Utility companies collect data from electric meters, smart roadways monitor weather conditions, and intelligent kitchen and bath fixtures create personalized experiences for homeowners. These are all examples of edge computing combining with the Internet of Things (IoT) to enable people to gain quicker insights at the edge. Edge computing is an architecture in which data is processed closer to where it originates.
Gartner Research projects there will be 20.4 billion connected things being used globally by 2020. Meanwhile, IDC forecasts the "global datasphere" to increase 10 times to 163 zettabytes (ZB) by 2025.
"Edge computing is moving compute as close as feasible and necessary to the physical world to meet certain performance needs, security elements, or costs factors," explained Jason Shepherd, IoT CTO at Dell Technologies. The company is one of a number of companies to offer IoT Edge Gateways that aggregate analytics at the edge from sensors. In edge computing, devices and sensors send data to a local gateway rather than transmitting all of the data back to the cloud.
The Connection Between Edge and the IoT
The IoT involves billions of sensors and devices that gather data at the edge to help companies gain insights and make business decisions. These decisions may include whether to change the temperature of a room or turn off the water in a field. It's not possible to take all of the data from the edge and transfer it right into the cloud. Therefore, you need to use a process of aggregation to organize the data into "smaller numbers," explained J. Craig Lowery, Ph.D., a Research Director in Gartner's Technology and Service Provider group.
Edge computing provides the opportunity to reduce latency when performing analytics in the cloud. But after data is collected on-premises, it is shared in the public cloud. "For the most part, edge computing is going to be about having some on-premises compute that works with the public cloud in a hybrid fashion," Lowery said.
The edge could be a rack of servers containing ruggedized hardware, a cell phone tower equipment shed, a manufacturing facility's data center, or a cable company's neighborhood office, Shepherd explained.
For an electric company, the process entails telemetry traveling through wires to a utility office. The data then gets sent to the public cloud where data analytics is performed. That data lets electric companies keep track of electricity consumption and get broader trends about consumption for electricity. You can also pull data at the edge in the digital signage, energy, industrial automation, and transportation markets.
The Edge Lowers Latency, Boosts Bandwidth
Performing data analytics closer to the edge brings more security, lower latency, and greater bandwidth, according to Shepherd. Latency can suffer when there is a long amount of time required for an edge device to send data to a network and then to a data center when performing analytics on data in the cloud. Edge computing provides a way to not only speed up data analytics but also lessen network traffic. Here are three examples of implementations of an edge IoT platform:
1) Microsoft Azure IoT Edge: In April 2018, Microsoft announced that it will invest $5 billion in IoT technology. As part of this investment, the company offers Microsoft Azure IoT Edge, a platform that transfers cloud analytics and business logic to devices. The platform includes three components: containers called "IoT Edge" modules that companies can deploy to edge devices and connect to the Microsoft Azure public cloud architecture, third-party applications, or their own code; an IoT Edge runtime that operates on IoT edge devices and manages the modules found on IoT edge devices; and a cloud-based user interface (UI) that lets companies monitor and manage the data from IoT edge devices remotely.
Image used with permission from Microsoft.
Microsoft Azure IoT Edge is a "good example of general-purpose edge computing because it's not really a private cloud," Lowery said. "It's like a public Azure coming to life in your data center. It's an extension of that service into your data center, running on equipment that is on your own premises. You may or may not own it. There are a lot of different ways that it's delivered from different services."
The IoT is a strong use case for Azure Stack, the extension of Azure that lets companies run hybrid apps on-premises, because customers purchase it as a service rather than having to deploy it, Lowery said.
Image used with permission from Microsoft.
2) Dell Edge Gateway for IoT: A hub for wired and wireless systems, the Dell Edge Gateway for IoT aggregates data and sends it to the cloud. The Dell Edge Gateway for IoT is ruggedized to withstand harsh conditions in a factory or on an oil rig.
With the millions of devices collecting data in the field, it takes a gateway to process all of this data, according to Shepherd. "The interesting thing with the IoT is that it flips it completely backward, and now I've got millions of devices that want to hit one server," Shepherd said. "So that's why you need to fan more compute even closer to those devices."
3) EdgeX Foundry: Hosted by The Linux Foundation, EdgeX Foundry is an open-source project designed to create interoperability around IoT edge solutions. EdgeX Foundry is modeled after Cloud Foundry, the open-source, multi-cloud app platform, and extends the same format to the edge, Shepherd said. It's a framework that brings together thousands of protocols, with the cloud on top and any other device below, according to Shepherd. EdgeX Foundry provides the flexibility to create solutions at the edge.
More than 70 companies contribute to EdgeX Foundry, including AMD, Dell EMC, Samsung, Zephyr, and VMware. But EdgeX Foundry is "a vendor-neutral, open-source project that [Dell] helped get started that decouples the edge from the cloud and the apps from the underlying infrastructure," Shepherd said. "You get to have open frameworks for underlying compute."
Dell has provided a Filesystem in Userspace (FUSE) source code base under Apache 2.0 for EdgeX Foundry. It also provided at least 12 microservices and more than 125,000 lines of code.
Image used with permission from EdgeX Foundry.
5G, AI to Shape the Future of Edge, the IoT
Artificial intelligence (AI) will be a key factor in the IoT and edge computing. Advanced AI will be able to run on devices close to the edge to enable analytics to be performed, notes Sam George, Partner Director for Azure IoT at Microsoft, in a blog post. Microsoft has an initiative called Project Brainwave that uses field-programmable gate arrays (FPGA) to perform AI calculations.
Another key driver of edge computing will be 5G, noted David Mayo, Senior Vice President and 5G & IoT Business Chief at T-Mobile, which recently launched the first nationwide narrowband IoT network.
"With edge computing, you're capable of distributing that processing capability closer to the user, and you've got the reduced or compressed latencies that a 5G network will provide," Mayo said. He sees 5G networks powering the sensors in autonomous cars and the array of sensors in oil and gas refineries.
With all of the data points that can be collected in real time, edge computing will have a beneficial effect on consumers, according to Mayo. "Over time, the need for computing to happen even closer to the user, whether it be a business or a consumer, is going to increase," he said. "Whether it's augmented reality [AR] or virtual reality [VR], the closer the compute occurs and exists to the edge, the better off the customer experience is going to be."