USE Method: Utilization, Saturation, Errors
You'll learn a systematic approach for finding performance bottlenecks by examining three key metrics across all system resources.
What is the USE Method?
The USE Method, created by Brendan Gregg, is a resource-focused methodology for quickly identifying bottlenecks in distributed systems. Unlike the Four Golden Signals (which focus on service-level behavior), USE drills down into individual hardware and software resources.
For every resource in your system (CPU, memory, disk, network interface), you ask three questions:
- Utilization: How busy is the resource? (percentage of time actively working)
- Saturation: How much excess work is queued or waiting? (backlog or waiting threads)
- Errors: Are there any failures occurring? (dropped packets, failed allocations, retries)
How It Works
Think of a highway system:
- Utilization = percentage of lanes occupied by cars
- Saturation = traffic backed up at on-ramps waiting to merge
- Errors = accidents or closed lanes preventing flow
Apply this to technical resources:
- CPU: 80% utilized, 5 processes in run queue (saturation), 0 errors
- Memory: 95% utilized, swap activity (saturation), OOM kills (errors)
- Network: 60% bandwidth used, retransmit queue growing (saturation), packet drops (errors)
- Disk: 70% I/O time, increasing wait queue (saturation), read failures (errors)
Why USE Method Matters
It provides a complete checklist for performance investigations. When users report slowness, systematically checking USE metrics across all resources prevents overlooking critical bottlenecks. High saturation often appears before utilization reaches 100%, making it an early warning signal.
Key Takeaway: The USE Method systematically examines Utilization, Saturation, and Errors for every system resource to quickly pinpoint hardware and infrastructure bottlenecks.