Conflicting transactions are required to be serializable to provide a consistent application service. In this paper, the SOR (Subject-Oriented with Read-data-from relation) serializability is newly proposed based on the significancy of roles granted to subjects and the read-data-from relation of transactions. Then, the EERO-RDF (Energy Efficient Role Ordering with Read-Data-From relation) scheduler is newly proposed to not only make transactions SOR-serializable but also reduce the electric energy consumption of servers. In the EERO-RDF scheduler, each read method reads the latest committed version of data in an object. Hence, each read method is not blocked by the other methods and the concurrency of transactions increases compared with the EERO (Energy Efficient Role Ordering) scheduler proposed in our previous studies. Evaluation results show not only the execution time of each transaction but also the electric energy consumption of servers are reduced by increasing the concurrency of transactions in the EERO-RDF scheduler compared with the EERO scheduler.