Less than a month since the industry turned the page on its calendar, organizations already have their analytics priorities mostly figured out for the new year. A new poll of 250 data scientists, business intelligence specialists and other key members of the Hadoop ecosystem from Syncsort Inc. reveals that nearly 70 percent rank Apache Spark as the top number crunching framework on their agenda. Just over half of the respondents expressed a similar level of interest in MapReduce, the current default processing option.  

The findings reaffirm an earlier study from Databricks Inc. that revealed Hadoop adopters are slowly but surely abandoning the batch processing engine in favor of its speedier successor. But due to its head start, MapReduce is still the most widely-used of the two and continues to dictate the day-to-day priorities of the surveyed analytics professionals. A quarter of the respondents to Syncsort’s study ranked the complexity of the framework as a top challenge in their work, while 14.5 percent pointed to the difficulty of integrating the software with other applications.

Interoperability is only set to become a bigger topic as organizations shift more and more work from their data warehouses to Hadoop in order to take advantage of its lower operating costs. Syncsort found that 32 percent are using the platform to process traditional relational records while another 8 percent have adapted their implementations for archiving purposes. Not surprisingly, the rest of the respondents to its study stated that they’re focusing on modern use cases involving unstructured information.

Predictive analytics emerged as the most common workload category in that camp with a 26 percent share of the votes, followed by data discovery and visualization, which garnered a distant 12 percent. Processing machine-generated information was cited by just 8 percent as the reason why their organizations are using Hadoop, landing third place, but that figure is likely to increase significantly in 2016 as top vendors like IBM Corp. double down on the connected universe. 

Photo via sethink