Research On Scheduling Algorithms Based On Docker

With the rapid development and popularization of cloud computing,distributed computing technology has rapidly become the mainstream data processing technology of Internet companies in the era of massive information and this technology has been widely used in the routines of these companies.Generally speaking,Infrastructure Services(IaaS),Platform Services(PaaS)and Software Services(SaaS)constitute the classic three-level architecture of cloud computing,which can provide the demanding service.Within the three layers,the roles of each layer are clearly distinguished and each layer performs its own duties.With the continuous progress of virtualization technology,the emergence of container technology Docker has subverted its rigid classification,and makes the hierarchical boundaries between IaaS and PaaS no longer obvious,and has a tremendous role in promoting the development of its service form.Docker Swarm,the native cluster scheduling tool,is the most important solution to the Internet application delivery platform and cluster resource scheduling management.Therefore,it is very important to select an efficient and reliable scheduling algorithm to highlight the advantages of cluster.However,at present,the scheduling algorithm of Docker Swarm does not perform well when the resource distribution is uneven.The original three Scheduling Strategies of Docker Swarm,Spread,Binpack and Random,can not achieve better load balancing,and the resource utilization rate is low,and the overall performance of the cluster is poor.An improved ant colony optimization(Ant Colony Optimization,ACO)algorithm is proposed to optimize the resource allocation of the container cluster in Docker,which can not achieve better load balancing and low utilization of the cluster.To overcome the shortcomings of ant colony algorithm,Firstly,MIN-MIN scheduling strategy is used to initialize pheromones to reduce the search time.Secondly,by introducing equilibrium factor and coordinating the update mechanism of local pheromone and global pheromone,ant colony can decide the next round strategy according to the current iteration load and pheromone updatemechanism.At the same time,the adjustment mechanism of volatilization coefficient is introduced to improve the global search ability of the algorithm and optimize the utilization of resources of each node.By setting up the task of creating containers for cluster distribution,the original scheduling strategy was compared with the improved ant colony algorithm for many times experiemently.The results show that the improved ACO can better balance the resource utilization of each node in the cluster.In addition,when the cluster load pressure is high,the running time of the improved ACO is shorter than that of the native algorithm.