Resolution of Parallel Deadlock by Partial Abortion
暂无分享,去创建一个
n A distributed system is composed of multiple objects intercon ected by communication networks. Each object is an abstract o data type. Users write transactions to manipulate distributed bjects in a nested fashion. Transactions are composed of o operations and also atomic units of work for users. Each peration can call operations on another objects. Suppose that one operation op on an object o calls two operations op on 1 1 2 p 3 2 3 and op on q. op and op can be called in parallel. This means that op can be executed in parallel. In this sense, nest 1 ed transactions can be executed in parallel in the distributed k p systems. In this paper, we would like to discuss deadloc roblems occurred when transactions are executed in parallel.
[1] Edgar Knapp,et al. Deadlock detection in distributed databases , 1987, CSUR.
[2] William E. Weihl,et al. Local atomicity properties: modular concurrency control for abstract data types , 1989, TOPL.
[3] Abraham Silberschatz,et al. A Formal Approach to Recovery by Compensating Transactions , 1990, VLDB.
[4] Jim Gray,et al. The Transaction Concept: Virtues and Limitations (Invited Paper) , 1981, VLDB.