DATA REPLICATION IN A DATABASE MANAGEMENT SYSTEM

摘要

According to an aspect, a database management system (DBMS) is operated to perform data replication. The DBMS includes a source database and a target database. The source database includes a plurality of first tables respectively comprising one or more first columns, and the target database includes a plurality of second tables respectively comprising one or more second columns. Each of multiple ones of the first columns is mapped to one of the second columns, and each one of the second columns is adapted to store replicated data changes of the first column mapped to the second column. Each of the second columns has an assigned update time, the update time being a moment in time when the second column most recently received incremental data changes from its mapped first column. The DBMS operates to process requested database statements while providing data replication.

主权项

1. A computer-implemented method for operating a database management system, the database management system comprising a source database and a target database, the source database comprising a plurality of first tables respectively comprising one or more first columns, the target database comprising a plurality of second tables respectively comprising one or more second columns,
wherein each of multiple ones of the first columns is mapped to one of the second columns, each one of said second columns being adapted to store replicated data changes of the first column mapped to said second column, wherein each of the second columns has an assigned update time, the update time being a moment in time when said second column most recently received incremental data changes from its mapped first column, the method comprising: the database management system receiving a request to perform a requested statement, the statement being a database statement; the database management system predicting one or more of the second columns that will be accessed when processing the requested statement; the database management system determining, for each of the predicted second columns, its assigned update time; the database management system identifying a syncpoint time, the syncpoint time being the most recent update time of all said determined update times; the database management system identifying a starting time, the starting time being the earliest update time of all said determined update times; based on the identified starting time differing from the identified syncpoint time, the database management system selectively and incrementally replicating data changes from all first columns mapped to any one of the predicted second columns to said predicted second columns such that all data changes imposed on said first columns between the starting time and the syncpoint time are replicated to said predicted second columns, wherein subsequent to said replication, the update time of each of said predicted second columns is set to the syncpoint time; the database management system delaying execution of the requested statement until a successful termination of said replication; and subsequent to said successful termination, the database management system executing the requested statement in the target database.