Existing high performance computing (HPC) systems are designed primarily for workloads requiring high rates of computation. However, the widening performance gap between processors and I/O systems, and trends toward higher data intensity in scientific and engineering applications, suggest there is a need to rethink HPC system architectures, programming models, runtime systems, and tools with a focus on data intensive computing. The Third International Workshop on Data Intensive Scalable Computing Systems (DISCS) builds on the momentum generated by DISCS-2012 and DISCS-2013, providing a forum for researchers and other interested people in the areas of data intensive computing and high performance parallel computing to exchange ideas and discuss approaches for addressing the challenges facing Big Data or data intensive computing at large scale.