Frequently Asked Question

What is Brendan Gregg's USE method and why is it the right place to start?

USE stands for Utilisation, Saturation, and Errors. Brendan Gregg's method says: for every resource in your system (CPU, memory, disks, network interfaces, I/O buses, interconnects), check those three numbers. Utilisation is the fraction of time the resource was busy; saturation is the queue or backlog it has accumulated; errors are the count of any kind of failure. If any of them is bad, the resource is a suspect. If all of them are clean across every resource, the bottleneck is somewhere else entirely (the application, a lock, a remote service) and you can stop chasing the hardware.

It works because it is exhaustive without being expensive. Instead of randomly running tools and hoping to spot something, you walk a fixed checklist and tick boxes. After a few investigations the sequence, top, vmstat, iostat, mpstat, ss, dmesg, sar, becomes muscle memory. The USE method's complement, RED (Rate, Errors, Duration) is the equivalent for services rather than resources; both belong in your toolkit.