It seems like what I wrote above is confusing. The objects that I am storing in S3 are associated to the individual events. I am not referring to the segments in the deep store. In my case, each event / row in the pinot table is associated with one or more files in S3. During ingestion, I receive events and associated files, I store the events in pinot, and I store the files in S3. The files in S3 have a naming convention that lets them be associated with a specific event in pinot (e.g. the pinot table has an event_id and the files have the event_id as a prefix which enables me to look up the files associated with a specific event).