Uma abordagem para a modelagem de desempenho e de elasticidade para Bancos de dados em nuvem

• Ano de defesa: 2016
• Autor(a) principal: Farias, Victor Aguiar Evangelista de
• Orientador(a): Machado, Javam de Castro
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Banco de dados; Computação em nuvem; Modelagem computacional; Cloud computing; Perfomance modeling
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/16374

Cloud computing is a successful, emerging paradigm that supports on-demand services. With the exponential growth of data generated by present applications, NoSQL databases which are inherently distributed systems have been used to manage data in the cloud. In this scenario, it is fundamental for cloud providers to guarantee Quality of Service (QoS) by satisfying tho Service Level Agreement (SLA) contract while reducing the operational costs related to both overprovisioning and underprovisioning. Thus QoS mechanisms can greatly benefit from a predictive model that estimates SLA-based performance metrics for a given cluster and workload configuration. Therewith, elastic provisioning strategies can benefit from these predictive models to provide a reliable mechanism to add and remove resources reliably. In this work, we present a generic performance modeling for NoSQL databases in terms of SLA-based metrics capable of capturing non-linear effects caused by concurrency and distribution aspects. Moreover we present a elastic provisioning mechanism based on performance models. Results of experimental evaluation confirm that our performance modeling can accurately estimate the performance under a wide range of workload configurations and also that our elastic provisioning approach can ensure QoS while using resources efficiently.