Performance Analysis: The USE Method
Central idea
For every system resource, check utilization, saturation, and errors before chasing more elaborate performance theories.
Summary
Performance investigations often begin with a user’s suspected cause and then anchor on the wrong subsystem. The USE Method provides a complete first pass: enumerate the resources on the actual data path, then check each for Utilization, Saturation, and Errors. Utilization asks how busy the resource is; saturation asks whether work is waiting; errors reveal failed operations even when averages appear healthy. Combine this resource-oriented scan with a precise problem statement and workload characterization before drilling down with specialized tools.
Key ideas
Start with an inventory
List CPUs, memory, disks, buses, network links, and software limits relevant to the request path.
Queues reveal pressure
A resource can be fully utilized without pain; saturation shows when demand is waiting and latency grows.
Avoid premature detail
A broad, structured scan prevents deep analysis of the wrong component.
USE
Utilization: how busy is it?
Saturation: is work waiting?
Errors: are operations failing?
Why it matters now
A shared first-pass method lets engineers and agents gather comparable evidence instead of producing disconnected performance guesses.
Continue with the original
This short version preserves the main argument. Follow the source for the complete talk, article, or book.