Optimization of $match+$sort on aggregation pipeline

[Prasad_Saya]

@sammy_Ma

I just want to know the stage execution sequence. Is $match before $sort or $sort before $match ?

The execution sequence is always $match followed by $sort (irrespective of the order of these stages). The optimizer makes sure that the sort stage has documents after they are filtered (that would be less number of documents than the input to the match stage).

More details at: $sort + $match Sequence Optimization

@sammy_Ma

Why the number of "keysExamined" is 10 but not 5?? In my opinion, $match will scan all documents in collections, and then get 5 documents.

There might be some index on the collection bar3, and this is affecting the $match stage, I think. You can post the details.

If possible, please run the explain for the aggregation query with “executionStats” mode.