Update the aggregation output to two different collections

[steevej (Steeve Juneau)]

@Santosh_Gonsalves

Yes, that can be done but the objective is to avoid querying the whole bunch of documents and processing them with the same logic again

That was clear from the beginning.

@Santosh_Gonsalves

You mean the second aggregation

No. I mean the first aggregation. You did not really shared the pipeline so I will try to illustrate my question.

db.FirstCollection.aggregate( [
stage-1 ,
...,
stage-5 ,
merge-into-A
stage-6 ,
... ,
stage-10
merge-into-B
] )

Is FirstCollection the same as A or B or is it a third one that is not A nor B? If it is a third one, then an idea that might work is that you create an on-demand materialized view from the first 5 stages. The you could $merge the materialized into A, and then start an aggregation on the materialized view with step-6 that you would merge into B.

This way step-1 to step-5 are executed only once when the materialized view is created.