Database Group Guest Talk

The Database Group will host Sam Arch from Carnegie Mellon University for a guest talk this Thursday, October 24th, from 11:00 AM to 12:00 PM (CT). The meeting will take place in person in Room CS 4310, and all are welcome to attend!

Abstract: User-defined functions (UDFs) are procedural functions (written in PL/SQL, Python) callable from SQL queries. However, queries with UDFs are notoriously hard to optimize as they mix declarative SQL and procedural UDF code. As a result, SQL and UDFs have had a problematic relationship spanning decades. In SQL Server 2019, Microsoft shipped UDF inlining, a feature eliminating all UDF code by translating entire UDFs to SQL. However, inlining often leads to sub-optimal performance. We propose a new optimization technique, UDF outlining, to intentionally hide pieces of a UDF from the query optimizer, resulting in significantly simpler and faster queries. Are SQL and UDFs finally ready to be together?

Bio: Sam Arch is a third-year PhD student in Computer Science at Carnegie Mellon University, co-advised by Prof. Andy Pavlo and Prof. Todd Mowry. Sam’s research interests are at the intersection of compilers and database systems, with a specific focus on improving the performance of SQL queries that invoke user-defined functions (UDFs). Before starting grad school, Sam worked as a Compiler Engineer at Apple, where he helped ship Metal-accelerated rendering in Blender 3.1. Before that, Sam completed his undergraduate degree at The University of Sydney, Australia.