Failover operation on a replicated distributed database system while maintaining access invariance

摘要

Systems and methods are disclosed herein for performing a failover operation between a first distributed database cluster and a second distributed database cluster. In one aspect, the method includes blocking a plurality of software processes from making any new access requests to the first distributed database cluster and completing a current access request to the first distributed database cluster that was initiated by one of the plurality of software processes before the failover operation was initiated, waiting for data to replicate from the first distributed database cluster to the second distributed database cluster, wherein replicating includes any changes to the data from the current access request, and allowing the plurality of software processes to make the new access requests to the second distributed database cluster once the replicating has completed.

主权项

1. A method of performing a failover operation between a first distributed database cluster and a second distributed database cluster, the method comprising:
blocking a plurality of software processes from making any new access requests to the first distributed database cluster; completing a current access request to the first distributed database cluster that was initiated by one of the plurality of software processes before the failover operation was initiated; waiting for data to replicate from the first distributed database cluster to the second distributed database cluster, wherein replicating includes any changes to the data from the current access request; and allowing the plurality of software processes to make the new access requests to the second distributed database cluster once the replicating has completed; wherein the software processes each have a database abstraction layer that performs the blocking of the plurality of software processes from making any new access requests to the first distributed database cluster, the completing of a current access request to the first distributed database cluster, and the allowing of the plurality of software processes to make the new access requests to the second distributed database cluster.