Orchestration Software for Resource Constrained Datacenters: an Experimental Evaluation

Abstract

The evolution of the cloud-computing technology has allowed the instantiation of resources almost anywhere. Handheld devices, edge/fog resources, and core cloud datacenters comprise a resource continuum that can be used for hosting almost any service. The rise of micro-services has allowed any application to be hosted over any type of compute resource, regardless of the underlying hardware architecture. In this work, we focus on the far-edge devices that participate in the resource continuum, located at the network access or the fog, and are usually resource constrained. We evaluate two lightweight frameworks which can be used for orchestrating micro-services on top of them. Our evaluation presents experimental evidence in terms of their capabilities for instantiating/tear-down of network services, and their dynamic adaptation to external workloads by using the respective horizontal scaling solutions, when tested under the same experimental environment. © 2022 IEEE.