Touching Lightly on a Killer Task
June 5, 2008
A colleague sent me a link to a white paper written by Bikram Sankar Das, the head of Tata Consultancy Services Business Intelligence and Performance Management practice in the UK and Ireland. Tata is a massive conglomerate known for outsourcing and buying aging automobile companies. Its consulting unit’s tag line is “Experience certainty”.
I enjoy reading white papers about business intelligence and content processing. Good papers give me useful anecdotes for my talks. Not-so-hot papers are less useful. “Business Intelligence and SOA: Making the Jump” tips toward the useful side. The wrap up struck me as the strongest section of the paper; to wit:
At the same time, there are still a number of challenges to be faced. One of the defining characteristics of BI-PM is its hunger for accurate information. As users become more and more accustomed to relying on analytical tools, their demand for new kinds of data capture and analysis increases. This leads to rapid database growth, and accelerating demand for storage capacity – pushing up costs and clashing with green IT policies.
The author makes an important point: users are going to grouse unless the systems deliver heterogeneous information properly parsed and sorted in a timely way. When systems don’t deliver what the marketers promise, users won’t use the system. Bad things happen when users get cranky and find other ways to get the information needed to do their jobs.
The weaker part of the paper is the hippity-hop over the problem of data transformation. As much as one-third of an information technology budget can be consumed fiddling with data so a fancy-Dan system can do its song and dance act. The author put my teeth on edge when he wrote:
Instead of a centralized information store, a federated approach can work well. With this approach, although the information is stored in a number of different databases, the databases themselves share a common protocol for information exchange through an ‘information bus’ – making it simple to compare and analyse data from different sources. To create a successful federated infrastructure, metadata must be carefully standardized across all systems, and a data/information governance model must be adopted across the entire organization. This can often necessitate a cultural change in the process of information creation, storage and consumption.
We’re talking big money, mucho time, and quite a bit of work to deliver standardized information to a business intelligence system. There simply is neither the money nor the programming resources to crunch through the large amounts of digital information. Users don’t know about these costs, nor do they care.
The hurdle next-generation business intelligence systems must get over is data transformation. A failure to explain the costs and complexities of this set of tasks fertilizes the ground for user revolt to take root.
Judge for yourself. You can download the essay here. The author minimizes what may be the most complicated work required for next-generation business intelligence.
Stephen Arnold, June 6, 2008
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
