Abstract

In this talk, I will present our ongoing research about evaluating policies to update- or reconcile master/slave replicas.

Master/slave replication is a situation in which the same data is correctly represented by a master replica and possibly erroneously represented by a slave replica. Typically, the slave replica is used for several business operations while the master replica is not. For example, in inventory management, the actual stock in the warehouse is the master copy and the stock registered in the information system is the slave copy. Even though the actual stock (master) is correct by definition, it is the stock registered in the information system (slave) that is used for business operations.

When organisations possess master/slave replicas, they should make sure that both replicas are in sync by following and implementing update- or reconciliation policies. If master/slave replicas are not in sync, inconsistencies in the slave replica could have a major impact on business operations.

Naturally, before update- or reconciliation policies are actually implemented, their effectiveness should be evaluated using, for example, simulation. An important component of simulation is a simulation model.

However, to the best of our knowledge, a general model to simulate update- or reconciliation policies is lacking from the literature.

Therefore, the aim of this research is to create such a simulation model and to demonstrate its applicability and utility by (1) making sure that it can simulate every aspect of update- and reconciliation policies, and (2) by applying it to simulate several candidate policies to reconcile master/slave replicas at YH Global and KBC Group NV.