摘要 |
PURPOSE: A database management device of a standby node and an active node in a main memory database management system and a method thereof are provided to prevent data mismatch caused by transaction loss by supporting the data recovery of a defective node based on dual information. CONSTITUTION: A memory logging unit records a database log time-sequentially allocating identification numbers to a log buffer of a memory of an active node. A replicating unit (115) replicates the database log to a standby node (120). A disk storage unit (116) periodically stores the database log in a disk of the active node. A recovery unit (117) transmits an identification number of the final recovery log to the standby node and receives all replicated logs belonging to a transaction including lost logs from the standby node in order to recover database of the active node. [Reference numerals] (110) Active node; (111,121) Memory; (112,122) Disk; (113,123) Communication interface; (114,124) Memory logging unit; (115) Replicating unit; (116,125) Disk storage unit; (117,127) Recovery unit; (118,128) Recovery information managing unit; (120) Standby node; (126) Log storage reporting unit |