Dynamic replica control based on fairly assigned variation of data with weak consistency for loosely coupled distributed systems
Tóm tắt
This paper proposes a replica control method based on a fairly assigned variation of numerical data that has weak consistency for loosely coupled systems managed or used by different organizations. This method dynamically distributes the variation of numerical data to replicas according to their demands while achieving fairness among them. By assigning the variation, a replica can determine the possibility that processed update transactions will be aborted and can notify a client of the possibility even when network partitioning happens. In addition, fairly, assigning the variation of data to replicas enables the disadvantage among replicas caused by asynchronous update to be balanced among replicas. Fairness control for assigning the variation of data is performed by averaging the demands in the variation that are requested by the replicas. Simulation showed that our system can achieve extremely high fairness while processing update transactions at the maximum rate.
Từ khóa
#Control systems #Availability #Scalability #Database systems #Delay #Telegraphy #Telephony #Computational modeling #Computer networks #HumansTài liệu tham khảo
helal, 1996, Replication Techniques in Distributed Systems
10.1109/90.365436
10.1145/185595.185651
10.1109/ICDCS.1993.287680
10.1109/RIDE.1992.227430
10.1109/69.476504
soparkar, 1990, Data-value partitioning and virtual messages, Proc ACM Symp Principles Database Systems
10.1145/224057.224070
10.1109/69.591451
10.1109/ICDCS.1998.679484
10.1109/ICDSC.2001.918944
10.1145/358468.358487
10.1007/978-1-4757-2491-2
10.1109/ICDSC.2001.918975
10.1109/MRD.1992.242627
10.1145/588111.588124
bersekas, 1987, Data Networks
10.1145/1994.2207
gray, 1996, The dangers of replication and a solution, Proc ACM SIGMOD International Conference on Management of Data, 173, 10.1145/235968.233330
10.1109/RELDIS.1999.805087
yu, 2000, Efficient numerical error bounding for replicated network services, Proc 20th International Conference on Very Large Databases, 123