Aggregation Granularity

Is there any way i can give segment granularity as DAY ( with time lines as 20:00 to 20:00, something dynamic) ?

Not right now, although there is a PR out to enable more descriptive segment granularities: https://github.com/druid-io/druid/pull/3766

I guess you can also use arbitrary granularity spec to achieve this.
docs can be found here - http://druid.io/docs/latest/ingestion/

Thanks Nishant. Not sure if it works with KAFKA Indexing Service - will try once and see.

I think it won’t, arbitrary granularity spec is more of a batch mode thing. For realtime / streaming ingestion you would want that PR I linked.

This would be really helpful. Though it’s not just a time zone issue.

We have many data sources that hand off on the same schedule, and if we could stagger their periods it would help to spread out the performance impact from hand off.

Is there anything planned for next release on this direction by any chance ?

Hi Gian, Do you know if arbitary granularity works with re-indexing of raw data ? Im unable to find any examples for this.

With reindexing using hadoop index task, arbitrary granularity should work.
Also, we are working towards improving this in 0.10.0 (Related PR - https://github.com/druid-io/druid/pull/3850)