| 摘要 |
Techniques are provided for replicating database changes among servers of a database system that effectively removes the inconsistency problems encountered after restoration from backup and due to removal and re-addition of partitions by assigning a new GUID to a server after it has been restored or after the partition has been re-added. This new GUID is used for identifying new changes made by the server after restoration or partition re-addition. By virtue of the use of the new GUID, new changes made after restoration will not be confused with any changes made by the server after the backup and before the restoration, which are identified by the old GUID of the server, thereby ensuring convergence of the servers' data through replication. Similarly, by virtue of the use of the new GUID, new changes made after re-addition of a partition will not be confused with any changes made prior to the re-addition of the partition.
|
