YeSQL

The diversity and complexity of modern data management applications have led to the extension relational paradigm with syntactic semantic support for User-Defined Functions (UDFs). Although well-established in traditional DBMS settings, UDFs become central many application contexts as well, such science, analytics, edge computing. Still, a critical limitation is impedance mismatch between their evaluation processing. In this paper, we present YeSQL, an SQL rich UDF along pluggable architecture easily integrate it either server-based or embedded database engines. YeSQL currently supports Python fully integrated queries scalar, aggregator, table functions. Key novel characteristics include easy implementation complex algorithms several performance enhancements, including tracing JIT compilation UDFs, parallelism fusion stateful seamless integration engine. Our experimental analysis showcases usability expressiveness demonstrates that our techniques minimizing context switching engine VM are very effective achieve significant speedups up 68x common, practical use cases compared earlier approaches alternative choices.