The Interest, the Organizers and Keynote Speakers Have Attempted to Capture the Key Insights from the Workshop for Wider
چکیده
......Processing in large-scale systems is shifting from the traditional computingcentric model successfully used for many decades into one that is more data centric. This transition is driven by the evolving nature of computing, which is no longer dominated by the execution of arithmetic and logic calculations but instead by the handling of large data volume and the cost of moving data to the locations where computations are performed. The computing-centric model where data lives on disk, or even tape, and moves as needed to a central computing engine across a deep storage hierarchy is sufficient when computational aspects dominate data movement aspects. In contrast, in the data-centric model, data lives in different storage levels within the hierarchy, with processing engines surrounding the data and operating on such data without moving it across the system. The trend toward big data is leading to changes in the computing paradigm, and in particular to the notion of moving computation to data, in what we call the near-data processing (NDP) approach. Data movement impacts performance, power efficiency, and reliability, three fundamental attributes of a system. NDP seeks to minimize data movement by computing at the most appropriate location in the hierarchy, considering the location of the data and the information that needs to be extracted from that data. Thus, in NDP, computation can be performed right at the data’s home, either in caches, main memory, or persistent storage. This is in contrast to the movement of data toward a CPU independent of where it resides, as is done traditionally. Examples of NDP already exist in systems that execute computations close to the disk, filtering or preprocessing the data streaming from the disks so that a minimal number of items are transferred for processing at other parts of the system. Conceptually, the same principle can be applied at other levels of a system’s memory and storage hierarchy by placing computing resources close to where data is located, and restructuring applications to exploit the resulting distributed computing infrastructure. At the MICRO-46 conference, a workshop was held to bring together experts from academia and industry, who presented recent advances in the development of large systems employing NDP principles (www.cs.utah. Rajeev Balasubramonian