Special Issue on Greener Data Management for Ultra Large Scale Systems

摘要

The convergence of HPC and Big Data systems offer a great opportunity to improve the, energy efficiency when processing massive amount of data. For the past years, HPC systems have considerably improved their energy efficiency, so opportunities exist to apply HPC solutions and technologies to Big Data systems. For instance, GPU have considerably improved the FLOPS/Watt ratio of HPC systems, and we could follow the same path for data analytics software. Such convergence is already happening. For instance, Big data systems, MapReduce, which have first addressed the issue of extreme scalability, are now considerably concerned about the responsiveness of the computation. As a result, we see more and more in-memory processing, which by alleviating access storage both improve the computation time and decreases the energy usage. For such kind of computation, a foreseeable trend would be to support high speed network commonly found in HPC platform.