Cloud computing provide an efficient big data processing platform for many small and medium scale enterprises, how to replicate and allocate data in clouds is a critical problem influencing cost consumption for small and medium scale enterprises. Cloud data management systems mainly serve two kinds of workloads, one is read-intensive analytical workloads (e.g. OLAP), the other is write-intensive transactional workloads (e.g. OLTP). It is essential to minimize data management costs like storage, communication bandwidth, update and power with guaranteeing the service level agreements. Toward two workloads, a cost-effective data replica placement approach for minimizing data management costs on cloud computing centers is proposed. The definition of different data management costs is identified first, then we construct the cost optimization model of the data replica placement problem. The paper proposes a hybrid genetic algorithm and a data support-based initialization method that addresses the problem. Experiments show that the approach result in significant reduction in total data management cost and the algorithm is with good performance.