You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It looks like under low load it wouldn't let the delegate limit to perform (while it probably should raise limit, especially if it was pessimisized before on overload)
And under high load it will work fine with averages, but with percentiles - you're either skipping too many samples(because they exceed window size) or letting actual logic not to be executed, because you will be setting window size to a big number. So picking window size became really tricky and hard to achieve properly
The text was updated successfully, but these errors were encountered:
IgorPerikov
changed the title
Trying understanding part of windowed algorithm
Trying to understand part of windowed algorithm
Dec 3, 2019
I'm trying to figure out, why (and does it?) this condition check for windowed limit is important https://github.com/Netflix/concurrency-limits/blob/master/concurrency-limits-core/src/main/java/com/netflix/concurrency/limits/limit/WindowedLimit.java#L155
It looks like under low load it wouldn't let the delegate limit to perform (while it probably should raise limit, especially if it was pessimisized before on overload)
And under high load it will work fine with averages, but with percentiles - you're either skipping too many samples(because they exceed window size) or letting actual logic not to be executed, because you will be setting window size to a big number. So picking window size became really tricky and hard to achieve properly
The text was updated successfully, but these errors were encountered: