Research On Container Cloud Scheduling Algorithm Based On Kubernetes

Due to the high flexibility and short deployment time of containers,more and more computer companies and Internet companies choose container cloud as a business deployment platform.Among many container cloud platforms,Google 's Kubernetes container cloud platform has emerged and occupied The mainstream of the container cloud field.Among the many components of Kubernetes,the scheduler is responsible for the scheduling of Pods.It is a key role that plays a vital role in the good operation of the platform.The existing Kubernetes scheduling strategy is too simple,and the resource requirements considered are not comprehensive enough.When triggering preemptive scheduling,only the priority of the Pod is considered,which is not conducive to the rational use of the platform's resources.In view of the above problems,this paper studies the resource scheduling and preemption scheduling in Kubernetes environment,and proposes an improved fireworks algorithm combined with simulated annealing algorithm,and uses the fireworks algorithm CLFWA that considers the platform load to improve the Kudnetes platform's Pod scheduling optimization strategy,The search power of the firework algorithm is linked to the platform nodes.At the same time,the use of multiple resources is considered.When the platform is under low load,the algorithm can search widely.When the platform is under high load,the local search ability of the algorithm is improved,and the accuracy of the algorithm is improved.degree.Aiming at the problem that the preemption scheduling mechanism is too simple,this paper uses CLFWA algorithm to analyze the historical resource usage data of the deported Pod,which reduces the impact of Pod deportation on the platform.In order to improve the firework algorithm and make it more suitable for the field of resource scheduling,this paper improves the iterative method of the firework algorithm,and also makes the firework explosion range non-linearly decrease with the number of iterations.Finally,the simulated annealing algorithm is combined with random The steps of selecting sparks are intended to improve the accuracy of the algorithm and the speed of calculation.This article uses Golang programming to experiment with the improved algorithm.A variety of benchmark functions are used to compare the improved firework algorithm with other typical algorithms.The experimental results show that the improved firework algorithm has better optimization and robustness in all benchmark function tests than the original firework algorithm and simulated annealing.Algorithm and enhanced firework algorithm.This paper uses the ContainerCloudSim simulation platform to verify the effectiveness of Kubernetes resource scheduling strategy based on the improved firework algorithm and Kubernetes preemptive scheduling strategy based on the improved firework algorithm.The experimental results show that the resource scheduling strategy based on the CLFWA algorithm is more balanced than the original Kubernetes resource scheduling strategy;the preemptive scheduling strategy based on the CLFWA algorithm makes the platform more stable and uses more resources than the original algorithm full.