C3: Compressing Correlated Columns - A Framework for Multi-Column Data Compression
Explore a technical talk that introduces C3 (Compressing Correlated Columns), an innovative compression framework designed to exploit correlations between columns for more effective data compression. Learn how this framework builds upon typical lightweight compression infrastructure by incorporating six new multi-column compression schemes, developed based on patterns observed in real-world datasets. Discover how C3's sampling-based algorithm selects optimal compression schemes for each column, resulting in approximately 20% higher compression ratios compared to traditional single-column compression methods when tested on the Public BI benchmark. Presented by Thomas Glas, a computer science master's student at the Technical University of Munich and researcher at CWI's Database Architectures Group, this 25-minute presentation from the Dutch Seminar on Data Systems Design offers valuable insights into advancing columnar data compression techniques.