Proposta de dissertação do MEI Título: Big Data computations with partially replicated data Proponente(s): Nuno Preguiça Margarida Mamede Créditos: 42 ECTS Área científica: Computer Systems and Networks Início preferencial: Qualquer semestre URL: Já estão em curso trabalhos preliminares executados pelo alunos: Breve descrição: Big Data has become mainstream, as companies can extract valuable information from the data they store. As cloud providers offer services in multiple data centers (DCs) across the globe (e.g. Amazon currently has DCs in 13 locations), an increasing number of entities use multiple cloud locations. This improves service to clients, with lower latency and better availability and fault tolerance. However, to provide a good service is often sufficient to store data in a few locations, thus minimizing the cost of the infrastructure. Efficiently executing computations in such settings is challenging. In a previous work, we have proposed an algorithm to optimize the execution of computations with data located in different DCs [1]. In this work we intend to extend our previous work by considering that the same data can be replicated at multiple locations. This requires deciding which replica to use or if using multiple replicas can even lead to a better execution time. Observações: [1] http://www.gsd.inesc-id.pt/~rodrigo/p72-kloudas.pdf