Why use exponential moving average for Average Queue Holding time?

Just curious why a exponential average was chosen instead of a simple average for queue hold times?

I did a bit of digital spelunking – the per-queue hold time estimates as we currently have them were proposed by David Troy (There’s a great commentary in the comments at https://issues.asterisk.org/jira/browse/ASTERISK-211). He originally described the algorithm as a “boxcar filter”. That being said, in 2008 David Wooley found that the comment in the code was wrong – that it wasn’t actually a boxcar filter, but was an exponential average (IIR) (see https://issues.asterisk.org/jira/browse/ASTERISK-12135 for more details.)


Thanks for taking the time to dig those up it was helpful for understanding why it was chosen.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.