SAHA: A Self-Adaptive Hierarchical Allocation Strategy for Heterogeneous Grid Environments

The allocation of data and I/O operations into nodes in a data Grid environment is a critical issue. Especially, the Grid nodes may have either different performance or different capacity in heterogeneous circumstance. If the system undergoes changes (for example, due to the insertions or removals of disks from Grid nodes), it may be necessary to redistribute the data object with a low time and space complexity. Previous techniques mainly focus on handling these requirements only in part. For instance, some standard hashing and heuristic schedules can be used to decrease the data replacement time, but they usually do not adapt well to a change in the capabilities (both the capacity and performance of Grid nodes). In this paper, we presented a novel self-adaptive hierarchical allocation (called SAHA) based on nodes capacity to satisfy the load balancing. Furthermore, we illustrated the performance evaluation of our placement strategies while we allowed the various test conditions to be changed.