Before talking about Docker, it would be better to discuss the technology behind the software container.

Virtual machines are a common virtualization technology and have been widely adopted by cloud providers and enterprise companies. Actually, a software container (or container for short) is also a kind of virtualization technology, but there is something different about them. The key difference is that every container shares the same kernel on the host machine, while each virtual machine has its own kernel. Basically, a container uses virtualization techniques at the level of the operating system, not the hypervisor. The following diagram shows a comparison between container and VM stacks:

Linux's container technology heavily relies on two important kernel capabilities, namespace and...

Containers in the past were quite difficult to manage and use. Docker is basically a set of technologies to help us prepare, manage, and execute containers. In the world of virtual machines, we need a hypervisor to take care of all VM instances. Similarly, in the world of containers, we use Docker as the container engine to take care of everything to do with containers.

Undeniably, Docker is the most popular container engine to date. When using Docker, we follow the three concepts build, ship, and run, recommended by Docker itself:

• The workflow of Build-Ship-Run is optimized by the philosophy of Docker. In the Build step, we are allowed to build and destroy container images rapidly. As developers, we can include the container building steps as a part of our development cycle.
• In the Ship step, we ship container images to places, from our development...

A cluster is a group of machines connecting together to do work. A Docker host is a physical or virtual machine with the Docker Engine installed. We create a Docker Swarm cluster by connecting many Docker hosts together. We refer to each Docker host as a Docker Swarm node, simply a node.

In version 1.12, Docker introduced Swarm mode, a new orchestration engine to replace the old Swarm cluster, which is now referred to as Swarm classic. The main difference between Swarm classic and Swarm mode is that Swarm classic uses an external service, such as Consul, etcd, or Apache ZooKeeper as its key/value store, but Swarm mode has this key/value built in. With this, Swarm mode keeps orchestration latency at a minimum, and is more robust than Swarm classic because it does not need to interact with an external store. The monolithic nature of Swarm mode is...

To help you better remember and understand the concepts and practices of Docker described in this chapter, try answering the following questions without going back to the chapter's contents. Let's get started:

1. What are containers? What's the key difference between containers and virtual machines?
2. What are the main features inside the Linux kernel to enable container technology? Please name at least two of them.
3. What are the key concepts of the Docker workflow?
4. What is a Dockerfile for? Which Docker command do you use to interact with it?
5. What is the ENTRYPOINT instruction inside a Dockerfile?
6. Which command do we use to list all Docker images?

7. Which command do we use to form a Docker Swarm cluster?
8. What is the key difference between Swarm classic and Swarm mode?
9. Please explain the relationship between services and tasks.
10. How can we create an...

This chapter started off by discussing the concepts of containers. Then we reviewed what Docker is, how to install it, and the Docker build, ship, and run workflow. We then learnt how to form a Docker Swarm cluster and Swarm master and worker nodes. We learnt how to properly set up a robust Swarm cluster with an odd number of master nodes. We then learnt the service and task concepts of Docker Swarm. Finally, we learnt how Docker fits into serverless application development.

In the next chapter, we will review serverless frameworks and platforms to understand the overall architecture and the limitations of them.