PREVENTION OF IMPLICIT DEADLOCK BY OBJECT REALLOCATION
スポンサーリンク
概要
- 論文の詳細を見る
Most database systems and distributed database systems have adopted locking schemes as the synchronization method of interleaved and concurrent executions of transactions. Problem is how to resolve deadlock. In the conventional systems, a deadlock manager detects a deadlock by constructing a wait-for graph and finding a directed cycle in the wait-for graph. The deadlock is resolved by aborting a process in the deadlock cycle. If the objects held by the selected process are arbitrarily allocated to the other process, another deadlock may occur since the other process which has waited on object. We say such deadlock which is caused by aborting the selected process an implicit deadlock. In this paper, we discuss not only how to detect the deadlock and select a process in a deadlock cycle but also how to allocate the objects held by the selected process to the processes which waited on them so as not to cause the implicit deadlock.
- 一般社団法人情報処理学会の論文
- 1990-03-14
著者
関連論文
- グループウェアの課題と展望 / 平成4年度前期第44回全国大会報告 ( パネル討論会)
- Distributed Checkpointing and Rollback
- INTEGRATION OF DATABASE SYSTEMS AT THE NAVIGATIONAL LEVEL BY USING PROLOG
- SEMANTIC CONCURRENCY CONTROL
- PREVENTION OF IMPLICIT DEADLOCK BY OBJECT REALLOCATION
- PROLOG INTERFACE SYSTEM ON DISTRIBUTED NAVIGATIONAL DATABASE SYSTEMS
- RELIABLE BROADCAST COMMUNICATION
- Totally Ordering(TO)and Partially Ordering (PO)Broadcast Protocol