Why is the Linux load average over the past 1, 5 and 15 minutes reported as the exponential moving average, instead of just simple (unweighted mean) moving average? It seems to me that the simple moving average value would be easier to understand. And having the three different load average values already communicates what the load looks like nearer in time, so it is not apparent what added value the exponential weighting brings.

For background, see the uptime man page and the Wikipedia entry on load.


Exponential moving averages are actually better for presenting random data in a time series due to the smoothing effect of data artifacts. Wikipedia has some decent information of the subject: https://en.wikipedia.org/wiki/Exponential_smoothing#Background

The point here is to eliminate artifacts such as data inversion, which can occur when using simple moving averages.

Not the answer you're looking for? Browse other questions tagged or ask your own question.