A Quasi-Metric lattice Approach for Access Prediction in Replicated Database Protocols

Replicated databases in WAN environments are an alternative to guarantee data availability for applications where there is an intense load of user operations and these come from diverse locations. Many of these applications show highly localized accesses and may take advantage of lazy or epidemic updates to send data changes where needed. In such a scenario our study makes use of the notion of a quasi‐metric lattice and a function, that represents the probability of access, as a prediction tool to mitigate the effect of worst cases in lazy‐update replication systems.