Job scheduling and dynamic data replication in data grid environment
The Journal of Supercomputing
Hi-index | 0.00 |
In data grid, replication data on multiple nodes can improve availability and response time. Yet determining when and where to replicate data in order to meet performance goals with many users and files, dynamic network characteristic, and changing user behavior is difficult. Also, it is a challenging problem to find the best-fit replica efficiently according to location and performance of physical storage systems at which the replicas dwell. To address these problems, we combine tree topology of grid architecture with multi-ring topology of peer-to-peer architecture to model replica network topology. Based on the proposed model, a novel dynamic adaptive replica allocation algorithm that can adapt to thenode insertions and deletions frequently is proposed. The algorithm is adaptive in the sense that it changes the replication scheme of the grid nodes as changes occur in the read-write pattern of the grid node. To address scalability and replica location efficiently, we use Chord algorithm with considering characteristics of locality in the process of replica location. We use gird simulator GridNet to evaluate our approach, simulation results demonstrate that our algorithms efficiently reduce the communication cost of object access in data grid environment and improve the replica access success rate.