Add a time series aggregation to tsdb track #348
Conversation
| } | ||
| } | ||
| }, | ||
| "min_available_memory": { |
There was a problem hiding this comment.
I am not sure why we are also including this pipeline aggregation. Do we expect to see changes in performance if we introduce some optimization for time_series aggregator?
There was a problem hiding this comment.
I added this because I think a pipeline aggregator is part of the usage pattern with time series aggregation.
All the time series buckets is an information overflow, getting min, max or derivative from all time series buckets will be useful informations to plot in a graph.
There was a problem hiding this comment.
Or do you think we should just focus on the time series agg itself in the benchmark? So removing the pipeline agg and maybe even the date histogram here? So it is just the time_series agg?
|
LGTM apart understanding the reason why we need the min_bucket pipeline aggregation. |
because time series agg can create many buckets.
|
Closing in favour of #351 |
No description provided.