A vector-based atomic broadcast protocol for deferred updates in replicated distributed databases

Deferred updates propagation enhances update performance and reduces transaction response time for software-based replication in distributed databases. However, it may also result in concurrency anomalies if not handled properly. In this paper, we investigate the replication models and propose one vector-based atomic broadcast protocol for deferred updates, which is superior to the global sequencer-based atomic broadcast protocol in robustness and efficiency. The strategy for dealing with the processing of the deferred updates and the strategy for determining the establishment of update stability are put forward. Two practical examples are also given to further exemplify the main ideas. Finally, the reach of consensus through group communication is discussed in brief. With the new approach, group members can detect and recover from communication failures and establish update stability. The research results are beneficial to other similar projects.