Research Of Task Scheduling And Link Load Balancing In Cloud Data Centers

The rapid growth of cloud computing requires large-scale and scalable data centers as basic network support facilities to provide storage,computing and other resources for massive network services.Software-defined network,as a new type of network architecture that realizes decoupling of control plane and data plane,can overcome the defects of traditional network architecture and provide users with diverse business support.The important development trend of the future network is to integrate cloud computing,software-defined networking,and network function virtualization technologies into one for effective integration to better serve users.This article focuses on the challenges of limited computing resources in the cloud data center and the surge in network transmission traffic.It focuses on solving complex task scheduling due to limited resources,large task calculations,and diverse types,and heavy link load caused by the surge in network traffic.Other difficult issues.The specific research content and main innovations include the following two points:(1)Task scheduling in cloud computing: It involves how to use a set of necessary cloud resources to efficiently schedule task execution processes.In actual engineering,interrupting tasks and subsequent task recovery processes will inevitably increase overhead,as this will consume additional CPU time.This article considers a more general case where a job consists of interruptible and non-interruptible tasks with specific deadlines and resource requirements.Therefore,this paper designs an algorithm to minimize the overhead of task interruption while ensuring the task completion deadline.Specifically,an integer linear programming(ILP)was first designed for offline optimization,and then a heuristic algorithm was proposed for online task scheduling and compared with the optimized ILP model.The performance and feasibility of ILP and heuristic algorithms were verified by simulation experiments.(2)Link traffic load balancing: Based on the SDN management idea,uniformly regulate and control the traffic of the data center,consider the situation of offload,and design a load balancing routing algorithm to balance the data center link bandwidth utilization and realize the data center chain Load balancing and minimize system cost.Specifically,in the case where the maximum link bandwidth utilization of each link in the data center is minimized,the cost of traffic splitting is minimized;the two factors of maximum link bandwidth utilization and system cost are combined to establish a linear programming(LP)Optimize the model to design the transmission route;Finally,verify the performance of the LP algorithm through simulation experiments.