Skip to main content

DDSketches for Approximate Quantiles module

This module provides aggregators for approximate quantile queries using the DDSketch library. The DDSketch library provides a fast, and fully-mergeable quantile sketch with relative error. If the true quantile is 100, a sketch with relative error of 1% guarantees a quantile value between 101 and 99. This is important and highly valuable behavior for long tail distributions. The best use case for these sketches is for accurately describing the upper quantiles of long tailed distributions such as network latencies.

To use this Apache Druid extension, include in the extensions load list.

druid.extensions.loadList=["druid-ddsketch", ...]

Aggregator

The result of the aggregation is a DDSketch that is the union of all sketches either built from raw data or read from the segments. The single number that is returned represents the total number of included data points. The default aggregator type of ddSketch uses the collapsingLowestDense strategy for storing and merging sketch. This means that in favor of keeping the highest values represented at the highest accuracy, the sketch will collapse and merge lower, smaller values in the sketch. Collapsed bins will lose accuracy guarantees. The default number of bins is 1000. Sketches can only be merged when using the same relativeError values.

The ddSketch aggregator operates over raw data and precomputed sketches.

{
"type" : "ddSketch",
"name" : <output_name>,
"fieldName" : <input_name>,
"relativeError" : <double(0, 1)>,
"numBins": <int>
}
propertydescriptionrequired?
typeMust be "ddSketch"yes
nameA String for the output (result) name of the calculation.yes
fieldNameA String for the name of the input field (can contain sketches or raw numeric values).yes
relativeErrorDescribes the precision in which to store the sketch. Must be a number between 0 and 1.no, defaults to 0.01 (1% error)
numBinsTotal number of bins the sketch is allowed to use to describe the distribution. This has a direct impact on max memory used. The more total bins available, the larger the range of accurate quantiles. With relative accuracy of 2%, only 275 bins are required to cover values between 1 millisecond and 1 minute. 800 bins are required to cover values between 1 nanosecond and 1 day.no, defaults to 1000

Post Aggregators

To compute approximate quantiles, use quantilesFromDDSketch to query for a set of quantiles or quantileFromDDSketch to query for a single quantile. Call these post-aggregators on the sketches created by the ddSketch aggregators.

quantilesFromDDSketch

Use quantilesFromDDSketch to fetch multiple quantiles.

{
"type" : "quantilesFromDDSketch",
"name" : <output_name>,
"field" : <reference to DDSketch>,
"fractions" : <array of doubles in [0,1]>
}
propertydescriptionrequired?
typeMust be "quantilesFromDDSketch"yes
nameA String for the output (result) name of the calculation.yes
fieldA computed ddSketch.yes
fractionsArray of doubles from 0 to 1 of the quantiles to computeyes

quantileFromDDSketch

Use quantileFromDDSketch to fetch a single quantile.

{
"type" : "quantileFromDDSketch",
"name" : <output_name>,
"field" : <reference to DDsketch>,
"fraction" : <double [0,1]>
}
propertydescriptionrequired?
typeMust be "quantileFromDDSketch"yes
nameA String for the output (result) name of the calculation.yes
fieldA computed ddSketch.yes
fractionA double from 0 to 1 of the quantile to computeyes

Example

As an example of a query with sketches pre-aggregated at ingestion time, one could set up the following aggregator at ingest:

{
"type": "ddSketch",
"name": "sketch",
"fieldName": "value",
"relativeError": 0.01,
"numBins": 1000,
}

Compute quantiles from the pre-aggregated sketches using the following aggregator and post-aggregator.

{
"aggregations": [{
"type": "ddSketch",
"name": "sketch",
"fieldName": "sketch",
}],
"postAggregations": [
{
"type": "quantilesFromDDSketch",
"name": "quantiles",
"fractions": [0.5, 0.75, 0.9, 0.99],
"field": {
"type": "fieldAccess",
"fieldName": "sketch"
}
}]
}