Minimizing container-based applications sla violations on edge computing environments

• Ano de defesa: 2022
• Autor(a) principal: Kayser, Carlos Henrique
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul; Escola Politécnica; Brasil; PUCRS; Programa de Pós-Graduação em Ciência da Computação
• Palavras-chave em Português: Computação na Borda; Aplicações Baseadas em Contêineres; Orquestração de Contêineres; Escalonamento de Contêineres; Edge Computing; Container-Based Applications; Container Orchestration; Container Scheduling.; CIENCIA DA COMPUTACAO::TEORIA DA COMPUTACAO
• Link de acesso: https://tede2.pucrs.br/tede2/handle/tede/10786

The emergence of applications with strict requirements such as low latency and privacy motivated the approximation of computing resources and users at the network’s edge due to the difficulties of the cloud computing paradigm in fulfilling such needs. In this new distributed computing paradigm, like cloud computing, container-based virtualization techniques are also considered for application provisioning due to low resource consumption, fast provisioning, and low storage footprint compared to virtual machines (VM). However, the high variability of the edge nodes’ computational capacity and bandwidth directly impact the applications’ provisioning time in an edge computing environment. In addition, the end-users location is also an important factor to consider when scheduling applications, as the distance between end-users and edge nodes impacts communication latency. In this context, this work presents a novel scheduling algorithm, called Latency and Provisioning Time SLA Driven Scheduler (LPSLA), which coordinates application provisioning on edge infrastructures to minimize latency and provisioning time Service Legel Agreements (SLA) violations. It considers the latency between the end-users location and edge nodes and the capacity of edge nodes in downloading the container-based applications. As a result, the proposed solution is capable of minimizing the SLA violations in all evaluated scenarios.