0

Sorry if I'm asking an obvious question. I have proposed a load balancing algorithm for a system of several devices. Suppose at first the number of tasks on each device is $device1 = 18$, $device2 = 8$ and $device3 = 2$. Now after load balancing the number of tasks on each device has become $device1 = 10$, $device2 = 8$ and $device3 = 10$.

I want to publish the results and say that the number of tasks on devices is moderated after load balancing. I mean at first for example one device had $18$ tasks, while another had $2$ tasks, but after load balancing this amount is almost equal between all devices and it is almost $10$ which shows that these tasks are distributed almost equally.

But how can I state this in mathematical language? I was thinking about "the mean value of tasks" before and after load balancing, but then I saw that they are the same. So is there any mathematical way to show this difference in number of tasks before and after?

Pablo
  • 123
  • Look up variance https://www.google.com/search?client=ubuntu&channel=fs&q=variance+and+standard+deviation&ie=utf-8&oe=utf-8 – Ethan Bolker Jan 27 '19 at 22:46
  • @EthanBolker Thanks Ethan, I looked up variance and standard deviation and I found out useful information. So can I say that since after load balancing the value of standard deviation is lower than before load balancing, the load balancing is useful? I mean, in my case is it true that: the less the standard deviation is, the better the result is? – Pablo Jan 27 '19 at 23:43
  • That is a reasonable definition of the improvement your algorithm provides. – Ethan Bolker Jan 28 '19 at 00:49

0 Answers0