What the heck is Business Intelligence anyway?
Posted at 8/21/2008 11:59:00 AM
In The Business Intelligence Model of Conflicting Interests on the B-Eye Network, Karien Verhagen states, “There are three parties involved in business intelligence, each of them having its own interpretation of good management information; however, their interests are quite different and even directly oppose one another.”She goes on to describe the painfully common scenario where operations folks use Excel, the IT folks guard their ivory tower data warehouse, and the executives find none of it particularly useful for either status reporting or strategic planning. Meanwhile, massive amounts of money are being spent on business intelligence data warehouse activities. Her proposed solution is “a combined initiative of all parties involved”.
If the solution is so simple, why is it that integrated solutions are the exception rather than the de facto standard in business intelligence? Conflicting requirements are the universal obstacle that needs to be overcome to make this happen.
While everyone agrees that a single source of data is a good thing, making that happen has been virtually impossible. Operations people want access to last-minute data and they need answers to different questions every day. IT people want a dependable, well-managed data store with a single source of metadata. Executives want quick and simple answers to status questions and powerful discovery and analytical capabilities for strategic planning.
The only realistic solution for the data management side of this problem is to have a data storage system that has solid administration features, a virtual physical structure, immediate access to newly loaded data and unrestricted flexibility in accessing the contents. Of course, front end BI tools by themselves can do nothing to help with this problem.
Relational databases do a reasonable job for the IT people but can’t hope to provide the level of functionality that is needed by most users. Column databases get closer to meeting these requirements but they are still inadequate, especially in areas such as flexibility of access. In-memory databases meet most of the speed and access requirements but can’t realistically scale up to anything near an enterprise data store.
The only solution left is to use something new and different like a correlation database for the storage system. Today, iLuminate is the only correlation database available but, considering the tremendous advantages this structure provides, many companies will soon be working on their own version of the next-generation database.
Eventually every organization with a database will be using a storage technology that eliminates the constraints of today’s relational, column-based and in-memory designs. Many are likely to implement a correlation database model to provide a single integrated BI solution.