| In this thesis,a set of components for Kubernetes is implemented to support some features of compute offloading technology on containerized cloud platform for the respective technical features of compute offloading technology and Kubernetes cloud platform.In the requirement analysis and design phase,this thesis provides a detailed analysis and design of the functional requirements of the compute offload component,such as obtaining parameters of task time delay and task energy consumption,task scheduling,task offloading granularity and mode of computation offloading.In the system architecture design,this thesis divides the whole system into monitoring component,task scheduling component and computation offloading component,where the monitoring component includes task reporting component,container monitoring component,node performance monitoring component and node performance evaluation component? The task scheduling component mainly implements the functions of scheduling policy,resource management and task status management? The computational offloading component includes the offloading decision component and the offloading execution component.By implementing these components,this thesis proves in practical tests that it can improve computational efficiency and resource utilization.In the test analysis,the performance of the compute offload system is tested from various perspectives such as comparing the performance of different tasks,different load balancing policies,different number of nodes and different concurrency degrees,and the results show that the compute offload system can have good computational efficiency and resource utilization while avoiding overloading of individual nodes.The research results of this thesis can provide a reference for compute offloading technology on containerized cloud platform,and also provide new ideas for the implementation of distributed computing. |
