Modern web services are required to fulfill important non-functional criteria such as availability, responsiveness, scalability, and reliability, which are outlined in Service Level Agreements (SLAs). These agreements include Service Level Objectives (SLOs), which set performance benchmarks like up-time, latency, and throughput, crucial for maintaining consistent service quality. Failing to meet these SLOs can lead to penalties and harm to the provider's reputation. Additionally, service providers must avoid over-provisioning resources, as this can lead to excessive costs and inefficient use of resources. To address this, autoscaling mechanisms dynamically adjust the number of service replicas to match user demand, while scheduling policies establish a placement for each replica. However, traditional autoscaling solutions typically rely on low-level metrics (e.g., CPU or memory usage), making it difficult for providers to optimize both SLOs and infrastructure costs. Furthermore scheduling policies do not consider the run time resource contention between replicas that can have an impact on the application response time. This paper proposes an SLO-aware autoscaling methodology, along with load-aware scheduling and descheduling strategies, for containerized workloads in Kubernetes clusters, integrating SLOs in the container orchestration process. This approach overcomes the limitations of conventional orchestration policies by making more efficient decisions that balance service-level requirements with operational costs, offering a comprehensive solution for managing containerized applications and their infrastructure in Kubernetes environments. The results, obtained by evaluating a prototype of our system in a testbed environment, show significant advantages over the vanilla Kubernetes platform.

The orchestration of distributed microservices-based applications, particularly within geo-distributed Cloud-to-Edge environments, poses significant challenges. While Kubernetes stands as the predominant container orchestration standard in Cloud data centers, its static container scheduling approach presents limitations in deploying complex, distributed microservices-based applications across Edge environments. Presently, the scheduling process in Kubernetes fails to consider current infrastructure network conditions, resource usage, or runtime application statecrucial factors for mitigating the heterogeneous and dynamic nature of Cloud-to-Edge infrastructure and optimizing application response times. In this study, we propose an enhancement of the Kubernetes platform by implementing a load and network-aware microser-vices scheduling and orchestration strategy. The idea is to extend the Kubernetes control and scheduling logic with a dynamic orchestration strategy, continuously adapting application place-ment based on the real-time state of both the infrastructure and the application itself. We evaluate the efficacy of our approach by comparing it with the default Kubernetes orchestration and scheduling strategy.

Defining a scheduling and orchestration strategy for modern distributed microservices-based applications is a complex problem to deal with, especially if they are deployed on geo-distributed Cloud-to-Edge environments. Kubernetes is today the de-facto standard for container orchestration on Cloud data centers. However, its static container scheduling strategy is not suitable for the placement of complex and distributed microservices-based applications on Edge environments. Current infrastructure network conditions and resource availability neither run time application state are taken into account when scheduling microservices. To deal with these limitations in this work we present an extension of the Kubernetes platform in order to implement an effective application and infrastructure-aware container scheduling and orchestration strategy. In particular, we propose an extension of the default Kubernetes scheduler that considers application and infrastructure telemetry data when taking scheduling decisions. Furthermore, a descheduler component is also proposed that continuously tunes the application placement based on the ever changing application and infrastructure states. An evaluation of the proposed approach is presented by comparing it with the default Kubernetes scheduling strategy.

With the diffusion of Fog and Edge Computing paradigms, new application categories have emerged with specific quality of service requirements, in terms of communication latency and throughput. The placement of these applications on distributed Cloud-Edge environments is a challenging task, because of the continuously varying node-to-node network latency and bandwidth on Edge infrastructure. Although Kubernetes is the de-facto standard for container orchestration on Cloud data centers, its scheduling strategy is not suitable for the placement of time critical applications on Edge environments because it does not take into account current network conditions, neither communication requirements between microservices during its scheduling decisions. In this work we propose a network-aware scheduler plugin that extends the default Kubernetes scheduler, in order to deal with variable network conditions on cloud-edge clusters and run-time communication requirements of microservices. A custom descheduler is also proposed that periodically monitors run-time network state and traffic exchanged between microservices and evicts Pods from cluster nodes if better placement decisions can be done. An evaluation of our scheduling and descheduling strategies has been carried out on a sample microservices-based application deployed on a test bed environment.