Architecture, Use Cases, and Examples

Key takeaways

Using the term “edge computing” connotes a physical location.
A simple edge device can either collect or send data, or both, simultaneously.
These technologies will allow for virtualization and automation that has not been explored yet, such as better vehicle autonomy and workload migrations to the edge, while making wireless networks more flexible and cost-effective.

Edge computing brings AI and analytic workloads closer to where data is created, and actions are taken—reducing latency, decreasing network demands, increasing the privacy of personal and sensitive information, and improving operational resilience.

*Rob High,*IBM Fellow, VP & CTO, Networking & Edge Computing, IBM Cloud

Introduction

The edge architecture is a distributed computing architecture that incorporates every component active in edge computing—everything from devices to sensors to servers to clouds—wherever data is handled or utilized in remote areas of the network.

Using the term “edge computing” connotes a physical location. Instead of depending on the “bigger” clouds to perform all the work, it is computing that is done at or near the source of the data. An advantage of edge computing is that it brings corporate applications closer to the areas where data is generated and action is required. What is the source of all this information?

Devices that govern data flow at the network boundary are edge devices. With the rise of cloud computing and the Internet of Things (IoT), the importance of edge devices has increased, necessitating more intelligence, computation, and AI services at the network’s edge.

The term “edge computing” refers to processing data at the place closest to the source or destination of the data in a system—where it comes from or goes to. By lowering latency and lag, edge design speeds up processing. As a result, applications and programs running at the edge can perform better for the user and enhance overall performance. Each dock where goods are loaded and unloaded and inspecting the product before it leaves the pier is an example of an “edge” for a shipping firm. Even if your company’s primary data center is remote, the edge is where your apes are being processed.

What is Edge Computing? #

With edge computing, you not only get a lot of data from a lot of different devices, but you also have to make sense of all that data and let the devices decide for you. We will talk about IBM’s offer and how to design an edge solution, and the different parts that make it up. Use cases from different industries that deal with scalability, latency, and security will be shown.

Edge computing architecture is an ecosystem of infrastructure components that have been spread out from the central location of an enterprise’s data center or main server to all of the company’s other locations. This is part of the company’s overall deployment. Computing and storage capabilities, applications and devices, and sensors are all part of this. They work together with IoT devices in a central data center or cloud.

When we talk about edge computing, they have enough bandwidth, memory, and processing power to collect, process, and act on data in real-time without the help of other parts of the network, so they do not need to ask for help from other parts of the network. Some connection with the network lets the device talk to a database in a single place. It is easy to move scaled-down, on-premises edge servers or data centers around and fit them into small remote places. Red Hat has a wide range of flexible topology options that can be used in small spaces or different environments, such as power/cooling and intermittent network connectivity. This is just one example of what Red Hat can do with edge computing.

Key Components of Edge Ecosystem #

Cloud

Clouds can be public or private, and they can be a place to keep the container-based workloads like applications and machine learning models that run on them. Also, these clouds run and host the applications that are used to manage and orchestrate all of the different edge nodes. There will be many interactions between the workloads on edge and the workloads on these clouds. The cloud can also be a source and a destination for any data that other nodes need to do their jobs well.

Loads are apes that need to run on this architecture layer because they cannot run on the other edge nodes or the management layers. Both application and network workloads need to be moved to different edge nodes by using suitable orchestration layers.

Edge Hardware #

An edge device is a piece of equipment made for a specific purpose and has a lot of computing power. Many exciting things can be done with edge devices. These include an assembly machine on the factory floor, a cash machine, an intelligent camera, or a car that can see things. People often see edge devices with ARM or x86 class CPUs with 1 or 2 cores, 128 MB of memory, and maybe 1 GB of storage that is not on the device itself. Edge devices can be more powerful, but they are not the norm now.

They are the systems that are on-site or at the edge of the network. They can be separate physical or virtual servers in different places, or they can be combined in a system called a hyper-converged system. There are two main parts to this layer of the architecture. It will be home to both the parts of the systems that will run these applications in these architecture layers and the applications on the edge of these devices.

Application Layer: People who want to run big apes that cannot run at the edge because they are too big for the device will run these apes in this layer. Video analytic and IoT processing are two examples of complex applications.

Network Layer: Physical network devices aren’t usually used because it’s hard to keep them up and running. Almost all of the network layer has been virtualized or containerized. These could be routers, switches, or any other network parts needed to run the local edge.

