Metodologia para avaliação da qualidade de experiência - QoE - de serviços em nuvem

• Autor(a) principal: Costa , Frederico Guilherme Irigoyen da
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da PUC_RS
• Texto Completo: http://tede2.pucrs.br/tede2/handle/tede/6356
• topic: INFORMÁTICA; COMPUTAÇÃO EM NUVEM; APLICAÇÕES PARA A WEB; INTERFACE COM O USUÁRIO; QUALIDADE EM INFORMÁTICA; ENGENHARIAS

In the information age, cloud computing has been touted as a revolutionary concept, since it enhances the quality of communication in a flexible environment, being highly cost-effective. Cloud services demand is supplied by an increasing number of cloud service providers (CSPs). This scenario has consolidated an environment of competition and pressure on prices, in which the perception of the user of the provided services has been gaining attention. In this competitive environment, the quality of experience (QoE) became a key factor not only in selecting cloud service providers, but also in defining the way of deployment services. In this context, this thesis proposes a methodology to evaluate the user quality of experience in cloud services, with focus on web applications, using the MOS (Mean Opinion Score), in a user-centered approach. The methodology estimates the QoE considering network, client application and server aspects. Moreover, it allows adjusting the user expectation according to the evaluation context.The proposed methodology has been applied to the development of the QoE Estimated Platform (PEQN). The implemented platform considers three cloud service providers, located in Brazil, Europe and USA. Several case studies, applied to different contexts of evaluation have been conducted. Case studies allow assessing different client application and server deployments, considering three server geographical locations. The results show significant variability in the MOS within the observed period, not only in the evaluated contexts but also in the assessed cases. Estimated QoE varies significantly even when the network parameters (latency, download rate, packet loss, etc.) remain stable. This result highlights the importance of QoE assessment by applying a methodology that is able to capture the performance variations not only regarding the network but also regarding the CSPs application and hardware. Finally, recommendations are presented, based on extensive literature research, methodology developed and obtained results, which could help on selecting the most suitable cloud providers.