On minimizing the resource consumption of cloud applications using process migrations

摘要: According to the pay-per-use model adopted in clouds, more resources an application running a cloud computing environment consumes, greater amount of money owner corresponding will be charged. Therefore, applying intelligent solutions minimize resource consumption is great importance. In this paper, we study problem identifying assignment scheme between interacting components application, such as processes and virtual machines, nodes system, that total consumed by respective minimized. Because centralized are deemed unsuitable for large distributed systems or large-scale applications, propose fully algorithm (called DRA) overcome scalability issues. DRA takes decisions concerning transition from one another dynamic way, based solely on local information. We also test two modifications basic deal better with heterogeneity servers terms capacity constraints. must note capture regarding network model. Through theoretical analysis, formally prove achieves convergence always provides optimal solution tree-based networks uncapacitated case. Moreover, through experimental evaluation up 55% cost reduction when compared most recent literature. show proposed improve algorithm's performance considerably case where have limited capacity.