New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Aggregations: Add an option to use HdrHistogram to compute percentiles #8324
Labels
:Analytics/Aggregations
Aggregations
Comments
It probably makes sense, as part of this change, to make the algorithm used pluggable in the same way that we have for the significance heuristic. |
imotov
added a commit
to imotov/elasticsearch
that referenced
this issue
Jan 21, 2020
The method parameter is not used in the percentile aggs, instead the method is determined by the presence of `hdr` or `tdigest` objects. Relates to elastic#8324
imotov
added a commit
that referenced
this issue
Jan 22, 2020
The method parameter is not used in the percentile aggs, instead the method is determined by the presence of `hdr` or `tdigest` objects. Relates to #8324
imotov
added a commit
that referenced
this issue
Jan 22, 2020
The method parameter is not used in the percentile aggs, instead the method is determined by the presence of `hdr` or `tdigest` objects. Relates to #8324
debadair
pushed a commit
to debadair/elasticsearch
that referenced
this issue
Jan 28, 2020
…#51272) The method parameter is not used in the percentile aggs, instead the method is determined by the presence of `hdr` or `tdigest` objects. Relates to elastic#8324
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
HdrHistogram like t-digest can be used to estimate percentiles with a fixed amount of memory and has the interesting feature that two instances can be merged (useful for the reduce phase). It is less flexible than t-digest (assumes positive integers) and has a different take on accuracy (based on significant digits on the values, while t-digest gives more precision to extreme percentiles). That said, this approach makes sense to compute percentiles on response times, which is a common use-case and the specialization allows HdrHistogram to be very fast (I have not performed any benchmark but adding a new value with HdrHistogram is a
O(1)
while with t-digest it is aO(log(compression))
).The text was updated successfully, but these errors were encountered: