Today technology is cluttered with buzzwords and hype-circles that prognosticators, media, and analysts pitch routinely. Edge computing is this era’s buzz word. Is edge computing a new hot topic of a recently discovered technology that will change the way we live? No, it is far from a new technical revelation, but it may change the way we interact, work, and entertain ourselves in a digitally connected world.

Edge computing has extensions to the Internet of Things (IoT), 5G and cellular communication, machine learning, and Industry 4.0 (all buzzwords to their own extent). You purchased this book for clarity and guidance around this subject and are trying to determine how edge computing may fit into your organization or business. Above all, you want to know: “what is it and what can it do for me”. In short, there is a massive amount of data being generated in the world. Each year 20% more data is generated than the previous year. This makes sense with nearly everything being digital and interconnected.

Consider these facts at the time of this writing:

• There are between 3.5 and 4 billion daily users of smartphone devices worldwide.
• Nearly 45 billion IoT connected devices and sensors are online transmitting data.
• In 2019, Netflix video streaming alone was responsible for 61% of all network bandwidth.
• 254,000,000,000,000,000,000 bytes of information were transmitted across the internet.

Each one of these devices whether it’s a sensor, a smartphone camera, or a server is a source of data generation. Much of the data created in the world is not even created by humans, they are autonomous machines creating data that are consumed by other machines. The availability of cheap computing, cheap sensors, nearly unlimited storage, and fast methods of data communication have allowed the internet and modern digital computing to impact nearly every aspect of life.

Edge computing serves a purpose by bringing programmable machines close to where data in generated or data is consumed. Its purpose is to fill a gap between computing in the cloud versus computing exclusively at the source of data. We will learn in chapter two that this is not a revolutionary concept and has been around for decades.

Edge computing for all intents and purposes are nearly the same from a basic hardware concept as a computer in the cloud. They have processors, memory, storage, and communications. However, the difference is in scale. Typically edge computers will not have the processing performance, storage capacity, and networking fabrics of a machine found in a hyperscale datacenter. They will be smaller, remotely located, and in some sense less capable. Edge machines pick up where the cloud leaves off. Think of an entire computing device providing the services of a large datacenter in the form factor of a small Raspberry Pi single board computer.

How to read this book

This book is intended to be focused on the high-level overview of edge computing and all the constituent components that make up edge hardware, software, and services. It is intended to quickly instruct and inform the reader on edge computing use cases, the myriad of technologies forming the edge, and the risks of improperly designed systems.

One does not need a deep technical background to get something out of this book. This work is targeted for both senior technologists and engineers that need to quickly ramp up their understanding to include edge systems. Someone versed in technology will find that edge computing involves a variety of disciplines from communications systems and theory to embedded systems design to security. This book is designed to also be understandable for project managers and senior executives that need to understand what the edge is and how to separate the hype from reality. The material should also be relatable by providing concrete real-world use cases that have been deployed and are commercially successful. After all, if the technology doesn’t make business sense, then it is more of a hobby.

What is the Edge?

We need to apply rigor in defining what this edge is. The edge has lots of definitions and many organizations misuse the term.

For now, the six basic goals of edge computing are as follows:

1. Reduce the latency in processing, storing, or transmitting data.
2. Reducing the cost of cloud and telecommunications by filtering data and aggregating data sources.
3. Bridging networks and communications from simple sensors and non-IP capable networks to IP and internet capable networks.
4. To provide resiliency of computing and critical communications to remote areas as well as devices that are moving. We all have witnessed dropped calls and lost connectivity, which is not acceptable for critical applications.
5. To provide security and data assurance. In some situation data must be scrubbed or “denatured” of content before being send to the cloud.
6. Edge devices can programmatically compute, but also are remotely managed/administered. In some cases, they form an extension of the cloud, in other cases they are more autonomous.

These requirements cannot be satisfied with today’s current infrastructure of cloud computing. The hyperscale public and private clouds provide relatively frictionless computing ability that can automatically scale globally. Cloud infrastructure consists of powerful server blades, high speed petabyte storage clusters, 400-gigabit optical networks, megawatts of power distribution, air and liquid cooling machinery, and the hardened security that surrounds them are few and far between. For example, worldwide Google has 20 datacenters spread between North America, South America, Asia, and Europe. Microsoft managed 58 data centers, and Amazon has 22. These are built in areas where there is geographical need as well as reliable and inexpensive power. The cloud serves a different function than the edge.

