Data-Intensive Scalable Computing for Scientific Applications

摘要

Increasingly, scientific computing applications must accumulate and manage massive datasets, as well as perform sophisticated computations over these data. Such applications call for data-intensive scalable computer (DISC) systems, which differ in fundamental ways from existing high-performance computing systems. The ever-increasing volumes of scientific data being generated, collected, and stored call for a new class of high-performance computing facility that places emphasis on data, rather than raw computation, as the core focus of the system. The system is responsible for the acquisition, updating, sharing, and archiving of the data, and it supports sophisticated forms of computation over the data. I refer to such systems as data-intensive scalable computer (DISC) systems. I believe that DISC systems will yield breakthroughs in many scientific disciplines.