The cloud community has quickly adopted Kubernetes, the de facto industry standard container orchestration system, in addition to containers in the past few years. All of the big cloud providers now offer Kubernetes-based Containers as a Service (CaaS). When CaaS is given to more than one independent user, or tenant, however, a multi-instance approach is used. In this case, each tenant gets its own separate cluster, which costs a lot of extra money because it uses virtual machines for isolation. If CaaS is going to be available in both the cloud and the edge cloud, where resources are limited, a different solution is needed. This paper presents an innovative classification of Kubernetes multitenancy into three methodologies: multi-instance via multiple clusters, multi-instance via multiple control planes, and single-instance native, based on the scientific literature. We suggest a single-instance multitenancy framework, which means that tenants are served from a single cluster's shared control plane. Our empirical findings indicate that the single-instance approach incurs significantly lower overhead compared to the other two methods. But it does mean that tenants share the compute nodes, which means that workload isolation is not as good. Even though this makes isolation weaker, there are ways to make up for it. We explain how our framework does this. The framework is free, open-source software that adds to Kubernetes and is available to everyone. Researchers are using it in the EdgeNet testbed for real work.
Mrs.K.Aparna et al. (Wed,) studied this question.