Native Docker Clustering with Swarm

By Fabrizio Soppelsa and Chanwit Kaewkasi

If you are a Linux admin or a Docker user who wants to natively manage Docker clusters, then this is the book for you.

• Language: English
• Publisher: Packt Publishing
• Release date: Dec 20, 2016
• ISBN: 9781786467607

Book preview

Table of Contents

Native Docker Clustering with Swarm

Credits

About the Authors

About the Reviewer

www.PacktPub.com

Why subscribe?

Dedication

Preface

What this book covers

What you need for this book

Who this book is for

Conventions

Reader feedback

Customer support

Errata

Piracy

Questions

1. Welcome to Docker Swarm

Clustering tools and container managers

Swarm goals

Why use Swarm

Real world use case examples

Pet versus cattle models

Swarm features

Similar projects

Kubernetes

CoreOS Fleet

Apache Mesos

Kubernetes versus Fleet versus Mesos

Swarm versus all

The Swarm v1 architecture

Terminology

Getting started with Swarm

Docker for Mac

Docker for Windows

Getting ready with Linux

Check that Docker Machine is available - all systems

Swarm, yesterday

Boot2Docker

Create 4 cluster nodes with Docker Machine

Configuring the Docker hosts

Starting Docker Swarm

Test your Swarm cluster

Swarm, today

Summary

2. Discover the Discovery Services

A discovery service

Token

Re-architecting the example of Chapter 1 with token

Token limitations

Raft

Raft theory

Raft in practice

Etcd

Re architecting the example of Chapter 1 with Etcd

ZooKeeper

Consul

Re architecting the example of Chapter 1 with Consul

Towards a decentralized discovery service

Summary

3. Meeting Docker Swarm Mode

Swarmkit

Versions and support

Swarmkit architecture

How a manager chooses the best node for a task

The heart of SwarmKit: swarmd

The controller of SwarmKit: swarmctl

Provisioning a SwarmKit cluster with Ansible

Creating a service on SwarmKit

Swarm mode

Old versus new Swarm versus SwarmKit

Swarm Mode zoom in

Integration into the engine

docker swarm command

docker node

docker service

Docker Stack

Etcd's Raft is integrated already

Load balancing and DNS

Promotion and demotion

Replicas and scale

Services and tasks

docker service scale

Summary

4. Creating a Production-Grade Swarm

Tools

An HA Topology for Swarm2k

Managers specifications

In case of Raft recovery

Raft files

Running tasks

Manager topologies

Provisioning the infrastructure with belt

Securing Managers with Docker Machine

Understanding some Swarm internals

Joining workers

Upgrading Managers

Live Upgrading the Managers

Monitoring Swarm2k

InfluxDB Time-Series Database

Telegraf Swarm plugin

Swarm3k

Swarm3k Setup and Workload

Swarm performance at a scale

Swarm2k and Swarm3k lessons learned

Summary

5. Administer a Swarm Cluster

Docker Swarm standalone

Docker Swarm Mode

Manually adding nodes

Managers

Workers number

Scripted nodes addition

Belt

Use Ansible

Cluster management

Nodes operations

Demotion and promotion

Tagging nodes

Remove nodes

Remove workers

Remove managers

Swarm health

Backing up the cluster configuration

Disaster recovery

Graphical interfaces for Swarm

Shipyard

Portainer

Summary

6. Deploy Real Applications on Swarm

Microservices

Deploy a replicated nginx

A minimal Swarm

Docker service

Overlay networks

Integrated load balancing

Connecting services: A WordPress example

Swarm scheduling strategies

Now, WordPress

Docker Compose and Swarm mode

Introducing Docker stacks

Distributed Application Bundles

Docker deploy

Another app: Apache Spark

Why Spark on Docker

Spark standalone without Swarm

Spark standalone on Swarm

Spark topology

Storage

Prerequisites

Start Spark on Swarm

Summary

7. Scaling Up Your Platform

The Spark example, again

Docker plugins

The lab

A unique key

Docker Machine

Security groups

Networking configuration

Storage configuration and architecture

Installing Flocker

Generating Flocker certificates

Installing software

Installing the control node

Installing the cluster nodes

Testing whether everything is up and running

Installing and configuring Swarm

A volume for Spark

Deploying Spark, again

Testing Spark

Using Flocker storage

Scaling Spark

Monitoring Swarm hosting apps

Prometheus

Installing a monitoring system

Importing Prometheus in Grafana

Summary

8. Exploring Additional Features of Swarm

Libnetwork

Networking plugins

Container Networking Model

Encryption and routing mesh

MacVLAN

Overlay networks

Network Control Plane

Libkv

How to use libkv

Summary

9. Securing a Swarm Cluster and the Docker Software Supply Chain

Software Supply Chain

Securing Swarm cluster

Securing a Swarm: Best practices

Certification Authorities

Certificates and Mutual TLS

The join token

Add TLS with Docker Machine

Form a cluster on a private network

Docker Notary

Introducing Docker secrets

Summary

10. Swarm and the Cloud

Docker for AWS and Azure

Docker for AWS

Docker for Azure

Docker Datacenter

Swarm on OpenStack

OpenStack Nova

The (deprecated) nova-docker driver

The reality - OpenStack the friendly way

OpenStack Heat

