基于SPN模型的可生存性DBMS中恶意事务修复算法的研究

SPN Model based Malicious Transaction Repair Algorithms in Survivable DBMS

在传统的数据库恶意事务修复方案的基础上,采用Petri网模型分析事务撤销冲突和操作执行序列异常检测;进而结合可生存性DBMS特征提出恶意事务静态和on-the-fly修复算法,并在此基础上给出随机Petri网恶意事务修复模型;在分析恶意事务修复随机Petri网模型和连续时间Markov链的一致性后,给出了连续时间Markov链的恶意事务修复模型求解.

Malicious transaction immediate repairing is an important aspect in building survivable DBMS. Based on traditional malicious transaction repairing solutions, this paper solves transaction undo collision and incorrect executing sequence of transaction operations using Petri net models. Considering characteristics of survivable DBMS, algorithms of static and on-the-fly malicious transaction repairing are provided. Further, related repairing system models are put up based on stochastic Petri net models. After consistency analysis of malicious transaction repairing stochastic Petri net model with continuous time Markov chain, this paper provides malicious transaction repairing solutions using continuous time Markov chain.