メインコンテンツまでスキップ
に最終更新

kll_sketch_merge_bigint function

Applies to: check marked yes Databricks Runtime 18.0 and later

Merges two compatible integer KLL sketches into a single sketch.

Syntax​

kll_sketch_merge_bigint ( sketch1, sketch2 )

Arguments​

  • sketch1: A BINARY expression containing a serialized integer KLL sketch.
  • sketch2: A BINARY expression containing a serialized integer KLL sketch.

Returns​

A BINARY value containing the merged sketch.

Notes​

  • Both sketches must be for the same data type (integer).
  • The merged sketch has the same k parameter as the input sketches.
  • Sketches with different k values cannot be merged.
  • Useful for distributed aggregation: create sketches per partition, then merge.

Examples​

SQL
-- Merge sketches from different data partitions
-- Use merged sketch for quantile estimation across all data
> WITH partition1 AS (
    SELECT kll_sketch_agg_bigint(value) AS sketch FROM VALUES (1), (2) AS T(value)
  ),
  partition2 AS (
    SELECT kll_sketch_agg_bigint(value) AS sketch FROM VALUES (3), (4) AS T(value)
  ),
  merged AS (
    SELECT kll_sketch_merge_bigint(p1.sketch, p2.sketch) AS sketch
    FROM partition1 p1, partition2 p2
  )
  SELECT kll_sketch_get_quantile_bigint(sketch, 0.5) FROM merged
  2