Mention the Internet of Things (IoT) in technology conversation, and chances are good that people are thinking about the devices—the things—and what they are sensing or seeing. Mention IoT at a business meeting for an organization that’s heavily invested in internet-connected things and is getting overwhelmed with IoT-generated data, and chances are good that people are thinking about how that data can better help the business and generate ROI. Oracle Magazine caught up with Harish Gaur, senior director, product management, Internet of Things at Oracle, to discuss IoT in the enterprise; connecting devices; analyzing data; machine learning; integrating devices, data, and applications; and more.
Oracle Magazine: What is IoT in general, and what is IoT specifically in the enterprise?
Gaur: Most of the focus on IoT today is in the consumer space.
IoT starts with the idea that every physical device in the world—whether it is a home appliance, an industrial machine, a commercial trailer, or a conference room—can be connected to the internet, and every device will be able to communicate and send information about its behavior to the internet. IoT then works with these connected devices—these interconnected physical devices—to collect and exchange data through the sensors that are attached to them. So that’s the textbook definition.
But if you focus on the textbook definition, IoT appears very hardware-centric, very sensor-driven, and much geekier than it really is.
At Oracle we look at the real value of IoT as much more than connecting sensor-equipped physical devices. The devices are a prerequisite, and all the devices need to be smart so that they can send and receive information. But the real value comes from making sense out of this data, from creating insights out of this data, and from taking actions based on this data. Oracle is focused less on the consumer IoT space and more on the industrial or—as I like to call it—the enterprise IoT space.
Oracle Magazine: How does IoT work with different types of information, and what is a typical industrial or enterprise IoT use case?
Gaur: People have different estimates for the number of physical devices connected to the internet. Some say there will be 20 billion devices, and some say there will be 50 billion devices.
The breadth of data that will come from sensors . . . will require different levels of analysis— real-time as well as predictive.”
Whatever the number, it is going to be so high that the amount of data coming from these devices is just mind boggling. The question then becomes: How do you work with that data and turn it into real actionable information?
So here is an example based on a customer. The company manufactures and sells industrial valves. These valves are sold and used in manufacturing plants at various other businesses, and the valves might be involved in producing high volumes of chemical products. So any malfunction of one of this customer company’s valves could have a big impact on the overall quality of the products at another company’s manufacturing plant.
So, first, there is a liability issue. Our customer does not want to be in a situation where its valve is functioning well but the quality of its customer’s product at a manufacturing plant is suboptimal. Our customer wants to be able to prove that its valve works fine. And if a problem arises with a valve, our customer wants to fix it as soon as possible.
Our customer is remotely monitoring these industrial valves. The valves are expensive—in the €50,000 to €60,000 range—so our customer can afford to instrument these valves and monitor their health, including the rate of flow and the opening and closing of a valve.
Our customer can monitor certain parameters and use that information to detect whether the valve is going to go bad or if it has already gone bad. And based on that information, our customer can dispatch a technician to go fix the valve.
This example is about monitoring raw sensor data, but more than that, it’s about using that data to understand whether the valve is currently malfunctioning, to predict future anomalies, and to use that intelligence to drive corrective action—such as creating an incident in a CRM [customer relationship management] application that will dispatch a technician to go fix the valve so that it doesn’t hinder any manufacturing processes. It’s about driving ROI from industrial or enterprise IoT.
Integration is about using what has come from connected devices and been analyzed and telling the logistics application or an asset management module, ‘Something has happened; take an action.”
Oracle Magazine: How does IoT use machine learning today?
Gaur: Machine learning is extremely important in the context of IoT. As I mentioned, the breadth of data that will come from sensors is going to be huge, and it will require different levels of analysis—real-time as well as predictive. The first level of analysis is what’s happening right now. And the challenge becomes: Can I process this data in real time and take action?
For example, if a fleet of commercial trucks is delivering perishable food to stores in a grocery chain, I want to monitor the humidity and the temperature inside the trucks so that perishable food does not spoil. And I want that information in real time. For example, if the humidity on a truck fluctuates plus or minus 10 percent beyond its hourly average within five seconds, I know there is a problem, and I want to take action. That’s real-time analysis of the data.
The next level of analysis is what we typically call predictive analytics, and that’s where machine learning comes in. Based on the past performance of this fleet, traffic patterns, weather patterns, and driver behavior, I know that I can predict an estimated travel time of three hours. And using that information, I can plan my next batch of products for dispatch, who should be driving, and what routes the trucks should take. Machine learning supports the predictions and recommendations based on the IoT data.
Oracle Magazine: What are the Oracle solutions for IoT?
Gaur: Oracle offers both SaaS and PaaS [software-as-a-service and platform-as-a-service] IoT services. For the most common use cases we see in our customer base, we provide Oracle Internet of Things Asset Monitoring Cloud Service; Oracle Internet of Things Fleet Monitoring Cloud Service; Oracle Internet of Things Production Monitoring Cloud Service; and coming soon, Oracle Internet of Things Connected Worker Cloud Service. These SaaS apps are built on top of the Oracle IoT PaaS platform: Oracle Internet of Things Cloud Enterprise.
Oracle Internet of Things Cloud Enterprise is a purpose-built IoT platform that our customers use to build their own IoT applications. The platform supports core feature levels to connect, analyze, and integrate devices, data, and applications.
Connect links the physical devices—the fleets, the machines, the assets, all the devices—and brings in the data. Oracle Internet of Things Cloud Enterprise connects the devices, pulls in data in a very secure fashion, and supports bidirectional connectivity, where the platform gets data and pushes data to devices.
Analyze is real-time analytics of data to see what’s happening right now. Oracle Internet of Things Cloud Enterprise supports split-second analytics and machine learning via an IoT analytics module. And that module is built on Apache Spark.
The integrate capability is about acting based on analysis and insights. Integration is about using what has come from connected devices and been analyzed and telling the logistics application or an asset management module, “Something has happened; take an action.”
LEARN more about
Photography by Rohan Makhecha,Unsplash