Re:USE Method: Linux Performance Checklist
by jamesaguilar on Tuesday July 01, 2014 @07:36PM
(#7974167)
A good collection of tools, but I can't figure out the distinction between utilization and saturation in this schema.
[ Parent
| on HN ]
Re:USE Method: Linux Performance Checklist
by georgebashi on Tuesday July 01, 2014 @08:41PM
(#7974427)
This is a summary / cheat-sheet for diagnosing performance problems with the "USE" method. It's explained in much more detail here: http://www.brendangregg.com/usemethod.html
Worth checking out the rest of the site too, really useful stuff for performance engineers.
For reference, the definitions are:
* utilization: the average time that the resource was busy servicing work
* saturation: the degree to which the resource has extra work which it can't service, often queued
* errors: the count of error events
He described saturation as how much the processes want to use the resource beyond the maximum 100% utilization. E.g. if you look at load average on Linux, that's the size of the run queue (processes that are ready to run + blocking on uninterruptable IO) versus just CPU usage which cannot go over the 100%. That requires there is a meaningful queue. There's lots more in his excellent "Systems Performance" book.