Fastest way of querying geospatial data in mongodb

Question

Currently I have my geojson data stored in this format:

  coord: [long, lat],
  time: unix timestamp,
  property: some property

I would like to find the nearest location with the closest timestamp (lte). The way how I am doing it right now is:

    collection.ensureIndex({loc: "2d"})
    collection.find(
            {coord : {
                    $near: [xval, yval],
                    $maxDistance: 200
                    },
             time: {
                    $lte: time
             }
            }).sort({time: -1}).limit(1).toArray(function(err, queryResult) {
  (did some return 404 and 200 here)
 }

When the data size is small, this works. But as my database has been increased to 50G+, this fails (always return 404 saying nothing's found) and I am thinking it is because the way how I query my data leads to slow performance. How should I change my query / data structure to improve and let it work again?

I don't think there is anything wrong with your query. You might have to scale your backend if you have 50G+ of data.... — Eric, Jan 11 '17 at 15:02
@Eric would you mind being a bit more specific about scaling my backend? — acbh, Jan 11 '17 at 15:06
@Eric more ram and cpu is out of question. How much would sharding improve my performance? — acbh, Jan 11 '17 at 15:41

score 0 · Accepted Answer · answered Jan 11 '17 at 17:23

You should first scale your mongodb vertically, by adding more ram and CPU power, and this until you reach your next plateau. Your next step is to scale horizontally (or shard) your DB by having more than one instance of MongoDB handling your dataset. Of course sharding improves the DB performance a lot. You can read this article which describe pretty much the steps to take to scale mongo.

Fastest way of querying geospatial data in mongodb

1 Answers1