Ask any business intelligence or analytics vendor to give a few customer examples of how their product is being used with data stored in Hadoop, and you’re likely to get some blank stares. That’s not a slam on the BI vendors. It’s partially an indicator of the state of maturity in the Hadoop market, and partially a reflection of just how hard BI is, regardless of the data source.
Organizations that have deployed and standardized on specific BI tools can’t just point those tools at a schema-less Hadoop cluster and start analyzing. Data needs to be aggregated or indexed. And it took years for SQL-based relational and multidimensional BI technologies to reach a state of maturity where they could deliver enterprise-class scalability, functionality and performance. It’s going to take some time before a new generation of analytic tools will meet the full range of requirements for Hadoop-based analytics.
Forrester analyst Boris Evelson wrote a great blog last September illustrating the issue. He noted that BI vendors are quickly responding to customer interest in Hadoop by announcing Hadoop integration. But he goes on to caution that when it comes to BI tools, not all approaches to Hadoop integration are created equal. In fact, he offers twelve questions customers should ask their BI vendors in order to get clarity on exactly what the vendor’s product can or cannot do.
The good news is that everybody is working on the problem. More to come! Also, be sure to checkout my first blog, Differentiation Across The Apache Hadoop Distribution Vendor Landscape, and second blog, Will Hadoop Replace Replace the Data Warehouse?, of this series.