Proposta de dissertação do MEI Título: Integrating Big Data Computations in a Geo-replicated Database Proponente(s): Nuno Preguiça João Magalhães 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 computations are usually run by specialized systems. Systems that execute batch processing, such as Hadoop and Spark, run periodically, reading data from the data store, executing the computation and storing data back to the store. With this approach, until the computations are run again, the results become outdated as new data is available. Additionally, it is complex to decide the best moment to execute computations. The AntidoteDB geo-replicated database we have been developing (http://www.antidotedb.org) has support to execute some computations (e.g. count, average, top-k) inside the store, with the state of an object being the result of a computation over all data in the storage. In the context of this work we want to study: (1) how this support can be used to execute a number of well-known Big Data algorithms; (2) improve the database as required for executing the previous algorithm efficiently. Observações: This work is executed in the context of GoLocal and EU LightKone projects.