approach he explained with glee, little knowing that the user was incompetent (he used much more descriptive words which I cannot repeat here).
So, I like to think that data modelling remains fundamental to successful information systems development, providing a sound foundation for business information. I would like to think that a business analyst today would also construct a data model or class model of the area under consideration. In doing so they would document the business objects and business rules using a technique that is rigorous and has an economy that sets it apart from many other techniques and concepts used within our world.
However, I do worry that too many projects nowadays concern themselves with process rather than content, and with slogans rather than techniques. Data modelling is a well-proven technique fundamental for successful business analysis. My data model for the shipping company, and the data structures that underpin it, still remains the basis for the design of the current software solution. The data model is hand-drawn (modelling tools were in their infancy in 1983) and has been heavily annotated by subsequent analysts. As the former managing director said, it is more battered than most of the ships that pull into the quay, but it has outlasted four IT directors, the quayside pub where we had our review meetings, the Vauxhall Chevette I used to drive to get there and the two companies that, back then, were the main customers of the shipping company.
Now; that's the power of data modelling!
The Business Alchemist