Hello
Lets say you timechart with span=1h and within that hour you have 10000 requests that you need to calculate the average service time, is it possible to do that with the time used by the 10000 requests and not the full hour? Lets say that the total time spent servicing those requests are 20 minutes and nothing happens for the rest of the hour. If you take an average for the full hour that value will be very different from the average if you only use the total time spent working, and the value with only the time spent working will be more correct?
This is already how timechart
behaves.
The span
parameter to timechart
doesn't affect the statistical calculations it's performing, it only dictates how events should be put in discrete buckets. When you do for instance timechart span=1h avg(time_taken)
, what it does is simply to take all the events and get an average of the time_taken
from them.
This is already how timechart
behaves.
The span
parameter to timechart
doesn't affect the statistical calculations it's performing, it only dictates how events should be put in discrete buckets. When you do for instance timechart span=1h avg(time_taken)
, what it does is simply to take all the events and get an average of the time_taken
from them.
Well it depends on how the events are distributed, but yes, if all the events end up into one of the 20 minute spans, the average value will be identical.
So, a span=1h will yield the same result as span=20m? within that hour