Data replication optimization using simulated annealing

Data replication is ubiquitous in a large organization where multiple IT systems need to share information for their operation. This function is usually fulfilled by an enterprise replicating software that is dependent on the configuration that the IT administrator sets. The setup specifies the tables and routes, but it may not be optimum to meet the workload, leading to replication’s lag and bottlenecks. This paper proposes an approach to solving the configuration optimization problem for the data replication software with the simulated-annealing based heuristic. Empirical results show that the configuration setting enables the replicating software to perform at least 5 times better than the baseline configuration set achieved by this approach.