Linux Performance Management Duration Audience Course Contents Course Objectives Instructional Technique Prerequisites Contents A. Performance Overview B. Performance Monitoring C. System Monitoring D. Process Management E. Memory Management F. Block I/O Performance G. Network Performance Management H. The Next Step

Linux Performance Management

Duration

This course is 3 days (24 hours). Customizing the topic list will affect the time requirement.

Audience

Performance management specialists and system administrators who are or will be responsible for the tuning and capacity monitoring of one or more machines running Linux.

Course Contents

Performance management requires an understanding of overall system architecture. This course takes the approach of tackling each major system of Linux individually, covering the necessary architectural background and then pointing out how resources are allocated, consumed, released, and reused.

1. Performance Overview
2. Performance Monitoring
3. System Monitoring
4. Process Management
5. Memory Management
6. Block I/O Performance
7. Network Performance Management

Course Objectives

Upon completion of this course, the student will be able to:

1. Understand the operating system subsystems and their interactions

2. Understand the basic performance trade-offs of resource allocation

3. Describe how requests for CPU time, disk space, and memory are handled

4. Understand the performance implications of kernel tunable parameters

5. Use performance diagnostic tools to monitor and plan for future upgrades

Instructional Technique

Students are invited to bring their current ideas and questions to the classroom for discussion. Case studies, lecture, and group problem solving will be used (online labs are difficult in this environment, although demonstrations can be performed if appropriate equipment access is available).

Prerequisites

Familiarity with Unix concepts such as the command line shell, pipes, I/O redirection, process ids, and paging space. All information required for a proper understanding of the course material will be provided; however, hands-on experience with the techniques listed above is required to fully master these topics.

Contents

A. Performance Overview

• Performance Basics

• Response Time vs. Throughput (or "You can't have it both ways")

• Processes and Threads

• Overall Performance Factors

• Metrics

• Caching vs. Buffering

• Architecture as a Factor

B. Performance Monitoring

• RedHat Monitoring Tools

• Other Optional Tools

• How is the /proc Filesystem Used?

• User-written Tools

• Kernel Tunable Parameters

C. System Monitoring

• Interactive Commands

  • top

  • gnome-system-monitor

  • htop

  • sar and isag

• Scripting Commands

  • sadc

  • vmstat

  • iostat

  • ifstat

  • netstat

• Programmer's Tools

  • lsof

  • size

  • ltrace

  • strace

D. Process Management

• Process Management Overview

• The Native POSIX Threads Library

• Resource Limits

• Process Lifecycle

  • The fork() Function Call

  • The exec() Function Call

  • The exit() Function Call

  • The wait() Function Call

• Time Slices

• Real-Time Threads

• CPU Affinity

• Task Scheduling

• The Nice Factor

• The O(1) Scheduler

E. Memory Management

• Virtual Memory Concepts

• Swapping vs. Paging

• Pageable Memory vs. Pinned Memory

• Process Address Space Management

• Address Translations

• Page Table Entries

• How Page Faults Are Resolved

• Memory Zones

• Subsystem Cache Allocations (SLAB and SLUB allocators)

• Shared Libraries

F. Block I/O Performance

• Overview of Linux I/O

• Filesystem vs. Block Device

• The ext3 and ext4 Filesystem Layout

• Measuring I/O Speeds

• I/O Schedulers

  • The Completely Fair Queueing Scheduler - cfq

  • The Deadline Scheduler - deadline

  • The Anticipatory Scheduler - anticipatory

• Tuning the ext3 and ext4 Filesystems

• Logical Volume Manager Tuning

G. Network Performance Management

• TCP/IP Layers

• Network Layers

• LAN Topology

• Socket Controls

• Network Services

• Network File System

  • UDP Optimizations

  • TCP Optimizations

H. The Next Step

• Techniques for Automatic Monitoring

• Post-processing Monitoring Data

• Planning Ahead