What if some documents don't have a field that is part of an index?

Question

A collection has an indexed involved field_A. But field_A is not required. So what happens if some documents do not have this field? Will the index still work for documents that do have this field?

Answer 1

Yes it works, here is a test:

db.collection.createIndex({ field_A: 1 });

for (let i = 0; i < 100; i++)
   db.collection.insertOne({ field_B: i });

db.collection.stats(1024).indexSizes

{ "_id_" : 20, "field_A_1" : 20 }

You see index field_A_1 has a size of 20 kiByte. This behavior is different to most relational DBMS database where such index would have a size of zero.

The index is also used by your query, if you use the field:

db.collection.find({ field_B: 1 }).explain().queryPlanner.winningPlan;

{
    "stage" : "COLLSCAN",
    "filter" : {
        "field_B" : {
            "$eq" : 1
        }
    }
}

db.collection.find({ field_A: null, field_B: 1 }).explain().queryPlanner.winningPlan;

{
    "stage" : "FETCH",
    "inputStage" : {
        "stage" : "IXSCAN",
        "keyPattern" : {
            "field_A" : 1
        },
        "indexName" : "field_A_1",
        "indexBounds" : {
            "field_A" : [
                "[undefined, undefined]",
                "[null, null]"
            ]
        }
    }
}