Leveraging the Internet of Things (IoT) allows you to improve processes and take your business in new directions. But it requires you to live on the edge. That’s where you find the ability to empower IoT devices to respond to events in real time by capturing and analyzing the relevant data.
Edge computing relies on squeezing the power and functionality of a data center into a micro site as close to data sources as possible to enable real-time tasks. Whether the task involves self-driving vehicles, online transaction fraud prevention or responding to alerts from health-monitoring devices, there’s no time to waste. Consider the potentially catastrophic outcome of two autonomous vehicles on a collision course or taking a beat too long to act on an alert from an implanted medical device.
In either case, an automated response — applying the brakes, dispatching an ambulance — can avert disaster. And that comes down to being able to act on data at the precise time it requires action. Deploying machine learning (ML) and analytics capabilities at the edge is what makes this possible.
The edge is a critical component of many digital transformation implementations, and particularly IoT deployments, for three main reasons — immediacy, fast-changing datasets and scalability.
The ability to react in real time to continuous data flows, and to quickly adapt to new datasets, makes companies more agile so they can improve their operations and accelerate go-to-market strategies. The result is to not only boost the bottom line but also deliver products and services your customers need, when they need them, to better their lives.
The IoT depends on edge sites for real-time functionality. Without them, data collected by IoT sensors, cameras and other devices would have to travel to a data center located hundreds or thousands of miles away.
In such a scenario, data latency is essentially unavoidable — and, when real-time action is required, inadmissible. It’s easy to understand why if we’re talking about a potential collision of autonomous vehicles or a warning that a patient is about to go into cardiac arrest.
Real-time analytics isn’t always about life-and-death situations, though. If you’re in financial services or retail, it’s about saving people from other dangers, such as cyber threats and fraud. Credit card companies use server logs and transactional data to prevent fraudulent transactions in real time. ML can stop a transaction if the algorithm detects anomalous behavior indicative of fraud.
Other examples include cyberbullying and dissemination of fake news through social media. Facebook and Twitter, for instance, have started using ML algorithms to detect and stop these types of activity. The algorithms are imperfect, but they will get better in time as they “learn” by processing more data.
Data analytics at the edge also allow organizations to cope with another significant challenge. As Bernard Marr, a futurist and technology consultant, explained in a Cloudera digital event, that today’s datasets have a short shelf life. “Datasets that are three months old are no longer relevant.”
Take the data that government agencies and medical researchers have tracked during the COVID-19 pandemic. New knowledge has come to light fast and furious, making data from even a week earlier no longer relevant, Marr pointed out.
The same is true about other aspects of life, such as consumer behavior. “Even things like credit scores that were relevant and meaningful six months ago are no longer relevant,” he said. With analytics at the edge, researchers can adjust their work as new data comes in. If the data has to land in a data lake before being analyzed, insights that can be acted on right now would take weeks or months to understand.
The edge also makes it easier to scale data-capture operations. Imagine if all the data your organization collects from hundreds or thousands of IoT endpoints had to be processed in a central location. If you’re in retail, finance, healthcare or another environment where real-time action is required, a central system would quickly get overwhelmed.
“Streaming analytics deliver predictive and prescriptive insights to prevent mishaps from happening as well as to capture ideal business opportunities at the right moment,” said Cloudera’s head of product marketing Dinesh Chandrasekhar in an interview with VMblog. “Agility is the order of the day and streaming analytics help deliver that.”
Streaming analytics makes it possible for a self-driving car approaching a red light in Seattle to know when to stop because an edge site nearby told it to. Or for a patient going into cardiac arrest in Denver to quickly get an ambulance. Or for an ML engine to stop a fraudster from impersonating an unsuspecting consumer in a financial transaction.
These split-second decisions can deliver a lifetime of benefits in a hyper-connected world — and make it an imperative to live on the edge.