Top 3 Analytics Trends for 2013
By Hannah Smalltree, Director of Product Marketing, ParAccel
This is a wildly inspiring, yet mildly overwhelming, time for analytics. It's inspiring because of the possibilities. What could you do if you could apply analytics across your entire datasphere with no constraints? What would you learn by connecting marketing response data, customer feedback, transactional data, predictive analysis, social sentiment, and product support records? How could you improve inventory management by looking across retail sales, marketing analytics, Web analytics, and supply chain data?
The possibilities are endless. However, analytics can also be overwhelming. What architecture will help you best bring together all of the data? How can you rationalize and consolidate the many tools, data silos, and shadow IT projects that characterize many analytic programs? How can you best arm all analytic stakeholders -- from casual users to data scientists -- with the capabilities they need to be analytically creative?
This is why the major trends in analytics in 2013 will focus on delivering the data and capabilities needed to become a truly analytic-driven enterprise.
2013 Analytics Trend #1: More organizations will move to cooperative processing architectures, such as Gartner's logical data warehouse
Data warehouses are straining from the surge of new data and complex analytic workload requirements. Traditional data warehousing wasn't designed for big data, real-time, social, and converged analytics, so the new expert-recommended architectures emphasize multiple, well-integrated systems working together to deliver a broad range of analytic capabilities. This basic concept is espoused by multiple experts and analyst firms, notably Gartner Inc. with its logical data warehouse, but also by Enterprise Management Associates, the 451 Group, Colin White, and other experts.
Although expert recommendations vary slightly, the underlying concepts of distributed, cooperative processing are similar. For example, an organization may have a traditional data warehouse for reports and business intelligence; Hadoop for big data ingest and processing; and an analytic platform for complex or ad hoc analytics. With the right architecture and tools, these systems can interact intelligently, sharing data and/or analytic results between components and minimizing data movement.
Integration is addressed agilely through data virtualization as well as more efficient, vendor-developed integrations between platforms (e.g., Hadoop connectors). The exact analytic components required are determined by business goals, but the underlying message is clear: a single system cannot cost-effectively handle the varied workloads of today's analytics. This infrastructure trend toward cooperative architectures lays the foundation for a flexible analytic environment.
2013 Analytics Trend #2: Converged analytics will become the new normal
Big data analysis in a silo often paints an incomplete picture. That is why organizations will move toward converged analytics in 2013, combining analytic results across domains (customer, supply chain, marketing, etc.) for more complete insights. Consider the retailer that wants better demand signaling, or the health provider aligning resources with patient requirements. It's not all about big data, but that has driven an uptick in the trend. For many big data projects, analytic value can often be greatly increased with additional context from other sources.
This long-held vision will be realized this year because of the technological innovations and the cooperative architectures described earlier. Converged analysis has been nearly impossible until now without serious custom development, but with the new cooperative processing architectures, organizations can have converged analytics capabilities without a major integration effort.
2013 Analytics Trend #3: Everyone will play with big data thanks to new big data applications
Big data has been decidedly DIY to date, embraced by tinkerers and architects. That's about to change. New big data applications offer prebuilt models and analytic functionality for specific business problems and data types. These are focused, single-purpose applications that can make a big impact to businesses -- such as analyzing sensor data from a particular type of machine or sifting through smart-grid data to identify potential failure spots in a power grid.
Big data apps represent an inflection point for big data. Now, companies hesitant to invest in early-stage experimentation have blueprints for leveraging specific big data types. Some offerings are better described as application platforms, designed for certain types of big data, with a framework and models on which companies can develop their own apps. Other industry-specific big data applications may be more likely to come in packaged form. Others may leverage analytic platforms to build their own big data applications. However implemented, the new big data apps promise to get companies to value creation, faster.
This trend will swing into full gear this year, with big data desires at an all-time high and the emergence of best practices and new applications emerging. Now, the many companies wondering "Do I need a Hadoop?(!)" can instead focus on what they can do with their big data. Although early big data apps primarily focus on marketing and customer data, new ones are predicted to emerge in other areas of common interest, including supply chain, health care, and retail.
Hannah Smalltree is the director of product marketing at ParAccel. You can contact the author at firstname.lastname@example.org.