Preface

The Linux (open source) operating system has benefited from a foundation of less expensive hardware, stability, and technical developments. The most popular use of Linux is as a web server. Today, it is doing a host of corporate services such as email, firewall, proxy, gateway, database, applications, file server, printing, and more. Linux is now poised to become a competitive operating system on the higher-end servers of the enterprise world. Major Wall Street firms, movie production companies, and many other businesses are now turning to Linux for their enterprise server needs.

As more servers are being deployed, their performance becomes a critical factor in the efficiency of the overall system and affects all users. Although replacing the entire server with a newer and faster one may be an alternative, it is often more appropriate to only replace or add to those components that need it, leaving the rest alone. Often, poor performance is due to bottlenecks in individual hardware subsystems, an incorrectly configured operating system, or a poorly tuned application.

The keys to improving performance are to understand what hardware and software configuration options are available, understand the performance tools and how to use them, and then analyze the results obtained from the tools so suitable changes can be made to positively affect the server.

This book provides the basic knowledge and skills needed to understand and improve the performance of Linux servers. It consists of several chapters written by Linux practitioners and, based upon their collective practical experience, describes the Linux performance monitoring, evaluation and measurement, analysis, and tuning of Linux servers. Also included is a discussion of the methodologies for improving and maximizing the performance of business server applications running on an Intel-based hardware platform and the Linux operating system.

The book uses the simple and progressive approach to performance tuning. It starts with a discussion on the Linux kernel followed by Linux servers, performance tools, and tuning techniques. It then takes the reader into the performance characterization of Linux servers followed by tuning examples that can be adapted for use in various situations.

The book focuses on two Linux distributions for illustration purpose: Red Hat Enterprise Linux and Novel SUSE Linux Enterprise Server. In addition, the Intel-based platform is assumed as the base server hardware. This is mainly the result of its popularity in the Linux community. The performance tuning of Linux servers on other platforms can be easily adapted. The server applications include web server, file and print server, database server, network management server, and web application server. The discussions and examples used in this book are based upon the Linux 2.4, 2.5, and 2.6 kernels.

The book consists of five parts: "Linux Overview," "Performance Analysis Tools," "System Tuning," "Performance Characterization of Linux Server Applications," and "Tuning Case Studies."

• Part 1, "Linux Overview," provides an overview of Linux installation, and the Linux kernel and server architectures.

• Part 2, "Performance Analysis Tools," provides a detailed review of performance tools in three areas: system performance and monitoring, system trace, and benchmarking.

• Part 3, "System Tuning," discusses performance tuning principles and strategy, as well as scheduler, memory, I/O, file system, and network and code tuning.

• Part 4, "Performance Characterization of Linux Server Applications," describes the performance characteristics of server applications, including factors that might affect their performance. It includes discussions on web, file, database, and application servers.

• Part 5, "Tuning Case Studies," provides several examples of how to integrate all the work in the previous sections to undertake a tuning project with different scenarios. Included are case studies on scheduling, I/O, file system, and network and commercial workload tuning.

• The book also includes an appendix which discusses many of the kernel tunable parameters and explains how to use the sysctl, /proc, and sysfs interfaces to tune the Linux kernel.