A Virtual Machine Scheduling Strategy with a Speed Switch and a Multi-Sleep Mode in Cloud Data Centers

Abstract

With the rapid growth of energy costs and the constant promotion of environmental standards, energy consumption has become a significant expenditure for the operating and maintaining of a cloud data center. To improve the energy efficiency of cloud data centers, in this paper, we propose a Virtual Machine (VM) scheduling strategy with a speed switch and a multi-sleep mode. In accordance with the current traffic loads, a proportion of VMs operate at a low speed or a high speed, while the remaining VMs either sleep or operate at a high speed. Commensurate with our proposal, we develop a continuous-time queueing model with an adaptive service rate and a partial synchronous vacation. We construct a two dimensional Markov chain based on the total number of requests in the system and the state of all the VMs. Using a matrix geometric solution, we mathematically estimate the energy saving level and the response performance of the system. Numerical experiments with analysis and simulation show that our proposed VM scheduling strategy can effectively reduce the energy consumption without significant degradation in response performance. Additionally, we establish a system utility function to trade off the different performance measures. In order to determine the optimal sleep parameter and the maximum system utility function, we develop an improved Firefly intelligent searching Algorithm.

Keywords

Cloud data center / virtual machine scheduling / speed switch / multi-sleep / matrix geometric solution / utility function / improved Firefly Algorithm

References

[1] Banik A, Gupta U, Pathak S. On the GI/M/1/N queue with multiple working vacations-analytic analysis and computation. Applied Mathematical Modelling, 2007, 31(9): 1701-1710.

[2] Chen G, Xia W, Shen L. Dynamic bandwidth allocation algorithm based on transmission rate adaptation. Journal of Communications, 2014, 35(5): 25-32.

[3] Chen Y, Chang M, Liang W, Lee C. Performance and energy efficient dynamic voltage and frequency 16 Jin et al.: A Virtual Machine Scheduling Strategy with a Speed Switch and a Multi-Sleep Mode in Cloud Data Centers scaling scheme for multicore embedded system. International Conference on Consumer Electronics, 2016

[4] Chou C, Wong D, Bhuyan L. DynSleep: Finegrained power management for a latency-critical data center application. International Symposium on Low Power Electronics and Design, 2016

[5] Dabbagh M, Hamdaoui B, Guizani M. Toward energy-efficient cloud computing: Prediction, consolidation and overcommitment. Network IEEE, 2015, 29(2): 56-61.

[6] Dabbagh M, Hamdaoui B, Guizani M, Rayes A. Energy-efficient resource allocation and provisioning framework for cloud data centers. IEEE Transactions on Network and Service Management, 2015, 12(3): 377-391.

[7] Duan L, Zhan D, Hohnerlein J. Optimizing cloud data center energy efficiency via dynamic prediction of CPU idle intervals. International Conference on Cloud Computing, 2015

[8] Farahnakian F, Ashraf A, Pahikkala T. Using ant colony system to consolidate VMs for green cloud computing. IEEE Transactions on Service Computing, 2015, 8(2): 187-198.

[9] Gandomi A, Yang X, Alavi A. Mixed variable structural optimization using Firefly Algorithm. Computers and Structures, 2011, 89(23): 2325-2336.

[10] Gao P, Curtis A, Wong B, Keshav S. It’s not easy being green. ACM SIGCOMM Computer Communication Review, 2012, 42(4): 211-222.

[11] Greenbaum A. Iterative Methods for Solving Linear Systems, 1997.

[12] Hameed A, Khoshkbarforoushha A, Ranjan R, Jayaraman P, Kolodziej J. A survey and taxonomy on energy efficient resource allocation techniques for cloud computing systems. Computing, 2016, 98(7): 751-774.

[13] Hintemann R, Clausen J. Green cloud? The current and future development of energy consumption by data centers, networks and end-user devices.. International Conference on ICT for Sustainability, 2016

[14] Latouche G, Ramaswami V. Introduction to Matrix Analytic Methods in Stochastic Modeling, ASA-SIAM Series on Statistics and Applied Probability, 2000

[15] Li K. Improving multicore server performance and reducing energy consumption by workload dependent dynamic power management. IEEE Transactions on Cloud Computing, 2016, 4(2): 122-137.

[16] Liao D, Li K, Sun G, Anand V, Gong Y, Tan Z. Energy and performance management in large data centers: A queuing theory perspective. International Conference on Computing, Networking and Communications, 2015

[17] Liu J, Jin S, Yue W. A novel adaptive spectrum reservation strategy in CRNs and its performance optimization. Optimization Letters, 2018, 12(6): 1215-1235.

[18] Neuts M. Matrix-Geometric Solutions in Stochastic Models, 1981.

[19] Qavami HR, Jamali S, Akbari M, Javadi B. Dynamic resource provisioning in cloud computing: A heuristic Markovian approach. International Conference on Cloud Computing, 2014

[20] Salimian L, Esfahani FS, Nadimi-Shahraki MH. An adaptive fuzzy threshold-based approach for energy and performance efficient consolidation of virtual machines. Computing, 2012, 98(6): 641-660.

[21] Shen Y, Bao Z, Qin X, Shen J. Adaptive task scheduling strategy in cloud: When energy consumption meets performance guarantee. WordWideWeb-Internet andWeb Information systems, 2017, 20(2): 155-173.

[22] Tian N, Zhang Z. Vacation Queueing Models Theory and Applications, 2006.

[23] Wang Y, Xie Q, Ammari A, Pedram M. Deriving a near-optimal power management policy using modelfree reinforcement learning and Bayesian classification. Design Automation Conference, 2011

[24] Yu S, Zhu S, Ma Y, Mao D. Enhancing firefly algorithm using generalized opposition-based learning. Computing, 2015, 97(7): 741-754.

[25] Zhang M, Hou Z. Steady state analysis of the GI/M/1/N queue with a variant of multiple working vacations. Computers and Industrial Engineering, 2011, 61(4): 1296-1301.