Databricks has introduced four new sketch functions designed to significantly accelerate analytical queries. These functions, built on Apache DataSketches, enable faster computation of percentiles, distinct counts, and top-K items by using bounded-memory approximations with a configurable relative error of 1-2%. This approach allows for orders-of-magnitude reduction in compute time and memory usage compared to exact calculations. The new capabilities also facilitate real-time analytics, such as audience overlap analysis and trending leaderboards, by allowing sketches to be stored, merged, and requeried without reprocessing raw data. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Accelerates analytical queries for decision support, reducing compute costs and enabling real-time insights.
RANK_REASON Databricks released new sketch functions for its analytics platform.