Queue is the current runtime queue for processes. Load averages are measurements of this metric over 5-, 10-, and 15- minute intervals. So, the runtime queue is a count of the number of processes ready to run at any given point in time.
$ uptime
14:17pm up 0:57, 6 users, load average: 0.37, 0.54, 0.44
That means, a value less than 1 is not bad and there is never a process idling in the queue waiting to be run. Therefore, when the load averages start to rise over 1, know there is always a process ready and waiting to be scheduled!
Obviously, when the load averages rise above 1 there is something in the system being delayed. If it's a mission critical system, running some serious stuff in a production environment, getting an additional CPU can solve the problem and ensure that there is a better chance of “schedulable” time.
These data can be monitored through:
$ uptime
$ top
$ vmstat
uptime somehow has a nice output of the load average data.
Read: