摘要
自Docker问世以来,微服务也得到了快速的发展,企业、组织等纷纷使用微服务架构进行容器化开发。为了管理数以万计的容器应用,各种容器编排框架应运而生,但容器调度过程中带来的能耗高、资源利用率低等问题非常显著。研究合理的容器放置,能有效的减缓此类问题。针对CPU、内存和带宽三类资源利用率低等问题,提出了容器多目标资源放置算法CMR(Container Multi-target Resource)。实验结果证明CMR算法能够将容器放置到与自身资源请求大小最吻合的虚拟机上,对比FF、LF、MF和RS算法能够同时节省CPU能耗34.0%,内存能耗33.8%,带宽能耗26.5%。
Since the advent of Docker,microservices have developed rapidly,and enterprises and companies have gradually begun to use the microservices architecture for container development. In order to manage tens of thousands of container applications,various container orchestration frameworks have emerged,which brings about high energy consumption and low resource utilization in the container scheduling process. Reasonable container placement can effectively alleviate such problems. We propose a container multi-target resource CMR(Container Multi-target Resource) for the low utilization of CPU,memory and bandwidth. The experimental results show that the CMR algorithm can place the container on the virtual machine that best matches the size of its own resource request. Compared with the FF,LF,MF and RS algorithms,it can save CPU power consumption by 34.0%,memory energy consumption by 33.8%,and bandwidth energy consumption by 26.5%.
作者
曹成成
李志聪
CAO Chengcheng;LI Zhicong(School of Computer Science Technology and Information Engineering,Harbin Normal University,Harbin 150025,China)
出处
《智能计算机与应用》
2020年第4期50-53,56,共5页
Intelligent Computer and Applications
基金
国家自然科学基金项目(61202458,61403109)
黑龙江省自然科学基金项目(F2017021)
关键词
容器
多目标
资源利用率
能耗
Container
Multi-objective
Resource utilization
Power consumption
