Abordagem icônica para modelagem e simulação de ambientes de computação em nuvem ?

• Autor(a) principal: Silva, Diogo Tavares da [UNESP]
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://hdl.handle.net/11449/138453; http://www.athena.biblioteca.unesp.br/exlibris/bd/cathedra/08-04-2016/000863022.pdf
• topic: Computação; Sistemas de computação virtual; Computação em nuvem; Metodos de simulação; Simulação (Computadores digitais); Computer science

The use of Cloud Computing is becoming increasingly more popular, driven by application's portability and the trend of reduction of investments in IT's physical infrastructure. These developments create scenarios where it is hard to know if the use of such environments is efficient or not. In this context, evaluate the performance of cloud computing systems is useful both for clients, that need to find the best resource configuration for their applications, as well as for providers, who need to evaluate which scheduling and resource and virtual machine allocation policies are more efficient. Although more accurate, the use of benchmarking is not an adequate option for this evaluation since it is expensive to use the system just for performance measuring. This makes simulation the most attractive option because it has a lower deployment cost and it is easier to reconfigure model parameters and reproduce the measurements in a system's model. Unfortunately, the known cloud computing simulators have issues related to their usability and modeling capability. This work presents the development of an iconic approach for modeling and simulation of cloud computing with the iSPD. The choice for iSPD as foundation for this project is justified because it is a simulator aimed to be user-friendly, offering an iconic interface to systems' modeling. Results achieved with cloud computing simulation show success, since the simulator was able to correctly mimic executions in a real cloud, with a reasonably low cost for modeling and execution