Linux iostat Command

Linux 命令大全Linux Command Encyclopedia

1. Overview of the iostat Command

iostat (Input/Output Statistics) is a powerful performance monitoring tool on Linux systems and is part of the sysstat package. It is mainly used to monitor system disk I/O activity and CPU usage.

1.1 Basic Functions

  • Monitor system disk I/O read/write speed
  • View CPU utilization
  • Summarize device load
  • Identify I/O performance bottlenecks

1.2 Typical Application Scenarios

  • Server performance tuning
  • Storage device performance analysis
  • System bottleneck troubleshooting
  • Capacity planning

2. Installation and Basic Usage

2.1 Installation Methods

Most Linux distributions do not install iostat by default; you need to install the sysstat package first:

Example

# Ubuntu/Debian
sudo apt-get install sysstat

# CentOS/RHEL
sudo yum install sysstat

# Fedora
sudo dnf install sysstat

2.2 Basic Command Format

Example

iostat [Options] [Interval] [Count]

Simple Example

Example

# Display statistics for all devices once
iostat

# Refresh every 2 seconds, display 5 times in total
iostat 2 5

3. Detailed Command Options

3.1 Common Options

Option Description
-c Display only CPU usage
-d Display only disk usage
-h Display in human-readable format (e.g., KB, MB, GB)
-k Display data in KB
-m Display data in MB
-N Display device mapping names
-p Display statistics for a specified device or partition
-t Display timestamp
-x Display extended statistics
-y Skip the first report (usually used with an interval)

3.2 Advanced Options

Option Description
-z Omit output for devices with zero activity
-j ID Display persistent names for specified devices
--dec={0|1|2} Specify the number of decimal places

4. Interpreting Output Results

4.1 CPU Statistics Section

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           5.32    0.00    1.06    0.25    0.00   93.37
  • %user: User-level (application) CPU usage
  • %nice: CPU usage of processes with adjusted priority
  • %system: System-level (kernel) CPU usage
  • %iowait: Percentage of time the CPU waits for I/O operations to complete
  • %steal: "Stolen" time in virtual environments
  • %idle: CPU idle time percentage

4.2 Disk Statistics Section (Basic)

Device:            tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sda               1.02        12.34         5.67    1234567     567890
  • tps: Transfers per second
  • kB_read/s: Amount of data read per second (KB)
  • kB_wrtn/s: Amount of data written per second (KB)
  • kB_read: Total amount of data read (KB)
  • kB_wrtn: Total amount of data written (KB)

4.3 Extended Statistics (-x Option)

Device:  rrqm/s   wrqm/s     r/s     w/s    rkB/s    wkB/s avgrq-sz avgqu-sz   await r_await w_await  svctm  %util
sda       0.00     0.50    1.00    0.50    12.00     4.00    21.33     0.02   10.00    8.00   14.00   6.00   0.90
  • rrqm/s: Number of merged read requests per second
  • wrqm/s: Number of merged write requests per second
  • r/s: Number of completed read I/O operations per second
  • w/s: Number of completed write I/O operations per second
  • rkB/s: KB read per second
  • wkB/s: KB written per second
  • avgrq-sz: Average data size per I/O operation (sectors)
  • avgqu-sz: Average I/O queue length
  • await: Average wait time per I/O operation (milliseconds)
  • r_await: Average wait time for read operations (milliseconds)
  • w_await: Average wait time for write operations (milliseconds)
  • svctm: Average service time per I/O operation (milliseconds)
  • %util: Device bandwidth utilization percentage

5. Practical Example Analysis

5.1 Monitor a Specific Disk

Example

# Monitor sda disk, refresh every 2 seconds, display 3 times in total
iostat -d -x sda 2 3

5.2 Comprehensive Monitoring of CPU and Disk

Example

# Monitor CPU and disk, refresh every 5 seconds, display continuously
iostat -c -d -x -t 5

5.3 Display in MB

Example

# Display disk statistics in MB
iostat -d -m

5.4 Monitor All Partitions

Example

# Monitor all partitions, including LVM
iostat -p ALL

6. Performance Metrics Interpretation and Optimization Suggestions

6.1 Key Performance Metrics

  1. %util: Device utilization

    • 80% indicates the device is approaching saturation

    • Sustained 100% indicates the device has become a bottleneck

  2. await: I/O wait time

    • Normally should be < 10ms

    • 50ms indicates possible performance issues

  3. avgqu-sz: Queue length

    • > 1 indicates the device may be overloaded

6.2 Common Problem Diagnosis

Problem 1: High %util but low throughput

  • Possible cause: Excessive random I/O
  • Solution: Optimize the application I/O pattern, consider using SSDs

Problem 2: High await but low %util

  • Possible cause: Controller or bus bottleneck
  • Solution: Check the HBA card or storage controller

Problem 3: Abnormally high rkB/s or wkB/s

  • Possible cause: Applications performing heavy reads/writes
  • Solution: Optimize the application or increase cache

7. Integration with Other Tools

7.1 Combining with vmstat

Example

# Monitor memory and I/O simultaneously
vmstat 1 5 & iostat -x 1 5

7.2 Combining with sar

Example

# View historical I/O data
sar -d

7.3 Combining with iotop

Example

# Identify specific high-I/O processes
iotop

8. Summary and Best Practices

8.1 Usage Recommendations

  1. In production environment monitoring, it is recommended to use the -x option for detailed data
  2. For long-term monitoring, use sar to collect historical data
  3. When analyzing performance issues, pay attention to both CPU and I/O data
  4. Comparing statistics from different time periods is more meaningful

8.2 Common Command Combinations

Example

# Comprehensive monitoring command
iostat -c -d -x -t -m 5

# Monitor only extended disk information
iostat -d -x 2

# Monitor a specific partition
iostat -p sda1 -x 1 10

By mastering the iostat command, system administrators can effectively monitor and analyze the I/O performance of Linux systems and promptly detect and resolve storage-related performance bottlenecks.

Linux 命令大全Linux Command Encyclopedia

Other Extensions