Understanding Server Metrics: CPU, Memory, Disk, and Load

If you're managing Linux servers, you need to understand what the numbers on your dashboard actually mean. This guide breaks down the four most important server metrics and explains when you should be concerned.

CPU Usage

CPU usage tells you what percentage of your processor's capacity is being used. It's reported as a percentage from 0% (idle) to 100% (fully loaded).

What's normal?

• 0-30%: Light load, plenty of headroom
• 30-70%: Moderate load, healthy for a production server
• 70-90%: Heavy load, monitor closely
• 90-100%: Critical — processes may be queuing

Important nuance: A server running at 60% CPU isn't necessarily in trouble. It depends on whether that's steady or spiking. Server Scout tracks CPU over time so you can spot trends.

When to worry: Sustained CPU above 85% for more than 15 minutes, or sudden spikes that correlate with user complaints.

Memory Usage

Memory (RAM) usage shows how much of your server's physical memory is in use. Linux memory reporting can be confusing because the kernel uses "available" memory for disk caching, which is a good thing.

Key values to watch:

• Used: Memory actively used by applications
• Available: Memory that can be immediately allocated (includes reclaimable cache)
• Total: Physical RAM installed

What's normal?

• Available > 20% of total: Healthy
• Available 10-20%: Getting tight, consider optimising or upgrading
• Available < 10%: Critical — the system may start using swap

When to worry: When available memory consistently drops below 15% of total, or when you see swap usage increasing. Swap is much slower than RAM and will degrade performance significantly.

Disk Usage

Disk usage shows how much of each filesystem's capacity is consumed. This is one of the most critical metrics because a full disk can crash applications and even prevent the system from booting properly.

What's normal?

• 0-70%: Healthy, no action needed
• 70-85%: Plan for cleanup or expansion
• 85-95%: Warning — take action soon
• 95-100%: Critical — immediate action required

When to worry: When disk usage crosses 85% and is trending upward. Log files, database growth, and temporary files are common culprits.

Pro tip: Set up disk usage alerts at 85% and 95% thresholds. The 85% alert gives you time to act; the 95% alert means act now.

Load Average

Load average is perhaps the most misunderstood metric. It represents the average number of processes waiting for CPU time over 1, 5, and 15-minute intervals.

How to interpret it:

The key is to compare load average to your number of CPU cores:

• Load < cores: The system has idle capacity
• Load = cores: The system is fully utilised
• Load > cores: Processes are waiting for CPU time

For example, on a 4-core server:

• Load of 2.0 = 50% utilisation (healthy)
• Load of 4.0 = 100% utilisation (at capacity)
• Load of 8.0 = 200% utilisation (overloaded)

The three values matter:

• 1-minute: Shows current state (can be spiky)
• 5-minute: Better indicator of sustained load
• 15-minute: Shows the longer trend

When to worry: When the 5-minute load average consistently exceeds your core count by 50% or more.

Putting It All Together

No single metric tells the full story. Here's how to read them together:

• High CPU + High Load: CPU bottleneck — your server needs more processing power
• High Memory + Swap Usage: Memory bottleneck — your applications need more RAM
• High Disk + Slow I/O: Disk bottleneck — consider faster storage or cleanup
• Low everything but slow: Likely a network issue or external dependency

Monitor with Server Scout

Server Scout tracks all four metrics in real time and stores historical data so you can spot trends before they become problems. Start your free trial and get visibility into your servers within 60 seconds.