There's a lot of talk around big data, but actually getting the ball rolling on how to deal with it is one of the biggest hurdles that enterprises are facing, according to Cisco Consulting Systems Architect Adam Radford.
The IT architecture required for big data differs a lot from the ones used for traditional business analytics and application delivery, which consists or a compute tier, a storage tier, and networking tied together in a centralised way.
For big data, it's a scale-out model, rather than a scale-up one — there would be a multiple clusters of compute and storage units, Radford said.
While businesses may think that this would require a significant investment in IT, he said that the cost of entry for dealing with big data is actually quite low.
"The cost of hardware, at least to get started, it's not massive," Radford told ZDNet Australia. "A lot of the big data software is public domain, so there's not a lot of licensing cost, and the hardware is very much a standard x86 compute platform."
Regardless, organisations are still grappling with changes in the method of handling an influx of data, he said, but it's important to take advantage of big data, as it provides much more insight into things like internal operations and customer sentiments.
However, big data and traditional analytics not only require different IT architectures, but approaches in retrieving information as well, Radford said.
"My number one advice to enterprises is to make a start in a very small environment, to understand the types of thinking and processes that apply to big data," he said. "It's very different from traditional business intelligence and data warehousing questions you have asked in the past."
"This is very much an interactive, experimental, and agile approach versus the standard-based questions we've had previously."
Big data also requires different skill sets. Radford said that skills needed in traditional data analysis have been commoditised over time, and not something average users have to understand anymore.
"It's so drag and drop," he said. "In contrast, for big data, it's really a very different type of skill set used, because it's in a very dynamic environment.
"The type of data we are looking at is inherently unstructured."