OpenStack Magnum

Architecture and core concepts

Install HA Magnum on Mirantis OpenStack

Configure an HA Magnum installation

Create a Swarm cluster on Magnum

Summary

11. What is next?

The challenge of provisioning

Software defined infrastructure

Infrakit

TUF - The Update Framework

Docker stacks and Compose

CaaS - Containers as a Service

Unikernels

Contribute to Docker

Github

File issues

Code

Belt and other projects

Summary

Native Docker Clustering with Swarm

---

Native Docker Clustering with Swarm

Copyright © 2016 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the authors, nor Packt Publishing, and its dealers and distributors will be held liable for any damages caused or alleged to be caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information.

First published: December 2016

Production reference: 1151216

Published by Packt Publishing Ltd.

Livery Place

35 Livery Street

Birmingham 

B3 2PB, UK.

ISBN 978-1-78646-975-5

www.packtpub.com

Credits

About the Authors

Fabrizio Soppelsa works as an Escalations Engineer for Mirantis Inc., the OpenStack company. Docker activist and advocate since Docker 0.3 and author of several articles on the Docker tools in three languages, he is also a concrete contributor to the projects, especially Machine. He currently lives in Moscow, Russia, where he is the Docker Meetup organizer with his spider Mosha.

I would like to thank ClusterHQ folks for their help on Flocker, especially, Ryan Wallner. the Yandex team and Denis Kutin for having given me a free OpenStack lab on which I could easily access. Mirantis for creating the (I think) best OpenStack distribution. The Docker team and the Docker community, for all the fun.

Chanwit Kaewkasi is an Assistant Professor at the School of Computer Engineering, Suranaree University of Technology, Thailand. Chanwit started contributing to the Docker Swarm project since 0.1 where he co-designed and implemented the strategy filters, the ZooKeeper discovery, and other features. He currently serves as Docker Swarm Maintainer and Docker Captain.

I'd like to thank my wife, Pitchaya, for her encouragement and best support to everything I'm working on, including this book.

Special thanks to the Docker Engineering team for their great software, Suranaree University of Technlogy for being my great workplace, and my parents for their support and lastly, Fabrizio for inviting me to co-author this great book.

About the Reviewer

Baohua Yang is a senior researcher at IBM. His interests include key areas in Cloud Computing, Fintech, distributed system and analytics, especially on emerging technologies, e.g., SDN/NFV, Container, BigData, Blockchain and cognitive computing.

As the chief architect, he has lead the architecture design and system implementation of enterprise productions, and helps solve key technical challenges for industrial solutions.

As an Open-Source contributor, he has submitted code, proposals, and presentations to numbers of projects, e.g., OpenStack, Hyperledger, OpenvSwitch, Docker, OpenDaylight, Kubernetes, and has lead several projects including easyOVS, Hyperledger Fabric-SDK-py and Cello. He is now the chairperson of Hyperledger Technical Working Group in China.

He has published 10s of papers in top-quality international conference and journals (e.g., IEEE INFOCOM, IEEE Trans on Computers), with several technical books and patents. He is now TPC member of numbers of academic conferences and journals.

His homepage is at https://yeasy.github.com.

www.PacktPub.com

For support files and downloads related to your book, please visit www.PacktPub.com.

Did you know that Packt offers eBook versions of every book published, with PDF and ePub files available? You can upgrade to the eBook version at www.PacktPub.com and as a print book customer, you are entitled to a discount on the eBook copy. Get in touch with us at [e-mail] for more details.

At www.PacktPub.com, you can also read a collection of free technical articles, sign up for a range of free newsletters and receive exclusive discounts and offers on Packt books and eBooks.

https://www.packtpub.com/mapt

Get the most in-demand software skills with Mapt. Mapt gives you full access to all Packt books and video courses, as well as industry-leading tools to help you plan your personal development and advance your career.

Why subscribe?

Fully searchable across every book published by Packt

Copy and paste, print, and bookmark content

On demand and accessible via a web browser

Dedication

I want to dedicate this book to the memory of my father, Donato.

Preface

Welcome to Native Docker clustering with Swarm! This is a book about containers and distributed systems. We’ll show how to use the native Docker tools to model micro services, spawn tasks, scale up the size of your application, and push your containers to the Docker clustering limit! In a word, we’ll discuss Docker orchestration.

With the recent rise of Swarm Mode and the enablement of Swarm inside the Docker Engine itself, it turned out that the best way to orchestrate Docker is… Docker!

