The human society is increasingly dependent on computing systems, powering entertainment (e.g., Netflix), social (e.g., Facebook) and economic interactions (e.g., Amazon). The ubiquity and large-scale of these systems make hardware and software failures both common and inevitable. Keeping multiple identical replicas is the standard way to build resilient fault-tolerant systems, but incurs in considerable management and energy costs. This research addresses the dilemma of improving the efficiency of resilient systems while keeping the software systems correct and easy to develop and use. The project proposes a novel concurrency control protocol for replicated systems, exploiting some new hardware features, and a new transactional API, promoting the concurrent execution of operations in all replicas and reducing the conflicts between those operations. The project's results have the potential to improve the resilience, and reduce the energy consumption, of computer systems.
