How to understand the reason for fetch behavior in MongoDB?

Question

MongoDB

Greetings. Input data:
- a collection of about 11.5 million records
- each document has a string format field, which can either have a string or an empty value
- each document has a numeric field from 0 to 100, which is used in sorts
Query 1 (1.74 sec) = 5971171:

db.getCollection('channels').find(
    { 'update': { $gt: '' }}
).count();

Request 2 (1.74 sec) = 5697946:

db.getCollection('channels').find(
    { 'update': { $eq: '' }}
).count();

Relatively close results, i.e. the entire collection is essentially divided approximately in half - empty / non-empty values. Next, I want to select documents with the highest sort index from these two groups:
Query by "not empty" (0.005 sec):

db.getCollection('channels').find(
    { 'update': { $gt: '' }}
).sort(
    { 'sort': -1 }
).limit(1);

Request for "empty" (12.4 sec):

db.getCollection('channels').find(
    { 'update': { $eq: '' }}
).sort(
    { 'sort': -1 }
).limit(1);

Indices:

_sort_ {
    "sort" : -1
}
_update_ {
    "update" : 1
}

What is the reason for this runtime difference? Especially when you consider that the index on the "update" field is increasing, and empty values ​​"" should be at the beginning of the index, which means it's much faster to get out...

Answer 1

Write -1 where there is no value.
Use {update: -1}to search for "blank" lines.
What does explain say?