While talking about a new year’s next cool thing or development is a thoroughly enjoyable ritual, discussing what will not change provides valuable lessons for technology adoption strategy and investment planning, and highlights potential future disruptions.
So what are the five things that will remain more or less the same in 2014 for big data analytics?
1. Hadoop will NOT REPLACE ETL: The nine-year old platform has achieved great traction, and its mindshare has significantly increased. Well-known analytics providers such as Cloudera, Hortonworks, and MapR have supported it for a couple of years, and even the big boys such as IBM and Pivotal have embraced it. However, Hadoop’s proponents are positioning it as a panacea for all the ills of big data. The antagonists are equally up to the task, denouncing it as one of the important, yet small, pieces of the puzzle. Most Hadoop proponents confuse ETL as an “activity,” rather than a “process.” The way in which ETL is performed in a Hadoop framework set-up may differ, but it does not make ETL redundant or replaceable.