improve query performance mongodb - node.js

Question

I am using aggregation query to retrieve 20000 records. while retrieving it is taking much time. I will mention my query below, Please help me to improve the query performance.
Query:
[err, data] = await to(LeadsLog.aggregate([
{$lookup:{
from: "leads",
localField: "leadId",
foreignField: "_id",
as: "leadId"
}},
{$lookup:{
from: "company_contacts",
localField: "leadId.assignedTo",
foreignField: "_id",
as: "assignedTo"
}},
{
$unwind:{
path: "$leadId",
preserveNullAndEmptyArrays: true
}
},
{
$match:{"leadId.assignedTo":new mongoose.Types.ObjectId(userId),
"result":{$eq:null}}
},
{ '$facet' : {
metadata: [ { $count: "total" }, { $addFields: { page: 1 } } ],
data: [ { $skip: 0 }, { $limit: 20000 } ]
} }
] ));
LeadId:
{
"_id" : ObjectId("617a84b401c98424e00d1310"),
"status" : true,
"address" : "Howmif Trail",
"city" : "Kinawnet",
"state" : "LA",
"country" : "LA",
"pincode" : null,
"extraFormObject" : null,
"lead_name" : "Jayden",
"phone" : "[phone]",
"email" : "niligis#taptehe.cm",
"company" : ObjectId("6155c2758609663d10fff796"),
"createdBy" : ObjectId("6155c2758609663d10fff798"),
"createdAt" : ISODate("2021-10-28T11:08:40.433Z"),
"updatedAt" : ISODate("2021-10-30T04:43:49.490Z")
}
LeadLog:
{
"_id" : ObjectId("617a84bf01c98424e00daf52"),
"callLogId" : null,
"result" : null,
"assignedTo" : ObjectId("6155c2758609663d10fff798"),
"extraFormObject" : null,
"subResult" : null,
"apptDate" : null,
"nextcallDate" : ISODate("2021-10-28T11:02:50.516Z"),
"callDate" : null,
"leadId" : ObjectId("617a84b401c98424e00d1310"),
"company" : ObjectId("6155c2758609663d10fff796"),
"createdAt" : ISODate("2021-10-28T11:08:50.962Z"),
"updatedAt" : ISODate("2021-10-30T04:43:50.281Z")
}
Please help me with better solution. thank you.

Answer 1

There are a few simple tweaks that you can improve your existing query:
make intermediate result as small as possible; one of the common ways is pushing $match stages as early as possible
use Pipeline Coalescence Optimization as much as possible; one of the common tuples would be $lookup + $unwind combination
index the $match fields and $lookup fields
Based on the first 2 points, here is my suggested form of your query:
You can see result : {$eq: null} is pushed to first stage. The performance gain will depends on the selectivity of the clause.
the $lookup and $unwind leads are grouped together to utilize the coalescence optimization.
"leadId.assignedTo": new mongoose.Types.ObjectId(userId) is moved earlier to minimize intermediate result size
Don't forget to index the relevant $match fields and $lookup fields. From my personal experience, good usage of index will help most with the performance.
[err, data] = await to(LeadsLog.aggregate([
{
$match: {
"result": {
$eq: null
}
}
},
{
$lookup: {
from: "leads",
localField: "leadId",
foreignField: "_id",
as: "leadId"
}
},
{
$unwind: {
path: "$leadId",
preserveNullAndEmptyArrays: true
}
},
{
$match: {
"leadId.assignedTo": new mongoose.Types.ObjectId(userId)
}
},
{
$lookup: {
from: "company_contacts",
localField: "leadId.assignedTo",
foreignField: "_id",
as: "assignedTo"
}
},
{
"$facet": {
metadata: [
{
$count: "total"
},
{
$addFields: {
page: 1
}
}
],
data: [
{
$skip: 0
},
{
$limit: 20000
}
]
}
}
]));