Partitioned garbage collection of a large stable heap

This paper describes a new garbage collection scheme for large persistent object stores. The scheme makes efficient use of the disk and main memory and does not delay applications. The heap is divided into partitions that are collected independently using stable information about inter-partition references. We use novel techniques to maintain this information using in-memory data structures and a log to avoid disk accesses. The result is a scheme that truly preserves the localized and scalable nature of partitioned collection.