I have data that deals with time spans, rather than discrete dates. I

store a start time and end time.

I need to produce a date histogram taking into account the time

spans. So for instance, if a document has a start date of 2012-03-01

and end date of 2012-03-10, and I produced the histogram with a day

interval, the document would be counted once on 9 different days.

I have many millions of records and the histogram interval may be much

smaller than the date range in some cases.

Is there any way to do this efficiently in Elastic?

Alternately, here's the real problem I need to solve:

Given many documents, some with time spans, some with discrete dates,

identify the time spans for which there's at least one matching

document per time interval (e.g. find blocks of time in April where

data is no more than an hour apart).

Is that any easier for Elastic to solve?

Thanks!