Is Druid suitable to be used to generate aggregated data for the following case?

Currently, I plan to use Druid, to keep track activity of a user, via browser session id.

timestamp browser_session_id type affiliate
2011-01-01T01:01:35Z 1 view google <-- matched with sales with browser_session_id 1
2011-01-01T01:01:63Z 2 view google
2011-01-01T01:02:35Z 3 view yahoo <-- matched with sales with browser_session_id 3
2011-01-01T01:03:36Z 4 view yahoo <-- matched with sales with browser_session_id 4
2011-01-01T01:04:51Z 1 sales null
2011-01-01T01:04:52Z 3 sales null
2011-01-01T01:04:53Z 4 sales null


I would like to know, which what is the number of sales generated from first time page view, if the page view is brought by google. For the above example, we can see google’s view with browser_session_id 1, does generate a sales with browser session_id 1.

May I know, is Druid suitable to generate the following aggregated outcome?

type affiliate num_of_sales_generated
view google 1
view yahoo 2



Only if browser_session_id is not stored as a dimension, but as a metric (hyperUnique or theta sketch), and the queryGranularity is set to non-millisecond.

See: about rollup