The Industrial Internet of Things (IIoT), also known as the Industrial Internet, brings together brilliant machines, advanced analytics, and people at work. It’s a network of a multitude of devices connected by communications technologies resulting in systems that can monitor, collect, exchange, analyze, and deliver valuable new insights like never before. These insights can then help drive smarter, faster business decisions for industrial companies.
Industrial companies will adopt IIoT for 3 distinct outcomes:
Driving Operational Efficiency: dynamically managing and optimizing usage and control systems to drive the best experience at the lowest cost.
Predicting Failures: conducting proactive maintenance to reduce unplanned downtime, minimize maintenance costs, and extend asset life.
Drive Digital Innovation: innovating across connected ecosystems and driving new business models and new revenue streams, e.g., the
Asset Administration Shell (AAS).
Today, innovations in hardware, connectivity, big data, analytics, and machine learning are converging to create massive value-creation outcomes. Numerous studies predict that the size of these opportunities could have a total economic output of many trillions of dollars based on the billions of sensors and devices that will be connected by the IoT players.
Yet only one thing catches my attention: data. The power of data to transform these industries is huge. But companies struggle with it. They are unable to capture and use relevant data from multiple IoT systems due to structural, commercial and technical barriers. This manifests itself as a lack of understanding of the potential use of data, an inability to store and transact from remote sites, and finally, as a challenge to extract actionable insights and drive closed loop action. For instance, it is one thing to trigger an alarm when a piece of equipment fails, but totally different to do an alert for condition-based-maintenance where inputs from various sensors are analyzed across multiple variables to predict a future failure.
IIoT data today has 3 distinct constraints that need to be addressed:
The 1% Problem: the McKinsey Global Institute states that only a very small percentage of IIoT data collected is actually ever used. IBM goes one level further, describing the data that is not collected as “dark data”. But the important aspect here is that even if the data is collected, it is only used for mundane control purposes and anomaly detection. It is not used to predict failures or to optimize operations proactively.
The Scale: this is one reason why industries don’t store a lot of data today, and the data remains dark. The scale of the information spit out by these machines and processes is huge, ranging from 400GB a day for a turbine blade, to 500M readings on a smart city meter, to petabytes of weather info in agriculture. An average Formula One race car has over 50 sensors that, in total, generate around 1GB/hr of data. Now imagine the data output on this car throughout the racing season.
The Shelf Life: to me, this is the most important thing. Even if we were able to get to all the data and store it, it is important to note that the usefulness of the data is very very short-lived. The value of information erodes so fast that decisions have to be made even faster.
This brings me to the elephant in the room. Are the transactional and decision analytics systems of today sufficient? How fast do decisions need to be made to be effective? Today’s transactional analysis is designed for actions every few minutes or even hours. It values consistency. The new requirements, however, are expressed in milliseconds or even microseconds, and consistency gives way to efficiently and effectively analyzing the information.
These new systems will allow us to:
Generate and capture valuable machine, operational, and environmental data.
Store, retrieve, and process massive data sets from disparate sources — including the silos of Industrial Information technology (IT) systems and Operational technology (OT) systems.
Turn that data into operational insight.
The good news is that companies are investing in these systems. The key element of these systems is a datastore that can scale infinitely, perform in milliseconds, and be cheap. Yes, I said it — cheap. IIoT will drive computing some of these decisions at the edge, since a lot of the data cannot be transmitted back and forth to a centralized data center. This means that these stores have to perform exceptionally well at the lowest possible price.
It always comes down to the bottom line, or as oilmen would say, “You can’t get the grease without a lease”.
For more information:
Unlock Possibilities in Milliseconds –
HTAP architecture and how it enables transactional analytics, see blog post “
Transactional Analytics and Gartner’s New Market Guide for HTAP.”
