Data replication in a distributed database system is a crucial topic for enhancing data availability and improving the system faults tolerance, in the distributed database we think that clients use a single database yet, in fact, a set of databases are stored on multiple servers is used.Replication often plays an important role in database management strategy; organizations use database replication for load balancing, offline processing, redundancy, and other typical administrative tasks.In this paper we describe our experiences with using Event-B formal method and show how to conduct an assessment of system reliability at the refinement stages rather than at the implementation level; we demonstrate by case study how to obtain a mathematically proven solution using Event-B formal method. Here we propose using the Rodin modeling tool for Event-B that integrates modeling and proving together, and as a case study; we present data replication technologies in SQL Server databases.