Master Linux I/O Monitoring: iostat, iotop, sar, dstat, nmon & More

1. System Built-in Tools

1.1 iostat

iostat is part of the sysstat package and reports I/O statistics in kB/s. Use the following command to collect data every 2 seconds for 10,000 intervals: iostat -x -k 2 10000 Important fields:

rrqm/s and wrqm/s – number of read/write requests merged per second.

rsec/s and wsec/s – sectors read/written per second.

rKB/s and wKB/s – kilobytes read/written per second.

avgrq‑sz – average request size.

avgqu‑sz – average queue length (shorter is better).

await – average I/O request latency (including queue and service time). Values >5 ms may indicate a problem.

svctm – average service time per request.

%util – proportion of time the device was busy; values near 100 % suggest the device is saturated.

1.2 iotop

Install with yum install iotop. Example usage to display active I/O processes every second: iotop -d 1 -o Key options:

-o – show only processes doing I/O.

-b – batch mode for logging.

-n NUM – number of iterations.

-d SEC – delay between updates.

-p PID – monitor a specific PID.

-u USER – monitor processes of a specific user.

Common shortcuts:

←/→ – change sort column (default is I/O).

r – reverse sort order.

o – toggle I/O‑only view.

p – toggle process/thread view.

a – show accumulated I/O.

q – quit.

1.3 sar

Run with: sar -p -d 2 1000 Relevant metrics:

tps – total I/O requests per second (read + write).

rtps – read requests per second.

wtps – write requests per second.

bread – bytes read per second.

bwrtn – bytes written per second.

Alternative command for block I/O:

sar -b 2 1000

1.4 dstat

dstat is a versatile replacement for vmstat, iostat, netstat, nfsstat, and ifstat, providing colored, real‑time output.

Common options:

-c – CPU statistics.

-C – per‑CPU statistics (e.g., -C 0,1).

-d – disk I/O.

-n – network statistics.

-l – system load.

-m – memory usage.

-g – page statistics.

-p – process statistics.

-s – swap usage.

-r – I/O request statistics.

-y – overall system status.

--output FILE – write CSV output to FILE.

2. Monitoring Tools

2.1 nmon

Installation on Linux:

wget http://sourceforge.net/projects/nmon/files/download/nmon_x86_12a.zip/download

unzip download

yum install ld-linux.so.2 libncurses.so.5

chmod +x nmon_x86_rhel52

mv nmon_x86_rhel52 /usr/local/bin/nmon

Collect data every 10 seconds for 60 samples (10 minutes): nmon -s10 -c60 -f -m ./ Check the background process: ps -ef | grep -i nmon Analysis can be performed with nmon_analyser or by opening the generated CSV in Excel.

2.2 Zabbix

Zabbix provides ready‑made IOPS monitoring templates that can be imported directly, producing graphical dashboards.

2.3 Cacti

Cacti is a mature monitoring platform with downloadable IOPS templates. Typical deployment steps:

Unzip the template package.

Copy net-snmp_devio.xml to the Cacti resource directory.

Import all *_TMPL.xml files via the Cacti “Import Templates” interface.

Import net-snmp_devIO-Data_query.xml last.

Add the “ucd/net – Get Device I/O” data query to an SNMP‑enabled host.

Create graphs for the desired disk or memory devices.