Project

Compressed Data Structures for GraalVM Native Images

Principal Investigator

Gustavo Alonso

ETH Zurich

Oracle Fellowship Recipient

Rodrigo Bruno

Summary

Native image generation performs a static analysis for allowing a closed-world assumption for the program. This research focuses on using the results from this analysis for applying data structure compression techniques in the resulting native image. The research includes identifying opportunities for compression in common workloads, proofing correctness of possible compression candidates, and modifications to the native image runtime and GC code to support such compressions. Success is measured in resident set size (RSS) usage for common we server workloads.