Cloud Server Resource Monitoring and Optimization Guide

Is your cloud server's CPU stuck at 95% but you don't know which process is causing it? Or is RAM running out and OOM Killer randomly killing services? Cloud server resource monitoring is the foundation of solving performance issues proactively rather than reactively. In this guide, we cover how to monitor CPU, RAM, disk I/O, and network traffic using Linux command-line tools, how to detect bottlenecks, and concrete optimization steps.

Why Is Resource Monitoring Critical?

Managing a server without resource monitoring is like driving a car without a dashboard. You only notice problems when it crashes. With regular monitoring:

• You can plan capacity By tracking resource consumption trends, you can determine in advance when you need to scale.
• You can quickly identify performance bottlenecks You can determine within minutes whether the slowdown is caused by CPU, RAM, disk I/O, or network.
• You can optimize costs Allocating more resources than needed is a waste of money. Based on actual usage data, you can choose the right-sized cloud server.

Monitoring and Optimizing CPU Usage

Real-Time CPU Monitoring with htop

htop is an enhanced, colorful alternative to the top command. It shows each CPU core's usage separately, allows sorting processes by CPU or RAM, and lets you terminate processes directly.

# Install htop (Ubuntu/Debian)
sudo apt install htop -y

# Run
htop

# Filter processes by a specific user
htop -u www-data

# Sort by CPU usage: inside htop press F6 → PERCENT_CPU

Per-Core Analysis with mpstat

mpstat (part of the sysstat package) shows the usage distribution of each CPU core. It's a critical tool for workloads concentrated on a single core (single-threaded applications).

# Install sysstat
sudo apt install sysstat -y

# Show all cores' usage at 2-second intervals
mpstat -P ALL 2

# Key metrics to watch:
# %usr  → User-space CPU (your applications)
# %sys  → Kernel-space CPU (system calls)
# %iowait → CPU waiting for disk I/O (high = disk bottleneck)
# %idle → Idle CPU (low = at capacity limit)

Monitoring and Optimizing RAM Usage

Memory Analysis with free and vmstat

The free -h command shows the current memory status. However, the "used" value in Linux can be misleading because the kernel uses free RAM as disk cache (buffer/cache). For actual available memory, look at the "available" column.

# Current memory status
free -h

# Example output:
#               total   used   free   shared  buff/cache  available
# Mem:          4.0Gi   2.1Gi  0.3Gi  128Mi   1.6Gi       1.5Gi
# Swap:         2.0Gi   0.5Gi  1.5Gi

# Monitor memory and swap at 5-second intervals with vmstat
vmstat 5

# Note: if si (swap in) and so (swap out) values
# are consistently greater than zero, RAM is insufficient

RAM Optimization Steps

Follow these steps to reduce RAM consumption:

1. Adjust MySQL/MariaDB buffer pool size: Set the innodb_buffer_pool_size value to 50-70% of total RAM. On a server with 4 GB RAM, 2-2.5 GB is an ideal starting point.

2. Limit PHP-FPM worker count: Each PHP-FPM worker consumes an average of 30-50 MB RAM. On 4 GB RAM, keep the pm.max_children value between 20-30.

3. Disable unused services: List running services with systemctl list-units --type=service --state=running and disable unnecessary ones.

SWAP space configuration is also a critical step when RAM is insufficient. Check out our VPS SWAP Space Configuration guide for more details.

Disk I/O Monitoring and Bottleneck Detection

Disk I/O bottleneck is the most common performance issue, especially in database-intensive applications. You can monitor disk performance with iostat and iotop tools.

# Disk performance with iostat (2-second intervals)
iostat -xz 2

# Critical metrics:
# await  → Average I/O wait time (ms). Should be <1ms for NVMe, <5ms for SSD
# %util  → Disk utilization rate. If 80%+, there's a bottleneck
# r/s, w/s → Read/write operations per second

# To see which process is doing the most disk I/O
sudo iotop -oP

Monitoring Network Traffic

Network performance is critical, especially for API servers and media services. You can monitor real-time bandwidth usage and connection states with nload and ss commands.

# Real-time bandwidth usage
sudo apt install nload -y
nload eth0

# Active connection count and states
ss -s

# Find IPs with the most connections
ss -ntu | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -rn | head -20

# Connection state distribution (TIME_WAIT, ESTABLISHED, etc.)
ss -ant | awk '{print $1}' | sort | uniq -c | sort -rn

For more detailed information on network optimization, check out our VPS Network Performance Optimization guide.

Monitoring Tools Comparison

Automated Alerting and Monitoring Setup

Command-line tools are excellent for real-time monitoring, but for 24/7 monitoring you need to set up an automated alerting system. You can monitor critical thresholds with a simple bash script:

#!/bin/bash
# Simple resource monitoring and alerting script

CPU_THRESHOLD=85
RAM_THRESHOLD=90
DISK_THRESHOLD=85

# CPU check
CPU_USAGE=$(top -bn1 | grep "Cpu(s)" | awk '{print 100 - $8}' | cut -d. -f1)
if [ $CPU_USAGE -gt $CPU_THRESHOLD ]; then
  echo "WARNING: CPU usage at ${CPU_USAGE}%" | mail -s "Server Alert" [email protected]
fi

# RAM check
RAM_USAGE=$(free | awk '/Mem/{printf("%.0f"), $3/$2*100}')
if [ $RAM_USAGE -gt $RAM_THRESHOLD ]; then
  echo "WARNING: RAM usage at ${RAM_USAGE}%" | mail -s "Server Alert" [email protected]
fi

# Disk check
DISK_USAGE=$(df / | tail -1 | awk '{print $5}' | tr -d '%')
if [ $DISK_USAGE -gt $DISK_THRESHOLD ]; then
  echo "WARNING: Disk usage at ${DISK_USAGE}%" | mail -s "Server Alert" [email protected]
fi

You can run this script every 5 minutes with cron: */5 * * * * /opt/scripts/monitor-alert.sh. For more comprehensive monitoring, consider the Prometheus and Grafana combination.

Frequently Asked Questions

Conclusion

Resource monitoring on your cloud server is critical not only for solving performance issues but also for preventing them. Regularly monitor processes with htop, disk performance with iostat, and memory status with vmstat. Set up automated alerts for metrics exceeding threshold values and adopt a proactive rather than reactive management approach.