Delay-based incrementally mapping of virtual machines in cloud computing systems
Chin‐Fu Kuo, Hsueh‐Wen Tseng · 2014
The aim of this paper is to investigate the assignment problem of a cloud computing system where user requests to run virtual machines continuously arrive and the associated virtual machines (VMs) must be hosted on physical machines (PMs). Each PM has limited resource amounts for four kinds of resources, i.e., CPU, disk, memory, and network bandwidth. Also, each VM has the resource requirements for the four kinds resources in a PM. We propose an on-line VM-to-PM assignment algorithm, the Delay-based Incrementally Mapping Algorithm (DIMA), to assign VMs to PMs. We attempt to reduce resource wastes on running PMs such that the number of PMs can be minimized. The algorithm uses a user-defined variable to control the number of VMs handled at some time unit. The corresponding VM for an arrival VM request does not be host by a PM until the number of waiting VM requests reaches the value of the user-defined variable. A series of experiments were conducted to evaluate the proposed algorithm. The experimental results demonstrate that the performance of the proposed DIMA scheme is better than the greedy algorithm.