Like any new class of technologies, all things related to Big Data in the enterprise can be controversial. In particular, proponents of SQL database technologies don't always appreciate the insinuation that SQL can't handle Big Data types. The actual issue is the cost analyzing Big Data types on an SQL platform versus the cost of using something like Hadoop and MapReduce on standard Intel servers.
IBM wants these two camps to better appreciate the complementary nature of Big Data and SQL technologies. To that end, the company today announced it is going to provide free access to 3,800 Big Data bootcamps across 38 IBM Innovation Centers and online via DB2 University.com.
Anjul Bhambhri, IBM vice president of Big Data and streams says that from an IBM perspective, the company sees SQL and technologies such as Hadoop as part of one continuous data management spectrum. Hadoop allows IT organizations to cost effectively analyze not just 30 days of data, but as much as seven years of data. SQL databases will still be required to analyze subsets of the data at a high speed, so Bhambhri says the two technologies are more complementary than competitive.
As part of that effort, IBM generally expects customers to analyze data increasingly in real time using data warehouses running on zEnterprise mainframes and Power Series servers such as the new Watson supercomputer, while making use of xSeries servers to host Hadoop applications. That approach, says Bhambhri, allows IT organizations to get the best of SQL and Hadoop without getting caught up in the distractions over whether the use of SQL is a good thing anymore.
With the advent of Hadoop there will be a definite shift in how data is analyzed in the enterprise. Business executives want to analyze more data than ever so they can detect patterns over a longer period of time. But Bhambhri says that requirement shouldn't be seen as a threat to SQL, but rather something that makes SQL more useful by relying on Hadoop to sort through all the "noise" in the data before drilling down on it using SQL.
In other words, Hadoop should be seen as a way to extend investments in SQL rather than a threat to a technology that will be with us for a very long time to come.