In emergency medicine, there is the concept of the ‘golden hour’, that period of time after a major traumatic injury where getting the right treatment is crucial to the eventual survival of the patient. Once this window of opportunity closes, the actions of the medical staff have less of an impact and the results can be catastrophic.
The same principle can be applied to the Internet of Things (IoT) and real-time analytics. The latest estimates place the number of connected devices anywhere between 20 billion and 50 billion by 2020 and the IoT is gathering significant momentum in both the home and the industrial sector. However, many of these systems and platforms use cloud-based analytics, which is effective, but increases the time to take actions on the device.
This time delay may be an annoyance when it happens within a residential scenario, but can be disastrous in time-sensitive and industrial environments where immediate detection and actions impact security, safety, and machine failures. With cloud-based solutions, data will not be provided quickly enough to enable the IoT to be fully effective for mission-critical applications. This issue is not only a huge concern now, but one that is going to be exacerbated as IoT deployments grow in scale.
The True Edge
To support a massive number of IoT devices and radically reduce decision and action time, analytics need to be pushed as close to the device edge as possible. Too much IoT data is dying in data lakes and edge analytics can solve this problem by providing real-time actionable insights as events occur.
Traditional Big Data methods are great for batch-oriented analytics, but aren’t suitable for the interactive, real-time analytics required for IoT. With the huge amounts of data produced by large machines and engines, or wide networks of millions of smaller devices, it is impossible to massively mine and still get real-time decisions. The data that flows through these networks is vast, fast, and needs an active analytics framework to identify issues and take actions as the data flows.
If you need real-time decisions, actions and insights, then you need an architecture which supports the streaming of edge analytics, cloud analytics, and real-time and historical blending. This principle applies to smart factories, smart cities, smart homes and any environment where “smart” means intelligence at the edge.
The IoT industry is beginning to understand that one of the best ways to provide these insights from the vast numbers of IoT devices is to do as much analytics as possible at the true edge of networks on the devices themselves. This architecture makes devices truly smart, allowing them to analyze, detect and act on the data as it flows to the device.
Traditional methods of passing raw data to the cloud, analyzing it and sending actions back to the device, introduces time delays which can be detrimental to operations. However, by implementing analytics and pattern detection at the edge, an almost-instantaneous response can be achieved in situations that require immediate action, such as alerting users that their device or network is under attack, or shutting down an engine which is exhibiting dangerous behaviors.
In a nutshell, the longer data travels for analysis, the longer an anomaly which could have a detrimental impact goes unnoticed. With edge analytics, machines, and smart sensors, we can collect information using contextual time windows using the cloud to aggregate non-critical data and combining it with data from devices in real-time to detect anomalies and performance issues.
The Internet of Data
Edge analytics platforms that deploy this model, such as Greenwave Systems’ AXON Predict, give developers an easy onramp to connect devices, allowing them to dynamically deploy analytic scripts to the edge and coordinate distributed analytics between cloud and edge. This allows organizations to realize the full potential of real-time analytics for the IoT with anomaly detection and instantaneous, actionable insights, and enables them to capitalize on a valuable market opportunity by providing a customized analytics solution that adds computational power and real-time intelligence throughout Industrial IoT segments. Looking to the future, this also provides a smooth path to add machine learning and Artificial Intelligence at the edge.
We now live in a world where everything is becoming connected. The whole dynamic of the marketplace is changing and operators, manufacturers, utility companies, and healthcare providers are all looking at incorporating the IoT in a way that delivers their services in the most efficient and profitable way. The only way this can be done is if the masses of data that are being created by the IoT are monitored, processed, and actioned upon in real-time. If this doesn’t happen, there is a risk that the IoT will end up being viewed as only suitable for trivial, small-scale applications instead of reaching its full potential.
The enterprise market needs to find solutions that fit neatly with what they are looking to do with the IoT. If they can manage massive amounts of data processing, monitoring, diagnostics and service management, they can get the most out of their IoT deployments, but it all starts at the true edge of the network. If they can acknowledge that, their critical data will not only stay alive, it will thrive.
About the author
John Crupi is currently the VP of Engineering & Systems Architect at Greenwave Systems where he focuses on real-time visual edge analytics and patterns. Prior to joining Greenwave, he was the Chief Technology Officer at Predixion Software where he was responsible for the Internet-of-Things (IoT) and Real-time Visual Edge Analytics. Prior to working at Predixion Software, Mr. Crupi was the Chief Technology officer at JackBe, which was later acquired by Software AG. He has also served as CTO at Sun Microsystems, where he was recognized as a Sun Distinguished Engineer. Having over 25 years' experience with enterprise systems and advanced visual analytics, John holds a BS degree in Mechanical Engineering from the University of Maryland and earned a Masters degree in Engineering Administration with a minor in Artificial Intelligence from George Washington University.