Resource Allocation And Orchestration Strategy For Kubernetes Container Cloud

Currently,container is becoming the new technology standard in the era of cloud computing,and a large amount of services are developing and evolving from the traditional monolithic architecture based on virtual machine to cloud-native architecture based on container.Along with the growing scale of business on cloud platforms,the cost-effectiveness problem of container clouds is becoming more and more prominent.Resource utilization and default rate of Service Level Agreement(SLA)are two important indicators of the container cloud platform,which also directly determines the cost effectiveness of the container cloud platform.The dissertation focuses on research and optimization of container resource allocation and container orchestration strategy of container cloud platform,so as to effectively improve resource utilization while ensuring the SLA of container cloud platform.Speaking specifically,the thesis mainly completed the following work:Firstly,taking no consideration of dynamic requirement of service-oriented container,the current container cloud platform adopts static resource allocation strategy,which results in the underutilization of allocated resources and fragmentation of resources.To this end,the Periodicity SLA Utilization Balance(PSUB),the container resource allocation strategy,is proposed.This strategy combines the frequency domain and time domain analysis methods.It screens out the candidate period of container resource allocation through the frequency domain and further selects period through time domain analysis.Then,the cost-benefit model is further constructed to solve the optimal container resource allocation sequence.Compared with the current static allocation strategy,the results show that the PSUB strategy can effectively reduce the debris generated by the container resource allocation,and improve the resource utilization rate while ensuring the container service quality.Secondly,the traditional container orchestration strategy lacks the analysis of the real demand for service-oriented container resources,and also fails to consider the complementarity of the demand for periodic container resource in the time dimension,resulting in high cost and low benefit of container cloud platforms.A container orchestration strategy,PBFDR(PSUB Best Fit Decreasing Container Rebalancing)based on PSUB,is proposed.This strategy optimizes traditional orchestration strategies in three core steps: container sequence generation,container orchestration,and optimization and reorganization.By comparing with the MRP(Most Request Policy)?BFD(Best Fit Decreasing)and OBFD(Overcommit Best Fit Decreasing)orchestration strategies and PBFD's own control strategy,the results show that the PBFDR orchestration strategy can effectively reduce the cost of the container cloud platform while improving the resource utilization efficiency.To sum up,the thesis shows that the PSUB and PBFDR strategy proposed in this dissertation can not only effectively improve the resource utilization rate of the container cloud platform,but also avoid the risk of resource conflicts,which is of good practical significance.