Transparent on-demand co-allocation data access for grids

With the conventional data sharing system, a dataintensive application usually accesses data by using either the pre-staging scheme or the on-demand access scheme. Pre-staging systems simultaneously download an entire shared file from multiple data sources even when only a tiny file fragment is required. Obviously, pre-staging systems consume unnecessary data transmission time and storage space. On-demand access systems, on the other hand, download only the necessary fragments from a single data source. Ondemand systems unfortunately do not fully exploit available network bandwidth. This paper presents a data sharing system which uses a hybrid scheme, designated as the On-Demand data Co-Allocation (ODCA). ODCA downloads the necessary file fragments on-demand from multiple data sources, thereby reducing data transmission time, wasted network bandwidth and required storage space. Besides, ODCA enable unmodified legacy applications to cost-effectively migrate to the grid environment by using the native I/O system calls. Experimental results show the ODCA scheme successfully reduces turnaround time in data-intensive applications.