Cardinality aggregation

[jpountz]

The cardinality aggregation is a metric aggregation that allows to compute approximate unique counts based on the HyperLogLog++ algorithm which has the nice properties of both being close to accurate on low cardinalities and having fixed memory usage so that estimating high cardinalities doesn't blow up memory.

Example:

{
    "aggs" : {
        "author_count" : { 
            "cardinality" : { 
                "field" : "author"
            }
        }
    }
}