摘要
在分析嵌入式实时数据库数据和事务特征的基础上,讨论了嵌入式实时数据及事务的一致性约束,包括内部一致性、时间一致性(外部一致性和相互一致性).为满足各种一致性,针对不同类型的数据和事务给出了识时恢复的正确性准则.基于这些约束,设计了实时日志及相应的恢复算法.性能分析结果表明,所给出的恢复方法在减低超截止期事务比率、节省系统开销等方面都具有较好性能.
In this paper, through analyzing the characteristics of transactions and data in embedded real-time databases, various consistency constraints are discussed, including internal consistency, external consistency and mutual consistency. For guaranteeing these constraints, the correctness criteria of time-cognizant recovery for different classes of data and transactions are given. In terms of these criteria, a time-cognizant recovery scheme based on real-time logs is presented. It is shown by performance tests that the time-cognizant recovery approach has better performances in reducing the ratio of transactions missing deadlines and system overhead.
出处
《计算机学报》
EI
CSCD
北大核心
2007年第4期672-679,共8页
Chinese Journal of Computers
基金
国家自然科学基金(60673128)
中国国家博士后基金(2003034482)资助