I hope my question doesn't double something already asked and answered here.
I'm searching for the best way to store aggregated data in a dedicated index. In the best-case scenario, it should include both historical (already indexed) and newly ingested data. I'm aware of aggregation transforms and rollup jobs and have been testing them out.
- Is there any way to get a number of unique timestamp values to store them in the aggregated/rollup index? We need it to monitor ingested data consistency. I can easily get this metric's visualization in Kibana for non-aggregated data, but I see no option to get this metric as a part of my aggregated index.
- Is there a way to count unique values and percentiles (namely, medians/50-percentiles) in a rollup job at all? A rollup job seems to perform way faster than an aggregation transfom, and would be more efficient if we'd to scale this practice.
Any thoughts and suggestions are very much appreciated!