Book Image

Native Docker Clustering with Swarm

By : Fabrizio Soppelsa, Chanwit Kaewkasi
Book Image

Native Docker Clustering with Swarm

By: Fabrizio Soppelsa, Chanwit Kaewkasi

Overview of this book

Docker Swarm serves as one of the crucial components of the Docker ecosystem and offers a native solution for you to orchestrate containers. It’s turning out to be one of the preferred choices for Docker clustering thanks to its recent improvements. This book covers Swarm, Swarm Mode, and SwarmKit. It gives you a guided tour on how Swarm works and how to work with Swarm. It describes how to set up local test installations and then moves to huge distributed infrastructures. You will be shown how Swarm works internally, what’s new in Swarmkit, how to automate big Swarm deployments, and how to configure and operate a Swarm cluster on the public and private cloud. This book will teach you how to meet the challenge of deploying massive production-ready applications and a huge number of containers on Swarm. You'll also cover advanced topics that include volumes, scheduling, a Libnetwork deep dive, security, and platform scalability.
Table of Contents (18 chapters)
Native Docker Clustering with Swarm
Credits
About the Authors
About the Reviewer
www.PacktPub.com
Dedication
Preface

Installing and configuring Swarm


We can now install a Swarm with our favorite method, as shown in the previous chapters. We'll have aws-101 to aws-103 as managers, and the rest of nodes except aws-104, workers. This cluster can be expanded even further. For practical things, we'll keep it at 10-nodes size.

We now add a dedicated spark overlay VxLAN network:

docker network create --driver overlay --subnet 10.0.0.0/24 spark

A volume for Spark

We now connect to any Docker host and create a 75G sized volume to be used to save some persistent Spark data:

docker volume create -d flocker -o size=75G -o profile=bronze --
    name=spark

The option to discuss here is profile. This is a sort of flavor of storage (speed, mostly). As explained in the link https://docs.clusterhq.com/en/latest/flocker-features/aws-configuration.html#aws-dataset-backend, ClusterHQ maintains three available profiles for AWS EBS:

  • Gold: EBS Provisioned IOPS / API named io1. Configured for maximum IOPS for its size - 30 IOPS...