云环境下基于改进蚁群算法的虚拟机批量部署研究

针对云计算中虚拟机批量部署问题,在定义虚拟机与服务器匹配距离的基础上,使用蚁群优化思路进行部署方案搜索,并有针对性地对蚁群算法进行了扩展改进。首先在蚁群算法随机比例规则中加入性能感知策略,以尽量避免将相同性能偏好的虚拟机部署在同一台服务器上,造成对硬件资源竞争的危险。同时增加了单一蚂蚁信息素更新规则,以减少错误先验知识对蚂蚁后续选择的误导。通过在CloudSim中的仿真实验,对算法参数选择进行了研究。与现有部署算法相比,本算法具有更好的系统负载均衡性能和资源利用率,以及比基本蚁群算法更快的收敛速度。

Research on Extended Ant Colony Optimization Based Virtual Machine Deployment in Infrastructure Clouds

Aiming at the virtual machine deployment problem in the cloud computing environment, based on the defining of the match-distance between the virtual machine and the server, the ant colony optimization(ACO) was used to re- search the deployment scheme. And the ACO was extended and modified for the deployment problem. Using the proba- bilistic tour decision with performance apperceive policy, the virtual machines with the same performance interest arc designedly placed in different servers to reduce the competition of the hardware resources. And using the single ant pheromone update rules, the misdirection of the inaccurate heuristic information is avoided. I}he parameter values for the arithmetic were researched with the experiments in C1oudSim. Finally, the performance of the extended ACC) was com- pared with that of the ranking deployment arithmetic and the original ACO. The experimental results show that the ex- tended ACO meets the need of the system load balancing better, and accelerates the convergence to the original ACO.