Performance-Power Tradeoff in Heterogeneous SaaS Clouds With Trustworthiness Guarantee
Xinghui Zhu,Zijie Di,Qingsong Yao,Xuewen Dong,Jiandong Wang,Yulong Shen
DOI: https://doi.org/10.1109/tc.2022.3214626
IF: 3.183
2023-05-12
IEEE Transactions on Computers
Abstract:Software-as-a-service (SaaS) clouds grow dramatically due to cost-effectiveness, availability, and flexibility. Quality of service (QoS) and power, which represent performance and cost, respectively, are conflicting yet critical issues in the service scheduling of SaaS clouds, and some researchers have investigated the tradeoff between them. However, existing works do not involve QoS attacks in which untrusted service providers provide fake QoS values to absorb service requests, resulting in lower user experience and system profits. In this paper, we jointly consider the QoS performance, queue congestion, and energy consumption to formulate the performance-power tradeoff while considering QoS attacks. To address this NP scheduling problem, we propose a Lyapunov-based decomposition strategy that converts the original problem into three equivalent subproblems. By aggregating the solving strategies for the three subproblems, we develop the online service selection and trustworthiness management algorithm that optimizes the performance–power tradeoff while resisting QoS attacks. In addition, a light-weighted trustworthiness management strategy is designed to update trustworthiness values without storing large amounts of past information. Mathematical analyses and simulations demonstrate that our proposed control framework realizes detection and resistance of QoS attacks and a [O(1/V),O(V)] tradeoff between performance and power with a performance-power tradeoff parameter V.
engineering, electrical & electronic,computer science, hardware & architecture