A Time-Based bi-Objective Virtual Machine Placement Algorithm in Cloud Computing Platform

As the user's task arrives randomly in cloud computing platform, the system resource state and the task resource requirements are dynamic as well. Most of the existing scheduling algorithms are static so that they have low flexibility and energy efficiency. On the basis of the problems mentioned above, firstly, the task request dynamic simulation model was established by sampling Gaussian model of CPU and RAM, which could describe the tasks with different resource requirements and real-time changes. Based on the established model, to construct the virtual machine(VM) placement model under multi-constraints and bi-objective, the time-based genetic algorithm(T-Gene) was proposed, which introduced temporal dimension into the genetic algorithm. In order to get better performance on dynamic scheduling, fitness function was designed with the power consumption and the number of physical machines(PMs). Finally, comparing T-Gene algorithm with First Fit Descending(FFD), Random and Genetic Algorithm(GA), the power consumption decreased by 16.6% at most, the number of PMs used was reduced by 14.50% maximally. The results indicated that T-Gene algorithm can better resolve the configuration problem of virtual machines(VMs) under dynamically changing tasks.