Linux Clustering: Building and Maintaining Linux Clusters - Softcover

Bookman, Charles

3

8 ratings by Goodreads

9781578702749: Linux Clustering: Building and Maintaining Linux Clusters

Softcover

ISBN 10: 1578702747 ISBN 13: 9781578702749

Publisher: Sams, 2002

View all copies of this ISBN edition:

Linux Clustering: Building and Maintaining Linux Clusters provides Linux users with information about building their own Linux cluster from the ground up. It gives best practices, helpful hints, and guidelines about building one server or hundreds of servers at a level that administrators at any experience level can understand.

From installation of the air conditioning and power in the data center, to alternative file systems to the final production run, this book provides you with everything that you need to know. Linux Clustering: Building and Maintaining Linux Clusters walks you through the initial design and selection of the best possible types of clusters, as well as covering monitoring tools and providing for disaster recovery. Not only does this book provide information on parallel and Beowulf type clusters, Charles Bookman goes into depth on high availability clusters, load balancing, and provides advice for writing your own distributing applications as well. Incorporating best practices and cutting-edge approaches, Bookman provides step-by-step and tried-and-true methods of bringing up a Linux cluster to production level.

"synopsis" may belong to another edition of this title.

About the Author:

About the AuthorAbout the Author Charles Bookman fell in love with the personal computer when he was 12. He would stay after school to teach himself programming on the Commodore Pet. His experiences with Linux started during the 2.0.30 days, when he took the work computer at his counseling job and converted it into something that nobody there had ever seen. After they fired him, he knew that he was on to something, and has been a Linux evangelist ever since. Charles' hobbies include being up to no good, using sarcasm, listening to music, being snobbish about art, playing bass and guitar, and painting and drawing when he finds the time. He'll kick your butt at table tennis, and is learning a great game of pool as well. Charles currently works at the University of the Pacific as a UNIX systems administrator, maintaining its high-performance cluster. He runs a design company, and consults in his spare time. � Copyright Pearson Education. All rights reserved.

Introduction

Introduction

Linux has been increasing in popularity ever since its inception. With theexplosive growth of the open-source movement and the viability of what it can dowith an alternative operating system, Linux has grown into a mature, stableenvironment in which to host mission-critical applications and research.

It's this very nature of Linux and its open-source code that makes itattractive for hobbyists, educators, and research environments. Thousands ofprograms are available to enhance the user experience and allow the seriousadministrator to enhance or create his environment with Linux. It's thisflexibility and maturity of the operating system that allows people to easilycluster their Linux machines.

Having two or more machines act in concert with one another isn't rocketscience; in fact, anyone with any motivation and a skill for reading a book canmake computers do wondrous things. From load balancing to developing a thousandnode parallel computer, Linux is more than able to handle the task.

This book takes you through the construction of several types of clusteringby using Linux, and explains step by step how to get there. We'll coverhandy tips on the way, as well as explain why things work the way they do. In notime, you'll be able to create high availability and Beowulf type clusterswith the best of them.

Who Should Read This Book

This book is for Linux enthusiasts and users who want to get a Linux clusterup and running with the least amount of fuss. Administrators will find this bookparticularly helpful in learning the why's and how's of getting Linuxto do amazing things with more than one computer at your disposal.

This book will help if you need to introduce a large parallel or distributedcomputing cluster to your environment, and the ways that one can solve largecomputational problems using Linux. Methods to install, maintain, and back uplarge cluster farms are discussed. We'll also cover the software andhardware needed to create such a solution.

If you're looking for an enterprise-level load balancing orhigh-availability solution to your environment, Chapters 5 and 6 coverinstallation and configuration of Linux solutions to these issues. Not only doesthis book cover the basics behind these, but it also tells the best practicesand methods to get a solution up in the quickest amount of time. Softwareinstallation and configuration examples are given to bring readers up tospeed.

Who This Book Is Not For

This book assumes that the reader is comfortable with Linux. It does not leadthe user though basic tasks, such as installation of the operating system,adding users, or window manager configuration. More knowledge might need to begarnered to bring the user up to speed.

This book is also not for seasoned professionals who have a deepunderstanding of clustering. This book does not cover performance tuning or deepcluster theory. This book is specifically designed for those who need a clusterup, now. This book is not designed for those who already have a cluster plannedor running in their environment.

Overview

This book is divided into four parts that describe the planning andconfiguration of Linux clusters. Part I is designed as an introduction toplanning and configuring the environment that is needed to bring a cluster up tospeed in the shortest possible time. Part II is devoted to the installation andsupport of the various types of Linux clusters. Part III introduces programminglibraries and what to do when a cluster fails.

Part I is separated into four chapters. Chapter 1, "ClusteringFundamentals," briefly introduces Linux clustering in general, and why itmight be applicable to you or your environment. Chapter 2, "Preparing yourLinux Cluster," discusses how best to select the parts that are needed toinstall your cluster and the environment in which it's housed. Chapter 3,"Installing and Streamlining the Cluster," discusses best practicemethods on how to install large numbers of machines in the shortest and easiestamount of time. Chapter 4, "Alternative File Systems," discusses thelimitations of the current standard file system, ext2, and discusses othermethods to improve on it, and file system tricks.

Part II includes four chapters that deal with the clustering technologiesthemselves. Chapter 5, "High Availability and Fault-TolerantClusters," introduces high-availability systems, how they work, and howbest to implement the hardware and software. Chapter 6, "LoadBalancing," talks about the best ways that Linux can direct traffic tomultiple machines. Chapter 7, "Distributed Computing," introduces thedistributed model as a means to solve large computational problems. Chapter 8,"Parallel Computing," also expands on this model in a slightlydifferent approach. This chapter is skewed to parallel and commoditycomputers.

Part III deals with the care and feeding of a cluster. Chapter 9,"Programming a Parallel Cluster," references programming libraries andprograms that illustrate these libraries. Chapter 10, "ClusterManagement," discusses best practices to keep your cluster in check andmonitored for problems. The final chapter, Chapter 11, "Recovering WhenDisaster Strikes," deals primarily with disaster recovery.

Part IV contains reference material for the cluster designer. Appendix Aincludes cluster resources on the web and where to find more information.Appendix B is a reference for Kickstart options for Red Hat Linux. Using theseoptions, one can populate a cluster in no time with an unattended install.Appendix C details DHCP options in greater depth for the cluster, so that onecan maintain large clusters in a central location with ease. Finally, Appendix Ddetails class attributes for Condor, which is a distributed environmentcluster.

If you're familiar with clusters in general, it might behoove you toperuse Part I and move right on to Part II where the cluster you'reinterested in is discussed. You'll want to hit Part III right after becauseof the information that discusses keeping your applications and servers up.

Conventions

This book follows a few typographical conventions:

Italics indicate new terms the first time they're introduced, and are used for emphasis.
A special monospace font indicates code, a filename or pathname, or an Internet address.

"About this title" may belong to another edition of this title.