The Edge Node #

An edge node is a general term for any edge device, edge server, or edge gateway that can be used for edge computing. It can be used to refer to any of these things, such as cameras, sensors, and other physical devices that run on-site at the edge are called edge devices. A simple edge device can either collect or send data, or both, simultaneously. The more complex edge devices have more processing power to do more things. In either case, it is essential to be able to put and manage apes on these edge devices so that they can work. Specialized video analytic, AI models that learn quickly, and simple real-time processing apes are some examples of applications that use this type of technology. IBM uses its IBM edge computing solutions to run and manage containerized applications on these edge devices.

Edge Cluster or Server #

An edge cluster/server is a general-purpose computer that is used in a place where people work, like a factory, retail store, hotel, distribution center, or bank. An edge cluster/server is usually made from an industrial PC or a computer rack. People often find edge servers with 8, 16, or more cores of computing power, 16GB of memory, and much storage, like hundreds of GBs. An edge cluster or server is usually used to run enterprise application workloads and other shared services, such as email.

Edge Gateway #

“Edge gateways” are usually edge clusters or servers that, in addition to running enterprise applications and sharing resources, also have services that can do things like translate protocols or connect to networks. A few edge devices can act as a limited gateway or host network function, but most edge gateways are separate from edge devices. Fixed-function IoT sensors collect and send data to an edge/cloud, but they do not have inboard computer memory or storage.

Let’s discuss a few use cases to understand the power of edge computing and how it will rule the world in the future.

Real-life Case Studies: #

There is a plethora of problems to be solved in every complicated workplace. On an individual basis, it will be necessary to decide if an edge computing solution is warranted for a given business challenge or use case. The following are some possible industrial use cases, as well as real-life solutions, for reference.

Case Study #1: Smart Cities #

Due to aging infrastructure, operational inefficiencies, and increased demands from inhabitants, operating and regulating cities has become an arduous undertaking. New technologies, such as IoT, edge computing, and mobile connection have made it easier for innovative city solutions to gain traction with residents and municipal officials.

IoT is the foundation of every smart city solution. These devices may be embedded within the city’s infrastructure and assets to monitor the functioning of these assets and offer meaningful information about their behavior. Edge computing is a prerequisite for delivering the city’s mission essential solutions, such as traffic, flood, security and safety, and critical infrastructure monitoring, due to the need to give a real-time decision and avoid transferring a significant quantity of sensor data.

Case Study #2: Manufacturing Sector #

As a result, a quick reaction is vital for reducing product failures and increasing efficiency in production. To maximize the efficiency of each machine, analytical algorithms keep track of how well it’s performing and adjust the settings—predicting when a failure will occur and scheduling maintenance between runs may be done with the use of analytic algorithms. Using different edge devices, it is possible to send important information to many edges, including the cloud.

Take a look at this: Reducing downtime is a priority for an electric bicycle manufacturer. Factory floor equipment includes presses, assembly machines, robots for painting, and conveyors. You lose $250,000 per hour if there is a problem with a manufacturing run. Making a profit is the difference between a company operating at maximum efficiency and experiencing unanticipated interruptions. The following are required to keep the factory running at maximum efficiency for it to function smoothly:

Methods for analyzing equipment performance and adjusting its operational settings to maximize efficiency.
To plan maintenance on equipment in between runs, you’ll need algorithms that can identify and forecast failures.
To accurately predict failure, models specific to each use case must be developed.

Conclusion

Edge computing is constantly changing and getting better. It uses new technologies and practices to improve its abilities and performance. Edge availability is one of the most important trends, and edge services are expected to be available worldwide by 2028. There are many different ways that edge computing can be used right now, but the technology is expected to become more common and change how the internet is used in the future. This will bring more abstraction and use cases for edge technology. In fact, there have been a lot of new products for edge computing, like computers, storage, and network appliances that are made for this type of technology. More multivendor partnerships will make it easier for products to work together and be flexible at the edge. There is an example of this, which is when AWS and Verizon work together so that the edge can get better connectivity.

Wireless communication technologies, such as 5G and Wi-Fi 6, will also impact edge deployments and use in the coming years. These technologies will allow for virtualization and automation that has not been explored yet, such as better vehicle autonomy and workload migrations to the edge, while making wireless networks more flexible and cost-effective.

Related reading