Edge devices touch the physical world, whether that’s a patient in a hospital, a fleet of trucks, a surveillance system in New York City, or users watching Netflix videos on a smartphone. These uses cases produce as well as consume massive amounts of data and require real-time control with as little latency as possible. Think of the edge as a “micro” datacenter that exists outside of the confines of the megawatt-capable billion-dollar datacenter.

It’s a fair question to ask: “why aren’t there only edge machines and why require a public cloud?” The answer is that the services provided by the cloud will still need to exist. There is no better method for much of the compute required for business applications and the internet in general. Software-only tasks such as global websites, large database applications, machine learning training, file and data archives, and office productivity/collaboration tools run seamlessly on the cloud. Individual edge devices have a limited set of resources. Clearly, a large datacenter can provide significantly more compute and storage resources than an edge device. However, datacenters are centralized and built to house thousands of servers close to the cheapest power plant or dam. Edge machines are found everywhere and often are in mobile or moving systems like vehicles. They reside in the harshest of environments and where communication and reliable infrastructure is something to be concerned about. Cloud customers can’t afford disruption of infrastructure. The edge is built to robustly handle disruption.

Another factor separating cloud computing from edge is ownership. Cloud technologies became prevalent and are now mainstream due to the digital transformation of on premise data centers that were owned, managed, and funded by companies and private enterprises. This came at a substantial cost in capital equipment like servers and networking systems. The cloud allowed companies to shed the up-front capital costs that did not amortize as well as using a hyperscale public cloud. The bottom line is that public clouds are financially and technically better than private datacenters (except in some cases of security and specialty systems). Edge devices are a different story. In many cases edge computing is owned by a service provider or the final customer. While some cloud providers are extending their reach into edge computing either through dedicated hardware or software middleware, most edge systems are privately owned. The reason is that edge systems are nuanced. There is no general edge computer you can deploy that will work in farm machinery as well as palliative care home health monitoring. Edge systems are also deeply customized for end solutions. They can be bespoke and custom designed for their mission.

How would current life be different without edge computing? How has it made any difference? Technologies that are dependent on real-time decision making, low latency, and reliability are built on the edge. Take Netflix for example. Originally, Netflix video content was archived and delivered from Amazon AWS cloud centers exclusively.

While this can work, three problems arose:

1. The amount of data Netflix consumer on the internet backbone fabric grew year over year.
2. Users demanded near instantaneous play without lengthy buffering.
3. Content size was increasing as users migrated from 1080p HD videos to 4K videos this increased file sizes by 4x – exacerbating the problem.

Without moving streaming systems to closer to the subscribers, the backbone would become saturated and lead to more bottlenecks and poor latency. Netflix decided to build edge streaming systems that could be deployed at various Internet Service Provider (ISP) datacenters that were closer in many cases to the subscribers. This allowed traffic on the backbone of the internet to be reduced while also eliminating several hops across internet routers to AWS. Netflix calls these edge systems “Open Connect Appliances” and they sit in secured ISP centers across various regions. These machines are essentially datacenter blades with similar memory and processing capabilities. The machines are solely managed by Netflix. As matter of fact, the machines cannot be accessed by the ISP operator and are intended to be self-monitoring and self-reliant.

bWhile we should now have a clear definition of edge computing, we will next describe what edge computing isn’t and importantly how it differs from cloud computing and the Internet of Things.

Edge Hype and What the Edge is not

While edge systems are certainly part of recent hype and conference buzzwords, it is not a new concept entirely nor is it revolutionary. Internet connected devices have been in existence for over 40 years. One of the first remote devices connected to the Internet was a soda machine in the computer science department of Carnegie Mellon University in the 1980s. It allowed computer science students to buy soda remotely via their workstations. The edge hover is different, and it complements cloud computing by extending its reach to remote devices and machinery.

However, innovative technologies generate hype. One can look at Google search trends to see the interest in edge computing. In 2015, there was barely a mention of edge computing in marketing nor technical literature. Interest continues to peak month after month well into 2022. Shown in figure 1, one can see the steep interest in edge computing.

Figure 1: Google Search Trends 2011 to 2022 on Edge Computing

With peaking interest comes hype. Edge computing serves many functions, and we will see how it is essential in different use cases. There are limits and edge computing does not have a place in some environments. This book should help you understand what the edge is and where to use it. Just as importantly, it should help you understand where not use edge technology.

