Disaster recovery is a diligent issue in IT business. This issue is progressively significant in cloud computing, since Cloud Service Providers (CSPs) are bound to provide all facilities to their clients regardless of whether the server farm is down, because of a disaster. During the disaster, the data may be lost. To overcome this problem, replication is generated for each input data. The main objective of this paper is to upload different data on optimal location of cloud. The proposed system consists of three modules, namely, replica generation; choose optimal location and recovery process. Initially, to avoid the data loss, the input data are replicated. After replication process, the data are stored on cloud with the help of oppositional gravitational search algorithm (OGSA) which then retrieves only the request based data. Hence, we could avoid the data loss due to disaster. The presentation of proposed methodology is analyzed by different metrics comparing with various methods.