You can use the value returned from data Source Key as a unique identifier for publishing data into the data source.
For Postgres I did add indexes on order date and status but other wise kept with defaults for both Postgres and Elastic Search.
This is not about getting the best numbers but instead looking at the variations we can see in changing selectivity of predicates and denormalisation.
It is also interesting to note that for Elastic Search this is a non-scoring query (filter query only) so it's good to see how it compares ballpark wise to Postgres for this simple query.
Elastic Search did very well but actually the object it returned contained some additional denormalised information that we didn't have in the Postgres query.