CODEWITHSUNDEEP
mongodb
Nick Caruso
Nick Caruso

Reputation: 507

Slow performance on MongoDB Full Text Search for 500,000 document collection

I have a Mongo v3.4 DB on MLab's M2 instance with 3.5 GB RAM on a dedicated server.

I have 500,000 documents storing tweets with the body of the tweet in a single string field. I have a $text index on that field.

When I query that field, it can take anywhere from a few seconds to over two minutes. My query is:

[{
    "$match": {
        "$text": {
            "$search": "game losing"
        }
    }
},
{
    "$sort": {
        "score": {
            "$meta": "textScore"
        }
    }
},
{
    "$limit": 10
}]

I have reviewed the following posts:

And have incorporated responses to them, but I still have very poor performance.

I do update the tweets consistently with any new stats (i.e. the like count of a tweet went up).

Upvotes: 3

Views: 2527

Answers (1)

Divyanshu Sah
Divyanshu Sah

Reputation: 258

MongoDB is specially designed to process millions of documents. 500K is a very small dataset.

I think you are loading all the data (when a user searched for that item) simultaneously on the client-side, and Paginates it. It's causing server timeout. Because data also can be so huge.

What you can do here is: Use Aggeration Framework. Limit the no. of data to send from the Atlas server.

See the docs here: https://docs.mongodb.com/manual/aggregation/

See the Examples in that docs. :)

Upvotes: 1

Related Questions