What are the commonly used tools for Linux system performance monitoring, and how to analyze CPU, memory, disk, and network usage?

Linux system performance monitoring is an important skill for operations engineers, requiring mastery of various tools to diagnose system bottlenecks.

CPU monitoring:

• top/htop: real-time view of CPU usage and process information
• vmstat: reports virtual memory statistics, e.g., vmstat 1 5 (update every second, 5 times total)
• mpstat: displays usage of each CPU core, e.g., mpstat -P ALL 1
• sar: system activity report, e.g., sar -u 1 10 (collect every second, 10 times total)

Memory monitoring:

• free: displays memory usage, free -h shows in human-readable format
• vmstat: view memory swap, cache and other information
• ps aux: view process memory usage (VSZ, RSS)
• pmap: view process memory mappings

Disk monitoring:

• df: view disk space usage, df -h shows in human-readable format
• du: view directory or file size, du -sh directory
• iostat: view disk I/O statistics, e.g., iostat -x 1
• iotop: real-time view of disk I/O usage (requires root privileges)

Network monitoring:

• ifconfig/ip: view network interface configuration
• netstat/ss: view network connections and port listening, e.g., netstat -tulnp or ss -tulnp
• nethogs: view network bandwidth usage by process
• tcpdump: capture and analyze network traffic, e.g., tcpdump -i eth0 port 80

Log analysis:

• /var/log/messages: main system log
• /var/log/syslog: system log (Debian/Ubuntu)
• /var/log/dmesg: kernel boot log
• journalctl: systemd log viewing tool, e.g., journalctl -f (real-time view)

Comprehensive monitoring tools:

• dstat: integrates vmstat, iostat, netstat and other functions
• glances: web-based system monitoring tool
• prometheus + grafana: enterprise-level monitoring solution

Performance optimization recommendations:

• CPU: optimize algorithms, reduce unnecessary calculations, use multi-process/multi-threading
• Memory: reduce memory leaks, optimize data structures, use caching
• Disk: use SSD, optimize file system, reduce I/O operations
• Network: use CDN, optimize TCP parameters, reduce network requests