Feature-Based Cloud Provisioning for Rehosting

Misaki Mito, Takatoshi Ohara, Ryo Shimizu, Hideyuki Kanuka, Minoru Tomisaka

Abstract: As IT infrastructure is increasingly rehosted to public clouds, IT system development must become more efficient. Traditionally, configuring and provisioning cloud services to meet the same requirements as on-premises systems is a time-consuming manual process that is prone to human error. Therefore, we define a cloud feature model based on rehosting requirements and propose an automatic provisioning method for cloud environments in which engineers select model elements from the same design perspective as on-premises cases. We evaluated our method’s effectiveness in two projects for rehosting to an AWS environment. The results show that our method reduced man-hours by 20% in the design process and by 60% in the construction process compared with the conventional method. In the design process, our method eliminated the trial-and-error consideration of cloud services. In the construction process, our method helped introduce Infrastructure as Code, and the automatic matching function immediately increased productivity after provisioning. Therefore, our method improved efficiency in IT system development associated with rehosting.