A new paradigm of Business Intelligence
The current definition of Business Intelligence (BI) puts the focus on data utilization. However, the usage of BI solutions in many companies has become unmanageable and, in numerous cases, developments have gone in the wrong direction.
Many companies are trying to simplify legacy complexity and merge everything into a central system of modular components. Producers have recognized this and now provide comprehensive BI solutions that fulfil as many requirements as possible: from reporting functions through to sophisticated interactive data mining tools.
Companies are well advised to take a holistic perspective on tomorrow’s BI world and place the focus on medium- to long-term planning.
However, two problems have not yet been adequately resolved: firstly, two different datasets are being used and, secondly, there is insufficient performance to handle more complex business questions. Newly developed in-memory databases suitable for deployment in enterprises are a very promising solution. These systems have the potential to merge operational and analytical data back into a uniform dataset, thus ensuring high system performance and flexibility.
The article from which this abstract is drawn examines what’s new about in-memory databases. It argues that, while acknowledging performance as the most-cited argument for in-memory systems, flexibility should have an equivalent level of importance.
While it is possible to accelerate access to the data in the data warehouse, extraction, transformation and loading processes are still subject to the familiar restrictions. Similarly, there have, so far, been no changes in the flexibility of the data analysis. Until this happens, many companies will never achieve added value.
The authors also provide some thoughts when considering the transition to in-memory solutions. They conclude that, on the one hand, many businesses have invested a lot of money in building an extensive BI landscape with clear segregation between operational and strategic data. In this respect, it may seem like a mockery to now propagate a return to a joint data model. On the other hand, the high level of diversification of BI solutions has revealed various inefficiencies and developments in the wrong direction. This makes it all the more important to properly plan the future BI architecture with a sensible compromise between evolution and revolution.
The complete article was written by:
Read the full article457.34 kB