Why edge monitoring is a must in an IoT-dependent world

As edge computing accelerates, small and large portions of business applications are migrating to the edge of the network

wireless network - internet of things edge [IoT] - edge computing - digital transformation
Thinkstock

The more intelligent the edge becomes, the more important it is to understand the performance of the application at the edge and how it impacts the overall business performance.

The growth of edge computing is being fueled by the internet of things. Vast amounts of data are being generated by sensing devices that capture information about the physical environment—everything from humidity and light to chemicals and vibration. According to Cisco, the data from smartphones and machine-to-machine modules is expected to grow at a rate of 49 percent a year for the next four years.

As the amount of data generated per hour rises from tens to hundreds of gigabytes, sending such mass quantities back to the cloud for analysis simply isn’t viable. The consequences for not meeting response time requirements due to network latency are too high. Connected cars receiving sensor input from the front and rear of the car via lidar/ladar systems and cameras cannot risk a poor response time when requesting an action like braking or deceleration from the car control system. Likewise, connected roadways that rely on roadside sensors to dynamically change road signs warning drivers about severe weather conditions like flash fog and flash floods must process data on humidity, wind speed, and acoustics in real time. Any delay creates a risk of congestion, accidents, and even deaths.

IoT has created a smarter and more complex edge

The need for real-time edge computing also comes into play when sensors are located in areas where network access is intermittent or is limited by cellular or satellite bandwidth. Such is the case with fleet management and with monitoring industrial machinery in remote areas such as mines or oil and gas fields.

While apps are increasingly moving to the edge, performance monitoring has only recently begun to follow. When the edge was regarded as dumb, there was little need for robust diagnostics. Sensors worked, or they didn’t. They weren’t parts of more complex systems that could suffer from problems like thread contention or unavailable memory independent of the sensors themselves.

With a smarter edge, performance monitoring is critical. Without performance monitoring there is very little visibility into issues faced by remote applications and microservices—they occur in the dark. If an IoT device stops responding—for example, if sensors in a parking garage or on a platform in the Gulf of Mexico go silent—someone usually has to go out to the site and investigate. Connected cars must be brought in for repairs. This lack of visibility means it is difficult to prioritize which issues need immediate attention and which issues are likely to have little or no business impact.

The biggest challenge for performance monitoring at the edge is the rise of new, more complex architectures. The more devices that are sending data, and the more data there is in need of processing, the more layers or tiers of computing at the edge are required. For example, the sensors on a manufacturing floor might send their data to a single gateway which might send data on to the cloud. Similarly, a single smart garage might collect data throughout the structure and process it at a single gateway, forwarding on to a single cloud application. However, a smart city managing parking in multiple garages in multiple neighborhoods likely requires more computational layers at the edge (e.g., each neighborhood might get its own server or other physical resource). Ensuring that every layer is performing well requires an end-to-end view of requests for computing services, often known as business transactions.

Mapping IoT data back to business initiatives

The ability to track important business transactions at the edge and to correlate those requests with the impact they have on the profit and loss of a business will be the ultimate key to success in an IoT-dependent world. In the case of consumer devices, the level of satisfaction a customer is getting from their new car, home entertainment system, voice assistant, etc could affect the customer’s future buying decisions or lead them to post negative reviews. Parking garages will be able to determine if unresponsive screens or card reader problems are preventing payment, leading to potential city revenue losses. Factories will be able to precisely schedule maintenance in order to maximize production. Fleet managers will know immediately if valuable cargo, such as biomedical supplies, is at risk due to a change in temperature or humidity—and if the failure of an individual sensor requires diverting the vehicle for emergency maintenance. They’ll also be able to determine if cargo has been tampered with.

IoT is still in its early days. Last year, 61 percent of IT and business decision-makers responding to a survey said they had barely begun to scratch the surface of what IoT technologies could do for their businesses.

As IoT continues to grow so will the app-ification of the edge and the need to monitor the business logic that resides there. For better or for worse, actions at the edge will increasingly affect business outcomes, determining whether SLAs are kept or broken, whether customer relationships are deepened or damaged, and whether new business is won or lost.

Copyright © 2018 IDG Communications, Inc.