A study on alternative container-based technologies for virtualization of components deployment in software product lines

Abstract

The application containerization approach allows creating virtualization environments that could be used as a code. It allows running application in the isolated container that could be reproduced on any other hardware or cloud environment. One of the benefits of the containerization approach is the possibility to allocate necessary hardware resources like a RAM, CPU and storage. An approach to support agile development of software product lines (SPL) by using variability management techniques within the framework of the Scrum methodology has been proposed in the article. The main goal of the work is to analyze containers for virtualization of the runtime environment when deploying SPL. The information base for the proposed approach to managing the variability of deployment has been structured. The role of the approach in the general method of Scrum has been shown, and a conceptual diagram of the management process at the stage of application deployment has been proposed. The experimental analysis has been carried out and metrics for two types of containers, Docker and Vagrant, have been calculated. The following two metrics, namely, portability and productivity, for both containers have been analyzed. These metrics for the test component software solution have been calculated and executed in cloud environment with different configurations.  The portability metric indicates how easily the application can be migrated to other platform basing on the time required to start container with the application. The second metric is the time necessary for the same operations in different container.