Linkerd: Up and Running

Chapter 5. Ingress and Linkerd

Whenever you work with Kubernetes, you always have to find a way for your users outside your cluster to be able to make requests of (some of) the services running inside your cluster. This is the ingress problem (see Figure 5-1): the cluster wants to protect everything inside from the big scary Internet, but that’s where your legitimate users are.

Figure 5-1. The ingress problem

There’s an entire class of applications out there, unsurprisingly called ingress controllers, whose sole purpose is solving the ingress problem. Linkerd does not include an ingress controller; instead, it allows you to mesh whatever ingress controller you like, as long as certain rules are followed. In this chapter, you’ll learn how to make Linkerd and the ingress controller of your choice play nicely with each other.

There are a lot of different ingress controllers, which approach the ingress problem in fascinatingly different ways. However, there are some common threads across all of them, shown in Figure 5-2.

Figure 5-2. Ingress controller high-level architecture

These common threads include:

• They are all designed to live right at the edge of a cluster (usually behind a Kubernetes Service of type LoadBalancer), exposed directly to the Internet so that their clients ...