Overview
Learn about Prestissimo, a groundbreaking C++ Native Worker project designed to replace Java workers in Presto clusters, in this technical talk presented by Aditi Pandit from IBM's Ahana team. Explore the architecture of this state-of-the-art query processing engine built on the open-source Velox library, which provides reusable data processing primitives including new aggregation, join, window operators, Hive connector, and native Parquet and Iceberg readers. Discover the significant advantages of the Native engine, including enhanced performance through vectorization, SIMD, and adaptive runtime optimizations, more predictable behavior through improved memory management frameworks, and superior operational guidance. Gain practical insights into implementing Prestissimo and understanding its query processing behavior for optimal performance in data processing operations.
Syllabus
Introduction to Prestissimo - Aditi Pandit, IBM (Ahana team)
Taught by
Presto Foundation