Next generation BI: potential solutions

Written By:
Content Copyright © 2006 Bloor. All Rights Reserved.

In the first two articles in this series I have outlined some of the requirements that are needed for a next generation BI solution, both from a user and a technical perspective. They raise two supplementary questions: first, is this feasible and second, when can we expect the likes of Business Objects and Cognos to deliver on this vision?

Let me take the second question first and answer it by saying that I don’t think we will. The traditional BI vendors are wedded to a traditional concept of BI and OLAP. It is appropriate to recall the joke about the Irishman who asked the way to Dublin and was told that “if I was going to Dublin I wouldn’t start from here”. So, if we are going to see a next generation BI solution, as outlined, we will have to look outside the usual suspects.

As to the question as to whether this vision is feasible, I believe it is. As far as visualisation is concerned Inflection Point is doing some clever stuff with user access to data, while FYI Corporation has developed new ways of presenting information. Tableau and Spotfire also have innovative visualisation capabilities.

As far as the abstraction of relationships is concerned, this is precisely what QlikTech does, combined with discovery when you load the data and the ability to calculate aggregates on the fly (in memory). Information Edge, on the other hand, has technology that allows you to support a single massive cube rather than multiple smaller ones, so that all aggregates can be pre-calculated, without any degradation of performance. My guess would be that any vendor planning to build all of this would probably want to incorporate at least a part of both of these elements.

In addition to QlikTech and Information Edge I should also mention Exeros and Sypherlink. These companies have mapping tools that will automatically discover the relationships that exist within and across data sources. At present, neither of these vendors has an adapter to OLAP sources but it could be used in conjunction with one of various products that present OLAP data in relational format. Potentially, either of these companies could not just form part of the relationship management aspect of a next generation BI solution but also help to resolve migration issues and enable the reuse of existing OLAP technology.

As far as text is concerned, we are really only at the embedded search stage at present though combining this with query capability should not be complex. More sophisticated text querying capabilities may be some way off, however, though IBM’s OmniFind could be used as the basis for such a solution. Going further, we can expect closer integration between relational and XML query capability in the near future, at least within the environment of IBM’s DB2 Data Warehouse Edition, where a logical outcome of the recent introduction of DB2 Viper would be this sort of capability.

And finally there’s the question of joint real-time and conventional BI capabilities. The issue here is really which of the slew of event processing vendors to partner with or acquire, as putting the relevant front-end on should not be complex. It would make sense if the target was SQL-based rather than rules-based, since this would make integration easier, but otherwise there are a number of vendors to choose from, including StreamBase, Leanway (which is based, unlike the other vendors, on a relational model) and Coral8 as general-purpose vendors, as well as SeeWhy and Syndera as vendors already focusing on the BI space. In the case of the latter, this company’s solution is process aware so this also fits one of the requirements detailed in a previous article.

To summarise then, the elements of a solution that would offer a truly next generation BI system are in place. But they are scattered and in different places. Who will bring them together? Watch this space.