Mobile databases have spread into today's communication and computing, with numerous applications relying on mobile technology. It is an ever growing field and very much in demand, thus proving to be an excellent area to carry out research. With a large amount of data, management of such data is the key issue. Existing policies consist of Message Digest algorithms used along with cache replacement algorithms. However, the Message Digest algorithms have a major drawback of data replication, leading to a large overhead on the memory constraints. This report discuses a data management policy for heterogeneous mobile databases. This policy involves a Synchronization Server, which contains only the most critical of data, instead of all the data as in case of a normal server. The synchronization server data is efficiently managed by classifying it under various states based on the frequency of data access, the time of data access, i.e., which data has been actively used over a period of time and which has not been used so. This aims to facilitate synchronization of data between heterogeneous databases using cache replacement and to improve the rate of synchronization between heterogeneous mobile databases and the database servers. This policy has shown better performance than Least Recently Used and Most Frequently Used algorithms in terms of the execution time.