用户手册 — IBM REDP-4285-00

目录

• Front cover

  1
• Contents

  5
• Notices

  9
  • Trademarks

    10
• Preface

  11
  • How this Redpaper is structured

    11
  • The team that wrote this Redpaper

    12
  • Become a published author

    13
  • Comments welcome

    13
• Chapter 1. Understanding the Linux operating system

  15
  • 1.1 Linux process management

    17
    • 1.1.1 What is a process?

      17
    • 1.1.2 Lifecycle of a process

      18
    • 1.1.3 Thread

      18
    • 1.1.4 Process priority and nice level

      19
    • 1.1.5 Context switching

      20
    • 1.1.6 Interrupt handling

      20
    • 1.1.7 Process state

      21
    • 1.1.8 Process memory segments

      22
    • 1.1.9 Linux CPU scheduler

      23
  • 1.2 Linux memory architecture

    25
    • 1.2.1 Physical and virtual memory

      25
    • 1.2.2 Virtual memory manager

      27
  • 1.3 Linux file systems

    29
    • 1.3.1 Virtual file system

      29
    • 1.3.2 Journaling

      30
    • 1.3.3 Ext2

      31
    • 1.3.4 Ext3

      32
    • 1.3.5 ReiserFS

      33
    • 1.3.6 Journal File System

      33
    • 1.3.7 XFS

      33
  • 1.4 Disk I/O subsystem

    33
    • 1.4.1 I/O subsystem architecture

      34
    • 1.4.2 Cache

      35
    • 1.4.3 Block layer

      37
    • 1.4.4 I/O device driver

      38
    • 1.4.5 RAID and Storage system

      39
  • 1.5 Network subsystem

    40
    • 1.5.1 Networking implementation

      40
    • 1.5.2 TCP/IP

      44
    • 1.5.3 Offload

      47
    • 1.5.4 Bonding module

      48
  • 1.6 Understanding Linux performance metrics

    48
    • 1.6.1 Processor metrics

      48
    • 1.6.2 Memory metrics

      49
    • 1.6.3 Network interface metrics

      50
    • 1.6.4 Block device metrics

      50
• Chapter 2. Monitoring and benchmark tools

  53
  • 2.1 Introduction

    54
  • 2.2 Overview of tool function

    54
  • 2.3 Monitoring tools

    55
    • 2.3.1 top

      55
    • 2.3.2 vmstat

      56
    • 2.3.3 uptime

      57
    • 2.3.4 ps and pstree

      58
    • 2.3.5 free

      60
    • 2.3.6 iostat

      62
    • 2.3.7 sar

      64
    • 2.3.8 mpstat

      65
    • 2.3.9 numastat

      66
    • 2.3.10 pmap

      66
    • 2.3.11 netstat

      67
    • 2.3.12 iptraf

      68
    • 2.3.13 tcpdump / ethereal

      69
    • 2.3.14 nmon

      72
    • 2.3.15 strace

      73
    • 2.3.16 Proc file system

      74
    • 2.3.17 KDE System Guard

      76
    • 2.3.18 Gnome System Monitor

      81
    • 2.3.19 Capacity Manager

      81
  • 2.4 Benchmark tools

    84
    • 2.4.1 LMbench

      85
    • 2.4.2 IOzone

      86
    • 2.4.3 netperf

      87
    • 2.4.4 Other useful tools

      90
• Chapter 3. Analyzing performance bottlenecks

  91
  • 3.1 Identifying bottlenecks

    92
    • 3.1.1 Gathering information

      92
    • 3.1.2 Analyzing the server’s performance

      94
  • 3.2 CPU bottlenecks

    95
    • 3.2.1 Finding CPU bottlenecks

      95
    • 3.2.2 SMP

      95
    • 3.2.3 Performance tuning options

      96
  • 3.3 Memory bottlenecks

    96
    • 3.3.1 Finding memory bottlenecks

      96
    • 3.3.2 Performance tuning options

      98
  • 3.4 Disk bottlenecks

    98
    • 3.4.1 Finding disk bottlenecks

      98
    • 3.4.2 Performance tuning options

      101
  • 3.5 Network bottlenecks

    101
    • 3.5.1 Finding network bottlenecks

      101
    • 3.5.2 Performance tuning options

      103
• Chapter 4. Tuning the operating system

  105
  • 4.1 Tuning principals

    106
    • 4.1.1 Change management

      106
  • 4.2 Installation considerations

    106
    • 4.2.1 Installation

      106
    • 4.2.2 Check the current configuration

      108
    • 4.2.3 Minimize resource use

      111
    • 4.2.4 SELinux

      116
    • 4.2.5 Compiling the kernel

      118
  • 4.3 Changing kernel parameters

    118
    • 4.3.1 Where the parameters are stored

      121
    • 4.3.2 Using the sysctl command

      122
  • 4.4 Tuning the processor subsystem

    122
    • 4.4.1 Tuning process priority

      123
    • 4.4.2 CPU affinity for interrupt handling

      123
    • 4.4.3 Considerations for NUMA systems

      123
  • 4.5 Tuning the vm subsystem

    124
    • 4.5.1 Setting kernel swap and pdflush behavior

      124
    • 4.5.2 Swap partition

      125
    • 4.5.3 HugeTLBfs

      126
  • 4.6 Tuning the disk subsystem

    127
    • 4.6.1 Hardware considerations before installing Linux

      128
    • 4.6.2 I/O elevator tuning and selection

      130
    • 4.6.3 File system selection and tuning

      135
  • 4.7 Tuning the network subsystem

    139
    • 4.7.1 Considerations of traffic characteristics

      139
    • 4.7.2 Speed and duplexing

      140
    • 4.7.3 MTU size

      141
    • 4.7.4 Increasing network buffers

      141
    • 4.7.5 Additional TCP/IP tuning

      143
    • 4.7.6 Performance impact of Netfilter

      147
    • 4.7.7 Offload configuration

      148
    • 4.7.8 Increasing the packet queues

      150
    • 4.7.9 Increasing the transmit queue length

      150
    • 4.7.10 Decreasing interrupts

      150
• Appendix A. Testing configurations

  153
  • Hardware and software configurations

    154
  • Linux installed on guest IBM z/VM systems

    154
  • Linux installed on IBM System x servers

    154
• Abbreviations and acronyms

  157
• Related publications

  159
  • IBM Redbooks

    159
  • Other publications

    159
  • Online resources

    159
  • How to get IBM Redbooks

    161
  • Help from IBM

    161
• Index

  163
• Back cover

  170