System deployment of a computer environment plays a critical role in the daily administration of computer systems, and tasks of massive deployments may take a lot of time for data center administrator(s). The existing solutions for massive deployment normally consist of storage spaces and extra computer servers for running deployment services. Existing solutions of multicast massive deployment are not robust because the overall deployment performance will worsen if one client machine fails. Because of the limitation of the network protocol, a multicast solution is not scalable. Although scholars have proposed solutions based on BitTorrent (BT) to overcome performance and scalability problems, solutions are not good enough because they still require the storage space to save the image file. In this paper, we present a novel mechanism of massive deployment called 'BT deployment mechanism from the raw device' (BDMfRD), which differs from conventional solutions in that it avoids creating any image file in the deployment process or using external storage for it. The proposed solution was verified by conducting 10 experiments to replicate the 50 GB system of the source machine to 1-32 destination computers. Experimental results showed that the proposed method reduced the total time for deploying 32 computers by 45.289%. The implemented software is the first massive deployment solution that provides light, robust, efficient, and scalable capabilities simultaneously.
ASJC Scopus subject areas
- 工程 (全部)