Deadlock detection is really cheap
暂无分享,去创建一个
The treatment of deadlock has previously been considered as expensive, and implementations of the involved algorithm have appeared rather complex. In this paper a new algorithm for the treatment of deadlock is described. It is not only very efficient (in the worst case O(e) where e is the number of the edges in the “wait-for” graph) but also fairly simple. With this algorithm one can detect all deadlocks as soon as possible and list all participators in the corresponding cycle of the “wait-for” graph. The algorithm is described for the use of database transaction management.
[1] Hans-Jörg Schek,et al. Architecture and implementation of the Darmstadt database kernel system , 1987, SIGMOD '87.
[2] Jim Gray,et al. Notes on Data Base Operating Systems , 1978, Advanced Course: Operating Systems.
[3] David J. DeWitt,et al. Deadlock detection is cheap , 1983, SGMD.
[4] Catriel Beeri,et al. A Resource Class Independent Deadlock Detection Algorithm , 1981, VLDB.