The Cost of Transparency:

Grid computing has been a hot topic for a decade. Several systems have been developed. Despite almost a decade of research and tens of millions of dollars spent, uptake of grid technology has been slow. Most deployed grids are based on a toolkit approach that requires significant software modification or development. An operating system technique used for over 30 years has been to reduce application complexity by providing transparency, e.g. file systems mask details of devices, virtual machines mask finite memory, etc. It has been argued that providing transparency in a grid environment is too costly in terms of performance. This paper examines that question in the context of data grids by measuring the performance of a commercially available data grid product – the Avaki Data Grid (ADG). We present the architecture of the ADG, describe our experimental setup, and provide performance results, comparing the ADG to a native NFS V3 implementation for both local and wide area access cases. The results were mixed, though encouraging. For single client local file operations, native NFS outperformed the ADG by 15% to 45% for smaller files, though for files larger than 32 MB ADG outperformed native NFS.