With the first mainstream deployments of Hadoop, IT will be challenged to demonstrate value and measure a return on investment of big data projects.
Predict: away from the hype of data science, big data platforms will be used in 2013 to offload “mundane” tasks that can benefit from extreme and inexpensive scalability.
Tremendous hype has grown about data science and the benefits that can be derived from mining social data, log files, behavioral analysis or other enterprise “dark data”. While several of these use cases are promising in the mid to long term, they remain experimental and will require new models to be developed.
In the short term, a more promising use of big data platforms is to offload some of the “mundane” tasks that are traditionally done by existing platforms, and could benefit from the scalability of Hadoop.
Such “mundane” tasks that will be offloaded to Hadoop in 2013 include for example:
- Offloading of ETL to the Hadoop engine via MapReduce code generation
- Online archiving of massive volumes of historical records that need to remain searchable and queryable