Does Druid Support Sliding Windows


I would like to know if Druid supports timeseries queryType with
1D of granularity and 5D of sliding window. More specifically, I
would like to compute the number of unique users (hyperLogLog
dimension) for every single day based on the previous two days.
Would that be possible?

Druid doesn’t have explicit support for sliding windows yet, although if you are really in a pickle you can compute them client-side by using a normal tumbling window query and then combining adjacent values yourself. For HLL this gets a bit complex but is doable. You’d want to pass “finalize”: false in your query context, deserialize the HLLs yourself, and fold objects together to make sliding windows.

I think this might be easier to achieve with theta sketches take a look on this link