IIoT applications must be able to handle large amounts of continuous data from business facilities. Find out why IIoT apps need to ensure top performance.
The Internet of Things consists of two types: consumer and industrial. Toaster ovens, remotely controlled lighting and automated thermostats that intermittently transmit Wi-Fi status messages typify the consumer variety. Industrial IoT (IIoT), also called Industrial Internet, is very different — a torrent of continuous data packets from sensors in jet engines, oil refineries, and factory floors that is aggregated, analyzed, reported on, and managed over the cloud from afar. Developers building applications for each must understand these differences and design accordingly to ensure top performance.
According to a November 2014 forecast published by MarketsandMarkets, the annual worldwide Industrial IoT market, already a healthy $181.3 billion in 2013, is projected to exceed $319 billion in 2020, a compound annual growth rate (CAGR) of more than 8%. Another report, published in August 2015, that divides IIoT into four distinct segments — manufacturing, energy and utilities, automotive and transportation, and healthcare — is even more bullish, forecasting a CAGR to 2019 of more than 26%.
Tom Hendersoncloud researcher, Extreme Labs
While the opportunity for developers in IIoT is seemingly limitless, the development tools for harnessing this expansion are themselves in a continual state of growth and development.
A driving force behind IIoT is its ability to leverage sensor data and long-standing manufacturing and facility-automation technologies that use vastly different data protocols. Combined with real-time analytics, it becomes possible to generate exception-based status reporting as needed: Thousands of sensor readings taken seconds apart that a jet engine is operating within normal specifications are superfluous; it’s only those that track an unexpected rise in operating temperature that need to be reported and acted upon.
The challenge, said Christian Renaud, research director for IoT at 451 Research, is creating cloud applications that are robust enough to handle the torrent of data without slowing down. “You’re dealing with zillions of miniscule packets containing telemetry and diagnostic data,” he said. “You can’t have an application start skipping packets because it can’t handle the pace.”
General Electric (GE) has put that torrent to good use, developing Predix, an IIoT analytics and big data platform that examines sensor telemetry from industrial machinery to minimize downtime. As the world’s largest maker of jet engines for commercial airliners, GE’s aviation division used Predix to analyze 340 terabytes of data from 3.4 million flights to improve asset performance and minimize disruptions.
With such critical status reporting a key component of IIoT, another major difference between it and consumer IoT is dealing with disconnected operation. Miss a reading from the thermostat in the living room and the consequences are likely nil. But, start missing temperature readings in an industrial furnace and the outcome can be catastrophic.
Consumer IoT products generally work out of the box with a wide variety of communications devices, providing setup simplicity for users. Having recently switched his own home lighting to Philips Hue, Pelegri-Llopart said he benefits from reduced power consumption, the convenience to control lights remotely and being able to switch between different room settings. “What I don’t want to do is mess with gateways.” That, he said, is the job of the applications developers.
IIoT, conversely, can have differing standards that complicate deployment and management. “The cost of failure is very high, so the infrastructure has to be very reliable,” Pelegri-Llopart said.
For Tom Henderson, longtime network and cloud researcher at Extreme Labs, this real-time requirement is the key differentiator. Industrial systems are more-often based on real-time operating system (RTOS) fundamentals, rapid reactivity and safety among them. “We’re no longer looking at an old-fashioned interrupt-driven model. IIoT is about the RTOS model.”
What that means, Henderson said, is that cloud infrastructures and the applications that run on them must be scalable, in order to handle high volumes of data packets, and also react to even higher spikes if machinery is reporting anomalies. “Industrial IoT is about reactivity to sensor flow, and that’s a much stricter hierarchy than consumer IoT when it comes to app development,” he said.
GE is now offering Predix as a beta platform as a service to support development of cloud applications that collect data from industrial machines and apply analytics to improve business outcomes. Two starter packs are available, one for monitoring industrial equipment and the other for building an interactive dashboard. Other services, including analytics and asset management, are also currently available, with nearly two dozen other services due to arrive in 2016.