We are trying to optimize prometheus for large workloads and we wonder if it is possible to have different interval between time series depending on their age. Like for newly ingested metrics scrape_interval gonna be 30s, but for samples older than 1 month we should have 1m interval between time series, then lets say for 6month and older 2m and so on.
Downsampling (reducing granularity as data ages) isn’t something that Prometheus itself currently supports, but it is something that other related projects such as Thanos do have.