Good, but what do we mean by orchestrate Docker? What is orchestration? Better, what is an orchestra?

An orchestra is an ensemble of musicians led by a conductor, who dictates tempo, rhythm and shapes the sound. Strings, woodwinds, percussions, keyboards and other instruments follow the conductor’s direction to perform an astonishing symphony, for example Beethoven’s The Ninth.

Similarly, in a containers orchestration system the musicians are tasks, and the conductor is a leader service (Swarm primitives). Tasks don’t play a symphony, or at least not only: More abstractly they execute some computational job, for example they run a webserver. The conductor, Swarm, is responsible for their provisioning, their availability, their linkage, their scaling. This (and more) is what we mean by Docker orchestration.

This book shows how to provision such Docker orchestras, how to guarantee the availability of the service, how to connect tasks and how to scale the platform, to play the symphony of your application.

What this book covers

Chapter 1, Welcome to Docker Swarm, introduces Swarm, and explains why you need a clustering solution for your containers. It illustrates the Swarm features, giving a high-level description of its architecture. We define some use cases and describe how Swarm is different from Fleet, Kubernetes and Mesos. The chapter proceeds with the Docker tools installation and finally with two Swarms provisionings: A local Swarm Standalone and a remote Swarm Mode cluster on DigitalOcean.

Chapter 2, Discover the Discovery Services, is a descriptive and mostly abstract chapter. We’ll learn what discovery mechanisms and consensus algorithms are, and why they are crucial for distributed systems. We’ll describe in detail Raft and its implementation Etcd, the consensus mechanism included in Swarm Mode. We will also show the limitations of the discovery mechanism used in Chapter 1, Welcome to Docker Swarm, by extending the local tiny example with Consul, re deploying it.

Chapter 3, Meeting Docker Swarm Mode, is about the new Docker kit that allows to create task clusters of any size. We will introduce Swarmit, the foundation of Docker Swarm Mode, showing how it works in Docker 1.12+, discuss its architecture, its concepts, how it’s different from the old Swarm, and how it organizes workloads by abstracting services and tasks.

Chapter 4, Creating a Production-Grade Swarm, shows and discusses the community-driven projects Swarm2k and Swarm3k, our 2,300 and 4,800 nodes Swarm clusters experiments, which ran hundreds of thousands of containers. We demonstrate how such huge clusters were planned, provisioned, and summarize the lessons we learned.

Chapter 5, Administer a Swarm Cluster, is a chapter about infrastructure. We will show how to increase or decrease Swarms sizes, how to promote and demote nodes, and how to update clusters and nodes properties. We’ll introduce Shipyard and Portainer.io as graphical UIs for Swarm.

Chapter 6, Deploy Real Applications on Swarm, is where we will put real applications in motion on Swarm and where we add to the discussion some notes about Compose, Docker Stacks and Docker Application Bundles. We will show the typical deployment workflow, how to filter and schedule containers over the cluster, launch them as services, handle containers as tasks. We’ll start defining a web service with Nginx, then we’ll deploy a mandatory Wordpress with MySQL example. We’ll finally move on with a more realistic app: Apache Spark.

Chapter 7, Scale Up Your Platform, will develop new topics from the previous chapter. Here we’ll introduce Flocker to add storage capacity to Spark on Swarm, and we’ll show how to install and use it automatically at a scale in conjunction with Swarm. We’ll refine our Spark example by running some real big data jobs and setting up a basic monitoring system for this infrastructure.

Chapter 8, Exploring Additional Features to Swarm, discusses some advanced topics important to Swarm, such as Libnetwork and Libkv.

Chapter 9, Securing a Swarm Cluster and Docker Software Supply Chain, will focus on security considerations for Swarm clusters. Among the arguments, certificates, firewalling concepts for platform, and a mention to Notary.

Chapter 10, Swarm and the Cloud, is a chapter illustrating the most popular options for running Swarm on cloud providers. We’ll install Swarm on AWS and Azure before introducing the Docker Datacenter, and we’ll finally move to OpenStack showing how to install and manage Swarms on the top of Magnum, the Container as a Service solution for OpenStack.

Chapter 11, What is Next?, concludes the discussion with an overview on the next Docker orchestration trends, such as software defined infrastructures, Infrakit, unikernels, Containers as a Service. The adventure continues!

What you need for this book

We assume the reader to have some experience using Docker from the command line: throughout the book we’ll continuously pull images, run containers, define services, expose ports and create networks.

Also, the ideal reader possesses a basic understanding of networking protocols and is familiar with public and private cloud concepts like virtual machines and tenant networks.

To follow the examples in the text, you will need Docker and its tools. Chapter 1, Welcome to Docker Swarm, covers their installation.

Also, to get the most from the examples, you will need access to a a public (for example AWS, Azure or DigitalOcean) or private (for example OpenStack) cloud to instantiate Virtual