Hi @Vatemecum
I believe a very similar question was answered by @Aasawari on TimeSeries last x documents please have a look at that thread and see if it helps your case.
Basically due to the way the time series collection is working currently, I think adding a timestamp window should make this better. For example, if you add {$match: {date_utc: {$gte: <some recent date time>}}} as the first stage of the pipeline, it should perform better.
If this doesn’t work, could you post the output of db.collection.explain('executionStats').aggregate(...) output, both before and after adding the timestamp window matching? Also some example documents will be helpful.
Best regards
Kevin