In recent years we have witnessed a Cambrian explosion in database research. Column-stores, NoSQL, MapReduce have broken many of the assumptions present in traditional (relational) database systems. In this talk, we start by looking briefly at the early days of database research, making parallels with the current state of affairs and drawing important lessons. We will then discuss a few selected database research topics, including some lesser-known techniques, discussing their impact given computer architecture trends as well as new application requirements. In particular, we will show how a few important facts about modern hardware design are conditioning the architecture of database systems, speculating how future database systems are expected to come about.
Miguel Branco is a postdoctoral researcher at the Data-Intensive Applications and Systems Lab at EPFL. Between 2003 and 2009, he worked at CERN developing the distributed data management system of the ATLAS Experiment. At EPFL, his work focuses on adapting data management technology to computer architecture trends, where he has co-developed techniques to efficiently run transactional database systems in heterogeneous multisocket/multicore machines, as well as novel database architectures for "in situ" query processing. He received a Ph.D. in Computer Science from the University of Southampton in 2009.