Multiserver configuration for cloud service profit maximization in the presence of soft errors based on grouped grey wolf optimizer

Keywords Cloud computing; Multiserver configuration; Deadline miss rate; Soft error reliability; Profit Abstract With the growing demand of cloud customers for computing resources, cloud computing has become more and more popular. As a pay-as-you-go model, cloud computing enables customers to use cloud services on demand anytime, anywhere over the Internet and it has become the backbone of modern economy. Obviously, profit maximization is especially important for cloud service providers (CSPs) in a competitive cloud service market. Extensive research papers have been conducted during the past few years for CSPs to optimize cloud service profit, whereas few of them considers the transient faults (resulting in soft errors) that may happen during service requests' execution and thus cause failed execution of these requests. In this paper, we study the multiserver configuration problem for cloud service profit maximization considering the deadline miss rate of service requests and the soft error reliability of the multiserver system. To solve the profit optimization problem, we first construct the models of multiserver system, deadline miss rate, and soft error reliability. Based on these models, we derive the models associated with cloud service revenue and cloud service costs. Then, we formulate the cloud service profit optimization problem and propose an effective grouped grey wolf optimizer (GWO)-based heuristic method that can determine the optimal multiserver configuration for a given customer demand to maximize cloud service profit. Experimental results show that the cloud service profit improvement achieved by our scheme can be up to 33.76% as compared with a state-of-the-art benchmark scheme. Author Affiliation: (a) School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China (b) Shanghai AI Laboratory, Shanghai 200062, China (c) Ministry of Education Engineering Research Center of Software/Hardware Co-design Technology and Application, East China Normal University, Shanghai 200062, China (d) State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China (e) National Trusted Embedded Software Engineering Technology Research Center, East China Normal University, Shanghai 200062, China * Corresponding author at: School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China. Article History: Received 20 December 2021; Revised 11 April 2022; Accepted 12 April 2022 (footnote)[white star] This work was supported in part by the National Key Research and Development Program of China under Grant 2018YFB2101300, the National Natural Science Foundation of China under Grants 62172224, 61802185 and 61872147, in part by the Natural Science Foundation of Jiangsu Province, China under Grants BK20180470 and BK20190447, in part by the China Postdoctoral Science Foundation under Grants BX2021128, 2021T140327 and 2020M680068, in part by the Postdoctoral Science Foundation of Jiangsu Province, China under Grant 2021K066A, in part by the Open Research Fund of the State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences under Grant CARCHA202105, in part by the Future Network Scientific Research Fund Project, China under Grant FNSRFP-2021-YB-6, and in part by the Open Research Fund of the National Trusted Embedded Software Engineering Technology Research Center (East China Normal University ). Byline: Peijin Cong (a), Xiangpeng Hou (a), Minhui Zou (a), Jiangshan Dong (b), Mingsong Chen (c), Junlong Zhou [jlzhou@njust.edu.cn] (a,d,e,*)