Big Data is not a new phenomenon and has been around for many years, noted Jim Goodnight, CEO at SAS, who spoke recently at the SAS High Performance Analytics Conference in Hong Kong.
He described how companies have always been aware of the massive amount of data they accumulate and due to constraints have had to analyze this data in subsets. This analysis has typically formed trends and insight to be then applied to larger data sets and scenarios. But this takes time and resource.
"With in-memory technology, new analytics tools and hardware developments, we can now analyze much bigger data sets and this can all occur much quicker than before," said Goodnight.
To CIOs, who are in fact the "information" officers of the business, this should be setting their world alight at the possibilities now available to them. But Goodnight observes that many CIOs are still burdened with trying to manage and maintain the infrastructure and have little time to support the information needs of business leaders. "Unfortunately when systems go down or when data is lost, it's the CIO that gets screamed at," he said. "It's a tough position for the CIO to be in but I think technologies exist today to help CIOs better manage data and provide improved access to that data for the business."
Information gap
Increasingly there seems to be an information gap in terms of what companies want to leverage for decision-making versus having that information available for analysis. "Companies today still struggle to effectively target customers with truly relevant offerings, despite having more customer data than ever before," said Goodnight. "Any company should be able to consolidate all its data and build a complete profile of a customer but often the data is not consolidated or available to the right people within the business."
Businesses everywhere face similar challenges of trying to bring together vast amounts of data that are often stored in siloed systems. "It's really the CIO's role today to have this data all staged in a way that business users can access easily and quickly," said Goodnight.
Banks are great examples of companies that have multiple operations and massive data repositories. Each operation or business unit often conducts processes such as risk analysis for purposes ranging from customer loans risk, trading transactions risk to anti-money laundering and fraud.
Single view
Goodnight noted that these often cover the same data sets, which means a lot of overlapping work and analysis. "CIOs have clear visibility into this which allows them to streamline and consolidate these tasks," he said. This presents potential efficiency gains, faster processing and joint insight which can be shared among related business units.
Ultimately how this data is analyzed, modeled and used by the different groups is down to the business users but CIOs and IT should take hold of the acquiring and access to the data to simplify the analysis process for the users. "So many companies are still working to get a single view of their data and it's the CIO's job to pull all this together," stressed Goodnight.
He also added that the industry still hangs on to the dated belief that there has to be a single repository for all data. "Data can and will reside in many different and separate systems," he said. "Let's leave the data there and focus on getting complete and quick access to this data as and when users need it."
SAS has developed tools to aid this process as well as deliver on the promise of providing users with easier access to data and giving more users the ability to leverage analytics. "Companies today want to give their staff the ability to manipulate and dissect data without the need for IT," Goodnight said.
IT organizations should not fear this prospect but find ways to enable it so they can free up further time to work on areas that create more value for the business.
Michael Barnes, vice president & research director at Forrester also noted the rising demand by information workers for more BI access and control.
He added that in Asia Pacific there is always an adoption lag for technologies particularly in BI and analytics. "Culturally, Asia has a greater tendency to "tinker" with applications and information and the typical lag will make this trend even more pronounced," said Barnes.
Asian activity on the rise
Because of the previous high cost of BI and analytics tools, users were limited to a small number of power users within organizations. But Barnes has noted over the last 12-18 months, more users are gaining access to BI tools and capabilities. "And this is driving the desire to want to learn, investigate, model and play," he added.
Forrester research indicates that BI maturity varies widely across Asia and Barnes believes cultural differences will continue to have a major impact on BI spending, adoption and usage patterns. He observes in general AP organizations have a higher demand for data exploration rather than analysis. However, he expects that lower cost tools and the exuberance of both end users and IT staff will increase penetration of access to BI capabilities.
Barnes also noted that in-memory analytics is gaining traction across the region, but is primarily growing in large enterprises with mature IT infrastructure and expertise. "In-memory is being adopted within areas where real-time or near real-time analysis of very large volumes of data is a primary objective, such as risk management, diagnostics, customer sentiment analysis," Barnes added.
Risky business
One area where SAS has helped a customer make real performance gains is a risk scenario at UOB Bank in Singapore. Its chief risk officer observed that it took them 18 hours to calculate risk for trading operations on any given day. This meant that the resulting data would have no bearing on the current operations, so he asked SAS if this could be accelerated.
SAS reviewed this and through new hardware and the latest in-memory analytics tools it was able to reduce this risk calculation down to 15 seconds. Goodnight noted this task from a technological standpoint was not complex as risk calculations are not new or much different from the past. The difference today is the sheer amount of data and processing required in producing the resulting risk models.
"Computational intensity is the challenge, we estimated the system has to speed up from processing two-billion instructions per second into 200 trillion instructions per second," Goodnight said.
