文章提出的恢复子系统模型是根据差分日志记录的特点,将已提交事务的日志经并行日志管理器收集后并行写入到多个磁盘,通过“协调者”全局检验点和站点局部检验点的协调操作,具有更高的并发度和更高的系统吞吐量,并且能快速地支持系统失败恢复.经过验证试验表明该子系统有效地减少了日志信息量,缩短了系统的恢复时间,提高了系统的性能.
Introduces a recovery subsystem model based on logical logging. The system writes log which has committed transaction into multiple disk through the parallel log manager. In the control of global and local checkpoint of coordinator,it has a higher degree of currency and higher system throughput, and can support system to recover fast. Experiments validate that this system reduces the amount of logging information effectively, shortens recovery time of system and improves the performance of the system.