Book Image

Edge Computing Systems with Kubernetes

By : Sergio Méndez
Book Image

Edge Computing Systems with Kubernetes

By: Sergio Méndez

Overview of this book

Edge computing is a way of processing information near the source of data instead of processing it on data centers in the cloud. In this way, edge computing can reduce latency when data is processed, improving the user experience on real-time data visualization for your applications. Using K3s, a light-weight Kubernetes and k3OS, a K3s-based Linux distribution along with other open source cloud native technologies, you can build reliable edge computing systems without spending a lot of money. In this book, you will learn how to design edge computing systems with containers and edge devices using sensors, GPS modules, WiFi, LoRa communication and so on. You will also get to grips with different use cases and examples covered in this book, how to solve common use cases for edge computing such as updating your applications using GitOps, reading data from sensors and storing it on SQL and NoSQL databases. Later chapters will show you how to connect hardware to your edge clusters, predict using machine learning, and analyze images with computer vision. All the examples and use cases in this book are designed to run on devices using 64-bit ARM processors, using Raspberry Pi devices as an example. By the end of this book, you will be able to use the content of these chapters as small pieces to create your own edge computing system.
Table of Contents (21 chapters)
1
Part 1: Edge Computing Basics
7
Part 2: Cloud Native Applications at the Edge
13
Part 3: Edge Computing Use Cases in Practice

Creating K3s single and multi-node clusters

In this section, you are going to learn how to configure K3s master and agent nodes on your Ubuntu OS for your ARM devices. To visualize what we are doing, let's take a closer look at Figure 2.10:

Figure 2.10 – The K3s cluster configurations

The preceding diagram shows that you can install a K3s cluster in the following configurations:

  • Single node cluster: In this configuration, you only have one node that assumes the role of a master and agent/worker node at the same time. You can use this type of cluster for small applications. This is not ideal for heavy workloads, as it can slow down all the components. Remember that this node works as a master and an agent at the same time.
  • Multi-node cluster: In this configuration, you have a master node that controls the agent/worker nodes; this configuration will be useful for high availability and heavy processing tasks.

With these brief descriptions...