11. Build Your Own HA Cluster

Overview

In this chapter, we will learn how Kubernetes enables us to deploy infrastructure with remarkable resilience and how to set up a high-availability Kubernetes cluster in the AWS cloud. This chapter will help you understand what enables Kubernetes to be used for highly available deployments and, in turn, enable you to make the right choices while architecting a production environment for your use case. By the end of the chapter, you will be able to set up a suitable cluster infrastructure on AWS to support your highly available (HA) Kubernetes cluster. You will also be able to deploy an application in a production environment.

Introduction

In the previous chapters, you learned about application containerization, how Kubernetes works, and some of the "proper nouns" or "objects" in Kubernetes that allow you to create a declarative-style application architecture that Kubernetes will execute on your behalf.

Software and hardware instability are a reality in all environments. As applications need higher and higher availability, shortcomings in the infrastructure become more obvious. Kubernetes was purpose-built to help solve this challenge for containerized applications. But what about Kubernetes itself? As cluster operators, do we shift from watching our individual servers like hawks to watching our single Kubernetes control infrastructure?

As it turns out, this aspect was one of the design considerations for Kubernetes. One of the design goals of Kubernetes is to be able to withstand instability in its own infrastructure. This means that when set up properly, the Kubernetes control...

How the Components of Kubernetes Work Together to Achieve High Availability

You have learned in Chapter 2, An Overview of Kubernetes, how the pieces of Kubernetes work together to provide a runtime for your application containers. But we need to investigate deeper how these components work together to achieve high availability. To do that, we'll start with the memory bank of Kubernetes, otherwise known as etcd.

etcd

As you have learned in earlier chapters, etcd is the place where all Kubernetes configuration is stored. This makes it arguably the single most important component of the cluster since changes in etcd affect the state of everything. More specifically, any change to a key-value pair in etcd will cause the other components of Kubernetes to react to this change, which could mean disruptions to your application. In order to achieve high availability for Kubernetes, it is wise to have more than one etcd node.

But many more challenges arise when you add multiple...

Building a High-Availability Focused Kubernetes Cluster

Hopefully, by reading the previous section, you're starting to realize that Kubernetes is less magical than it may seem when you first approached the topic. It is an extremely powerful tool on its own, but Kubernetes really shines when we take full advantage of its capability of running in a highly available configuration. So now we're going to see how to implement it and actually build a cluster using a cluster life cycle management tool. But before we do that, we need to know the different ways that we can deploy and manage a Kubernetes cluster.

Self-Managed versus Vendor-Managed Kubernetes Solutions

Amazon Web Services, Google Cloud Platform, Microsoft Azure, and practically every other major cloud services provider has a managed Kubernetes offering. So, when you are deciding how you are going to build and run your cluster, you should consider some of the different managed providers and their strategic offerings...

Summary

Highly available infrastructure is one of the key components to achieving high availability for applications. Kubernetes is an extremely well-designed tool and has many built-in resiliency features that make it able to withstand major networking and compute events. It works to keep those events from impacting your application. During our exploration of high-availability systems, we investigated some components of Kubernetes and how they work together to achieve high availability. Then, we constructed a cluster of our own on AWS that was designed to be highly available using the kops cluster life cycle management tool.

In the next chapter, we're going to take a look at how we make our applications more resilient by leveraging Kubernetes primitives to ensure high availability.