Load average

Load, in computing, is a measure of the amount of processing a computer system is currently performing, usually in the form of a scalar and as some variation on a percentage. They come in different forms in the cases of differing operating systems. The average load of a computer system is known as the load average.

• On Unix and Unix-like systems the load average is generally given as three scalar numbers, representing the average number of processes ready to run in the last one, five and fifteen minutes. (Note: On Linux – and possibly other, similar Unix variants – one should divide the load average as reported by the uptime or top commands by the number of CPUs before comparing to load averages for any other system. That's because a multi-processor system has a number of processors to select from among the runnable processes. In some systems the number of CPUs is pre-factored into the load average reporting).

• On Microsoft Windows systems, a percentage of CPU load is generally used, with 100% being full CPU utilization.

Note that the load average is a measure solely of CPU utilization. It is only one factor in overall system performance (and often is the least significant).

The optimum load average is approximately 1 (or 100%) since the system is then not wasting any time and no processes are stalled waiting for CPU time.

However, processes which are awaiting I/O are "sleeping" or "blocked" and don't contribute to load average. On modern systems it is common for the performance bottlenecks to be disk I/O, network I/O (latency or bandwidth) and memory constraints (which often are translated into disk I/O through the commonly available virtual memory "paging" (and swapping) mechanism).