Best approach to find points within radius

Elastic Stack Elasticsearch
1

Hi guys. I would like to get you insight on what is the best approach to use radius filtering - i.e. get points which are within a given range from the specified center.

In Java API there are two ways, based on the format in which points coordinates are stored.

1). Coordinate has a "geo_point" mapping:
"location": {
"lat": -0.32005,
"lon": 51.51321
}

Then query will look like:
QueryBuilder qb = QueryBuilders.geoDistanceQuery("location")
.point(0.32003, 51.51328)
.distance(1, DistanceUnit.KILOMETERS)
.optimizeBbox("memory")
.geoDistance(GeoDistance.SLOPPY_ARC);

2). Coordinate has a "geo_shape" mapping:
"location": {
"type": "point",
"coordinates": [
-0.32005,
51.51321
]
}

Query will look like this:
QueryBuilder qb = QueryBuilders.geoShapeQuery("location",
ShapeBuilder.newCircleBuilder()
.center(0.32003, 51.51328)
.radius(1, DistanceUnit.KILOMETERS),
ShapeRelation.WITHIN);

Both queries yield the same result. However, I am not sure which one is a recommended approach.

In my application one of the coordinates has to be geo_shape - because I perform other geo shape queries on it (like find points within geo_shape polygon). Now the question is, should I add another field for storing the same coordinates in geo_point format (in addition to geo_shape) - just for the sake of using geo distance query? Will it provide any performance benefits as compared to approach 2)?

2

Hi Alexey,

heres an older forum thread pointing out some differences and tradeoffs between the two approaches. Not sure how much of this has changed in recent versions but as a starting point this might help.

1 Like
3

Thanks a lot, Christoph!
This link gave me the needed answer.