Resumen de DataMonster: building a cost-effective, high-speed, heterogeneous shared file system
In this article I describe how we built DataMonster, a high-speed heterogeneous shared file system. The project was built for the Scientific Computing Division (SCD) located at the National Center for Atmospheric Research (NCAR), where we support both supercomputers and very large datasets. The datasets are located on a multi-petabyte archival system made up of StorageTek�s silos housed within SCD. The most popular datasets needed to be more readily accessible to NCAR�s sixty-sixmember university users conducting atmospheric and earth science research. DataMonster would provide the means to accomplish this. Over the course of this project, DataMonster�s nature and composition changed, from a high-priced racing car to a cost-effective commuter car with race-car performance, by juggling choices for hardware and software.
