Comparação de algoritmos para alocação de infraestruturas virtuais

With the emergence of Cloud Computing, companies began to adopt it due to its several advantages, such as low initial investment and high scalability. In this context, a provider of Infrastructure as a Service (IaaS) provides computer resources to its clients in a virtualized way. Users can perform requests of virtual machines to deploy and provide their applications. A recent proposal has emerged to improve the services of IaaS providers, enabling the allocation of a network of virtual machines and routers interconnected by virtual channels. In other words, a Virtual Infrastructure (VI). However, the task of allocating VIs on the physical substrate is complex, characterized as an NP-Hard problem, where optimal algorithms are impractical in real scenarios due to its exponential execution time. Considering this fact, this work extends the simulator CloudSim implementing four on-line algorithms to allocate VIs, and compare metrics such as acceptance ratio, substrate load and the distance between the virtual resources. These metrics represent the perspectives of providers and users, respectively.