弱一致性模型下的分布式垃圾回收算法RRDGC

RRDGC: A Distributed Garbage Collection Algorithm on Relaxed Consistency Model

垃圾回收有效地减轻了编程人员的负担、优化了程序的设计.在基于弱一致性模型的分布式共享存储系统中进行自动垃圾回收是非常困难的和低效的.本文提出了一种基于跟踪策略的实时分布式垃圾回收算法—RRDGC,使得各节点回收进程可以和应用程序并发执行,同时有效减少了弱一致性模型下额外的通信开销,降低了程序的响应时间.算法确保了垃圾回收的正确性和有效性,体现了增量和实时性的特点.最后通过仿真实验对比论证了该算法的正确性和高效性.

Garbage collection can effectively relieve the programmers＇ load and optimize the design ot programs. It Is mefficient to perform garbage collection in the distributed shared memory system （DSM） on relaxed memory model with the traditional methods. This paper proposes a trace-based garbage collection algorithm for the DSM system based on relaxed memory model, RRDGC. It makes use of the relaxed consistency model＇ s features to reduce the communication cost and response delay. This algorithm shows its advantage of real-time and increment, and is proved to be correct and effective.