The edge is not a replacement for public cloud services. The cloud and massive hyperscale datacenters will continue to serve a function for a vast majority of information technology. Often it is the case that native cloud applications act as the overall administrator and orchestrator of “worker” edge nodes. The cloud serves this function well as it is centrally managed and scales with application growth.

Edge systems, also, are not standalone embedded systems and microcontrollers. We have had embedded machines running everything from televisions to toasters for decades. What these embedded machines and computers lacked was connectivity. Edge systems require communication in multiple directions.

Edge computing is also not another term for the Internet of Things (IoT). While most IoT devices are edge computers since they are connected and intelligent, not all edge computing systems are IoT devices. An example of this is Netflix edge computing to aid in streaming video. These are public cloud servers that are found close to the consumer. IoT devices are smaller and technically simpler than edge computers.

Edge Architecture

We should set the record early in this book: edge computers and edge computing are simply the same computers and services that would see in a cloud datacenter or in an embedded device. The difference between traditional computing and edge computing is the function they serve and where they live. There are also constraints within the edge that need to be reckoned with.

There are several viewpoints that also define edge computing as “any form of computing outside of a datacenter”. This is broad and could include even the smallest embedded systems in a vacuum cleaner or a microwave. Others claim edge computing to be a form of computing that reduces latency. It can be argued that latency is just one problem resolved by edge devices. Furthermore, depending on the distance and number of hops to a traditional datacenter, latency may not be an issue.

Our definition of edge computing is a form of computing services and hardware that provides processing, storage and programmable control close to the sources of data generation or data consumption. Edge computers may be extensions of applications and services hosted in a datacenter, but they do not necessarily need to be. One aspect of all edge systems is that they are connected by networks to relay data and for management. Therefore, simple embedded microcontrollers found in the bulk of machinery are not considered edge computers.

A typical edge topology is shown in following figure 2 below. The cloud provides the central authority for services like data warehousing and archiving, control and administration of the collective system, and for overall security and provisioning. We show three edge nodes that in in three disparate locations. Typically, edge systems will be associated with a location while the cloud has no physical presence.

Figure 2: Cloud to Edge Overall Architecture

Another key aspect in the figure is the communication between the edge nodes and the cloud. We call this “North-South” traffic. We speak of “North-bound traffic as originating from the edge and propagating to the cloud. “South-bound” traffic is the reverse where traffic originates in the cloud and moves to the edge. Communication between the edge and cloud can be raw data for archival or exchange of security keys. The other form of traffic that is important in edge architectures is “East-West” traffic. This is data exchange between various edge nodes with each other and do not directly involve communication with the cloud. Think of this communication between two edge computers managing an industrial production line. They communicate by providing time-critical data between a machine in final assembly and a machine in manufacturing to alert each other if there is a disruption in service or if manufacturing should ramp down in case of decrease demand.

What will you learn from this book?

text goes here

1. You will develop an understanding of hardware and software stacks of edge computing and IoT systems.
2. You will be introduced to networking and communications technologies which make up the heart of edge computing. This includes PAN, WAN, AND WLAN systems and architectures.
3. You will learn of the broad nature of edge security risks and how to prevent embarrassing and crippling cyberattacks.>
4. You will be presented with a myriad of comprehensive real-world use cases both in consumer and enterprise applications of edge computing.
5. You will be instructed on monetizing edge systems for commercial viability.

Summary

text goes here

• Edge computing is a fast-growing technology and research area that is finding its application in a number of industries.
• We have established that edge computing serves a viable function when there are constraints that traditional cloud-based infrastructure cannot remediate, nor can an information device exclusively reside without larger cloud connectivity.
• If a system is constraints by too much added latency in migrating data to and from the cloud or if the cost of that data movement is too high an edge system has a purpose.
• Additionally, relying exclusively on a cloud system can add risk if a system needs to resilient.
• Bridging networks and communications systems is also a common trait of edge machinery by providing the ability for non-typical networks like Bluetooth to communicate through the Internet.
• Edge devices also extend the cloud to devices that normally have never been connected to the Internet.
• Edge devices can filter data the customer does not want to migrate to a public cloud.

If you want to learn more about the book, check it out here.