How Docker Swarm Container Networking Works
Docker 1.12 is a release loaded with a lot of great features. With built-in orchestration and by removing dependencies on the external KV store, Docker Swarm allows DevOps to quickly deploy a multi-host docker cluster that “just works.” Although not without controversies, when compared to Kubernetes, Docker Swarm’s ease-of-use is one of it’s most cited advantages. Perhaps in response, Kubernetes simplified its cluster bootstrapping process with the introduction of kubeadm in its 1.4 release.
In this post on Swarm container networking I’ll assume you have mastered the basics of creating a swarm and starting services. I won’t be covering general concepts of managing swarm nodes and services. Instead, the focus will be on explaining how Docker uses various Linux tools to virtualize multi-host overlay networks.
First, let’s create an overlay network and deploy our containers. Our docker swarm cluster has two nodes. I will create three services, each has one running instance.
Read the entire article here, How Docker Swarm Container Networking Works