Garbage collection: an exercise in distributed, fault-tolerant programming
Two garbage-collection algorithms are presented to reclaim unused storage in object-oriented systems implemented on local area networks. The algorithms are fault-tolerant and allowed parallel, incremental collection in an object address space distributed throughout the system. The two approaches allow multiple collectors, so some unused storage can be reclaimed in partitioned networks. The first method makes use of fault-tolerant reference counts together with an algorithm to collect cycles of objects that would otherwise remain unclaimed. The second method adapts a parallel collector so that it can be used to collect subspaces of the entire network address space. Throughout this work concern is with a methodology for developing distributed, parallel, fault-tolerant programs. Also, there is concern with the suitability of object-oriented systems for such applications.
- Research Organization:
- Washington Univ., Seattle (USA)
- OSTI ID:
- 7055388
- Resource Relation:
- Other Information: Thesis (Ph. D.)
- Country of Publication:
- United States
- Language:
- English
Similar Records
Global garbage collection for distributed heap storage systems
Distributed resource management: garbage collection
Related Subjects
DISTRIBUTED DATA PROCESSING
FAULT TOLERANT COMPUTERS
PARALLEL PROCESSING
PROGRAMMING
ALGORITHMS
COMPUTERS
DATA PROCESSING
DIGITAL COMPUTERS
MATHEMATICAL LOGIC
PROCESSING
990210* - Supercomputers- (1987-1989)
990300 - Information Handling