Parsimonious Edge Computing to Reduce Microservice Resource Usage
share
Cloud Computing (CC) is the most prevalent paradigm under which services are provided over the Internet. The most relevant feature for its success is its capability to promptly scale service based on user demand. When scaling, the main objective is to maximize as much as possible service performance. Moreover, resources in the Cloud are usually so abundant, that they can be assumed infinite from the service point of view: an application provider can have as many servers it wills, as long it pays for it. This model has some limitations. First, energy efficiency is not among the first criteria for scaling decisions, which has raised concerns about the environmental effects of today's wild computations in the Cloud. Moreover, it is not viable for Edge Computing (EC), a paradigm in which computational resources are distributed up to the very edge of the network, i.e., co-located with base stations or access points. In edge nodes, resources are limited, which imposes different parsimonious scaling strategies to be adopted. In this work, we design a scaling strategy aimed to instantiate, parsimoniously, a number of microservices sufficient to guarantee a certain Quality of Service (QoS) target. We implement such a strategy in a Kubernetes/Docker environment. The strategy is based on a simple Proportional-Integrative-Derivative (PID) controller. In this paper we describe the system design and a preliminary performance evaluation.
READ FULL TEXT