In 2013, more companies will begin developing Hadoop-based applications for their most critical Business Intelligence (BI) tasks. That’s one of the predictions in Forrester Research’s crystal ball for BI in the new year.

The BI predictions, made by Forrester analyst Boris Evelson on his blog, include two new ones and eight updates to last year’s. Evelson said the predictions for 2012 “did a pretty good job,” so he revised and re-used the ones he said proved to be most accurate, using Forrester research from the past year.

Bottom-Up Vs. Top-Down

Evelson predicts that more companies will begin to “invest in custom coding that addresses current Hadoop limitations” in order to achieve cost savings by dropping reliance on commercial software licenses and proprietary hardware.

Hadoop is an open source framework that allows applications to utilize big data distributed over many sources, but, in Evelson’s words, its functionality is “immature at best” in such areas as transactional controls, multiphase commits and rollbacks of incomplete transactions.

Evelson’s other new-for-2013 prediction is that, in the struggle between bottom-up and top-down approaches to building BI systems, a hybrid approach might become popular. The hybrid way emphasizes top-down’s waterfall approach with business requirements for a specific use case, followed by implementation, but, in Evelson’s articulation, adds a “tax,” or what he implies is an additional budget set-aside, to fund bottom-up, figure-out-needs-as-you-go-along evolutionary approach to strategic architecture development, such as a data warehouse.

The revived 2012 batch includes the prediction that “the best tool for each BI job trumps IT standards.” In other words, an over-reliance on enterprise standards is not “flexible or agile enough” to accommodate the many BI tools that have been emerging, and will continue to flourish.

Self-Service, Data Discovery

End users will increasingly demand more BI control, Evelson re-predicted, and self-service features of BI tools, like semantic layers and search capabilities, “will become increasingly critical.” He noted that many BI vendors are releasing products with functionality that allows user creation of BI apps.

Mobile BI apps will continue to move beyond “nice to have” and become must-haves, according to another Evelson prediction, and cloud-based implementations will continue to gain ground over on-premise ones. Database management systems (DBMSes) that have specifically been architected for BI using agile approaches are emerging and will continue to do so, he forecast. He noted that BI-specific DBMSes started going mainstream this year, and, within the next two years, projects that over 20 percent of all BI applications with utilize this technology.

Other revised predictions include big data moving out of silos and into enterprise IT, data discovery and exploration becoming a regular part of BI suites, and BI tools integrating with email, search, portals, social communities and other parts of the information workplace.