Hey Brian,

First, I'll have our team cleanup the docs a bit. While CPU is definitely a large factor, load averages include a number of other metrics (disk wait, network wait, memory waits due to pressure, etc) as components to the overall number.

In general, there's an analogy floating around with sys admins about load relating to a highway. A server with two CPU threads would be a two lane highway. If that same server had a load of 1.0, that highway could support double the traffic without any backups. A load of 2.0 would imply that highway has cars bumper-to-bumper, but they're all moving at exactly the speed limit (max throughput).

Simply, a sustained load of 2.0 with a server that has 2 cores would be 100% utilization. Though again, this is a generalization as each workload will be somewhat different: is a workload network heavy? disk heavy? etc.

We have some users that try to "pack" servers dense enough to average 80% of their "max load" to get the most out of their nodes but still leave room for spikes.

Regarding your numbers: LOAD 0.74, 0.64, 0.61

0.74 is load over the last minute, 0.64 is load over the last 5 minutes, and 0.61 is load over the last 15 minutes. If this is a two core machine, you're using about 25% of your resources.

Hope this helps!

Very Helpful. Thank you!