It's All About the Data: The Internet of Things Raises Connectivity to a New Level

<< back Page 2 of 3 next >>

With this need to manage and draw insights from potentially vast sources of information, the practice of data management is being stretched well beyond the realm of databases and analytic platforms. IoT is an enterprisewide concern, requiring the direct involvement of a range of specialists and stakeholders. Companies will need technology “to develop product-service hybrids, enable the development of applications by third parties, and provide APIs for sharing data and control the channel for delivering services to customers,” Banerjee said. “Asset owners and operators will use these platforms to operate equipment and applications, to deliver and analyze data, to link and control processes, and to connect with other companies in their ecosystems.”

Data and IT leaders are the best equipped to take the IoT-enabled enterprise to the next level. While many skills will come to the fore, enterprises will need the ability to identify, secure, and integrate data from a wide array of sources. “Each type of thing on the Internet of Things will have its own data streams with its own data formats, and its own APIs,” said Gorbet. “Accommodating new data from new things and making it available in a way that can be used alongside other data from existing things is an incredibly challenging data integration problem.”

IoT Has a Different Take on Data

The bottom line is that “IoT data has to be analyzed to be useful,” Scott Sweet, VP and head of Capgemini’s North America Business Information Management practice, pointed out. “Internet of Things” may be a static term, but what enterprises actually need to do is pursue “Analytics of Things,” Sweet said. Any and all devices, he continued, “can have disconnected and connected analytics and each scenario has differing architectures, storage needs and processing. AoT’s success will be contingent upon engineers, business users, and others learning how to extract, locate and analyze this data to derive valuable insights to inform business decision making.”

IoT data must be loaded and queried simultaneously to avoid missing out on actionable insights.

IoT differs from traditional data management in another way as well, employing streaming data, versus the typical approach of data at rest, said Scott Jarr, co-founder and chief strategy officer for VoltDB. “This data is key to improving two things: the customer experience and revenue, and operations.”

This requires that IoT-based data be approached in an entirely different manner, as associated applications “generate streams of sensor data from many sources, and this data must be correlated and analyzed to identify perishable business opportunities,” said William Bain, CEO and founder of ScaleOut Software. “While BI excels at finding patterns in historical datasets, it cannot respond to the low-latency requirements imposed by fast-changing sensor data. Analyzing real-time data streams generated in IoT use cases requires tracking fast-changing data with extremely low latency, analyzing it quickly and scaling to meet the needs of production deployments.” The goal is to be able to “detect patterns, trends, and opportunities on a second-by-second basis,” he added.

Enterprises may need to take a closer look at new forms of computing—such as cloud—to handle the demands of IoT. “IoT data is time-sensitive, so speed matters in a highly competitive environment,” noted Zubin Dowlaty, head of innovation and development for Mu Sigma. “The key is to get the data, process it fast with smart algorithms, pass it to the decision science teams for more analysis and then to business. This cycle time will determine how much value businesses realize, hence the more integrated the design, the better it is.” Such information would also need to be context-sensitive, “meaning integration with existing data would drive incremental value to existing insights,” he added. “For example, a demand forecast or market mix can exponentially benefit as data gaps are closed from IoT.”

The first step in this process, data collection and integration, remains a challenge “because there is currently a lack of common—vendor and platform-agnostic—connectivity standards in the industry,” said Healey. “It’s critically important to be able to consume or read many diverse data sources, streamlining and accelerating data integration. In addition, IoT data must be able to be loaded and queried simultaneously to avoid missing out on real-time, immediately actionable insights. By the time the data is loaded into a database and analyzed, an organization may have missed a critical chance to respond or act upon a small window of opportunity with a connected product.”


<< back Page 2 of 3 next >>

Related Articles

HP has introduced a new offering called HP Haven Predictive Analytics to accelerate and operationalize large-scale machine learning and statistical analysis.

Posted February 18, 2015