Efficient dynamical system resource management method in cloud computing

The Journal of Engineering

Xiaoyu Shi¹  Shuai Wang²  Tianshu Wu² 

[1] Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences;College of Computer Science, Chongqing University;

关键词: resource allocation;    virtualisation;    cloud computing;    linear quadratic control;    optimisation;    virtual machines;    dynamical optimisation problem;    dynamical workloads;    robust workloads;    constrained boundary linear quadratic control method;    optimal resource allocation scheme;    system input;    system noise;    authors;    efficient dynamical system resource management method;    cloud computing;    virtualisation system;    efficient manage system resource;    optimal load;    different virtual machines;    single physical server;    dynamical web workloads;    system resource management problem;   

DOI  :  10.1049/joe.2018.9138

来源: DOAJ

【 摘 要 】

The resource management of virtualisation system is a core issue in the field of cloud computing. For achieving the goal of efficient manage system resource without performance loss in clouding environment, the authors propose an optimal load balancing controller to assign the resource to different virtual machines, which run on a single physical server. Here, the authors focus on dynamical web workloads that the system resource management problem can be formulated as a dynamical optimisation problem. In the face of dynamical and robust workloads, the authors present an alternative approach based on the constrained boundary linear quadratic control (cBLQC) method, the optimal resource allocation scheme is solved by minimising the quadratic cost function with constrains on the system input and output. Different with existing works that assume a particular distribution for system noise or ignore the noise directly, the authors loosen the restrictions on system noise, which simply assume that system noise belongs with some reasonable norm-bounded set. Experiments on the Xen-based platform with a set of workload patterns show that the efficiency of authors’ method in terms of control precision and stability.

【 授权许可】

Unknown