The Internet of things (IoT) generates a plethora of data nowadays, and cloud computing has been introduced as an efficient solution to IoT data management. A cloud resource administrator usually adopts the replication strategy to guarantee the reliability of IoT data. This mechanism can significantly reduce data access time, and evidently, more replicas of data increase the data storage cost. Furthermore, the process of selecting mini clouds for replica allocation and sorting replicas in mini clouds is considered an NP-hard problem. Therefore, this paper proposes an approach based on the harmony search (HS) algorithm to allocate replicas to the IoT data in the cloud computing environment in order to mitigate the data access cost. The HS algorithm was employed in the proposed approach to determine the best location for data replication in the cloud computing environment. According to the implementation results, the proposed approach outperformed the other methods and managed to significantly decrease data access time and delay as well as energy consumption.