IBM is one of the leading enterprise data cloud solutions for small and large-scale data warehousing and analytics, providing users with flexible access to all their data needs. Anjul Bjambri, VP of big data at IBM, shared her view with Dave Vellante, co-founder, Wikibon and Jeff Kelly, big data analyst, Wikibon, on evolution of big data from warehousing to analytics and history and Big Data story on the Cube at IBM Information on Demand 2012 event.

Bjambri said data warehouse architecture is evolving to an overall bigger ecosystem, its architecture extended not only to include structured data but also unstructured data, sandboxes and deeper predictive capabilities.  In the past, there was one monolithic structure in the form of enterprise data warehouse, but my, how things have changed. Today, the data warehouse remains a critical part of the total big data architecture, now enhanced to support new capability that most likely will live outside the walls of the data warehouse.

As more data and workloads are added to the system, warehouse complexity increases causing companies to focus more on analytics, and performance management. Apache Hadoop (and its MapReduce functionality) is a natural evolution of next generation of data warehousing, thanks to its scalability and flexibility.  Companies have started adopting Hadoop for strategic roles in their current warehousing architectures, such as extract/transform/load (ETL), data staging, and preprocessing of unstructured content.

Bjambri said IBM is helping companies to enhance and explore complex, huge amounts of data sets by introducing connectors (a gateway to connect big data with warehousing) to deal with leading-edge big data applications.

However, Bjambri goes on to note that getting full business intelligence (BI) value out of Big Data requires some best standards, practices and tools for enterprise data warehouses. Big Data tends to get big because new source of data are pouring in such as Web sites, social media, robotics, mobile devices, and sensors. Deciding which pieces are best and which of these are useful ranging from structured to semi-structured to unstructured for BI purposes can lead to better integration and quality of data.

See Bjambri’s full video below.