Processing big data does not have to take a lot of time: SAS

Processing big data does not have to take a lot of time: SAS

Vendor sheds insights into how it is cutting down processing time with big data

If you are going to mine your unstructured data, then you better make sure that the solution is up for the task.

SAS senior vice-president and chief marketing officer, Jim Davis, offered this recommendation during a keynote at the company’s High Performance Analytics solution event in Sydney.

When speaking about big data, Davis broke it down into the three pillars of data, analytics, and platforms.

“The data you have can be either structured or unstructured, and it requires information management,” he said.

“This is what we typically refer to in the industry as ‘big data.’”

In simple terms, Davis sees big data as data that exceeds the processing capacity of conventional database systems, which puts a burden on the three V’s of volume, velocity, variety.

“The only reason we talk about big data is because it forms an obstacle,” he said.

“We can’t get to it and make sense of it, let alone incorporating it into our day-to-day activities.”

In the area of analytics, Davis sees the analysis of data being useful in predicting outcomes and judging quantifiable benefits.

“The traditional approach of analytics has been reactive, making use of alerts, On Line Analytical Processing, Ad hoc reports, and standard reports,” he said.

“The proactive approach focuses on optimisation, predictive modeling, forecasting, and statistical analysis.

Davis says that SAS’ High Performance Analytics solution is the culmination of the company’s work to enable “people to look at future risks.”

When it comes to platforms, there is the Cloud to consider as well as mobile.

According to Davis, high performance analytics are only feasible with in-memory architecture.

“Our competitors are SQL based and not in-memory,” he said.

“You can’t do high performance analytics via SQL.”

When addressing what analytics can be used for, Davis picked out descriptive statistics, predictive analytics, model development, text mining, forecasting, and optimisation as some of the top reasons.

“Analytic applications include retail planning, revenue and marketing optimisation, stress testing, liquidity risk management, and fraud detection,” he said.

In Davis’ opinion, mobile computing should be “more than just reporting” and “an extension of the overall process.”

“In addition to a dramatic reduction in processing time, High Performance Analytics has been designed to provide you with a visual exploration of Big Data,” he said.

SAS senior research statistician developer, Oliver Schabenberger, was also on hand to give an insight into the technological side to High Performance Analytics.

“Parallelisation was the key to success,” he said.

“The architecture is scalable to run thousands of processes simultaneously.”

According to Schabenberger, data movement is what has typically been the “performance killer” in big data analytics, but is taken out of the picture by putting the data into memory and processing with powerful hardware consisting of numerous CPU cores and lots of RAM.

“We aimed to create stability and not change the way you interact with software,” h esaid.

“At the same time, we added high performance to further enable it.”

Follow Us

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.
Show Comments