Chinese Journal of Computers   Full Text
Title	An Optimistic Data Consistency Maintenance Method Based on Data Dependence
Authors	ZHOU JingWANG Yi-JieLI Si-Kun
Address	(National Key Laboratory for Parallel and Distributed Processing, School of Computer, National University of Defense Technology, Changsha 410073)
Year	2008
Issue	No.5(741—754)
Abstract &   Background	Abstract This paper proposes an optimistic data consistency method according to the question about data dependence in data consistency. In the method, data object is partitioned into data blocks by fixed size as the basic unit of data management. Updates are compressed by Bloom filter technique and propagated in double-path. Negotiation algorithms detect and reconcile update conflicts, and dynamic data management algorithms accommodate dynamic data processing. The results of the performance evaluation show that it is an efficient method to achieve consistency, good dynamic property, and strong robustness when choosing the size of data block appropriately. At the same time, a feasible way is put forward on how to choose appropriate data block size. Keywords P2P distributed storage system; data replication; data consistency; data dependence; update conflict Background Data replication is an important technique to improve the performance of distributed systems. P2P systems replicate objects on multiple nodes to improve availability and performance. Data replication has many advantages such as avoiding single server failure, reducing access response time, reducing communication overhead and balancing load, but it unavoidably introduces well-known consistency issues. Data updates may be issued simultaneously on different replicas in optimistic consistent systems and they may reach other replicas in an arbitrary order. While being intuitive, two updates conflict when a replica issues an update before it receives another update that is already circulating among replicas. However, the two updates do not conflict radically if each update operates just on different segment of data object. Additionally, if the outcome of one of the two updates changes the condition of another, eventual data consistency by sequentially applying them breaches user view virtually. All the above questions are brought by data dependence in data consistency, which is inevitable for describing updates by semantic and often needs to be resolved manually. In this paper, based on data dependence in data consistency in P2P distributed storage system, the optimistic data consistency method is proposed. It partitions large object into data blocks, uses Bloom filter compressed representation technique to summarily represent updates, and treats blocks as management unit to detect and reconcile conflicts. In recent years, Peer-to-Peer (P2P) computing has become a popular network computing paradigm. Researches on and applications of P2P computing have spread into many fields. P2P distributed storage systems, constructed by P2P computing technique, can offer data sharing and storage services for massive users and massive data. Data replication, one of the crucial technologies for managing massive data, can improve data availability and data access performance, but bring about problems of maintaining data consistency. Compared with other distributed systems, Peer-to-Peer systems exhibit some special characteristics, such as large-scale, dynamic, and heterogeneity, which has brought many challenges for data consistency maintenance technique in P2P distributed storage systems. Based on characteristics of massive data and P2P systems, an intensive study is conducted of data consistency maintenance technique for P2P distributed storage systems, including a multi-replica clustering management method based on limited-coding, an optimistic data consistency maintenance method to improve update propagation and reduce the space overhead of write-log and an optimistic data consistency maintenance method based on key-attributes. The relevant papers of study have been published.