OaaS Based on Temporal Partitioning with Minimum Energy Consumption

Abstract

Cloud computing is an economical solution for industry which is highly scalable and useful of virtualized resources that can be used on demand. It will have a significant impact on companies with the introduction of orchestration platforms as a Service (OaaS) to perform the services that support a variety of business processes such as BPEL. It's becoming an adoptable technology for many of the organizations, thanks to its flexibility and because it reduces total cost of ownership. Thus, an effective OaaS must meet several requests simultaneously; ensuring scalability and optimizing the use of shared resources in order to minimize energy consumption. In this paper, we will investigate three issues i) exploiting the minimum of resources to execute a maximum number of processes, ii) Preventing possible overload to the server, and iii) minimizing dynamic energy consumption which becomes one of the main challenges for large-scale computing, such as in cloud data center. As a solution for these challenges, we propose to use Workflow partitioning technique and this based on temporal dynamic reconfiguration approach. Our work aims to reduce the dynamic energy consumption; especially in communication buffers between partitions of BPEL process during partitioning. The proposed approach is based on two main steps: 1) Estimate the energy consumption of BPEL processes 2) Temporal and dynamic partitioning of BPEL process based on reconfigurable architecture in order to minimize overall energy consumption on each BPEL process.