Analysis Of Fault Tolerance Approach For Data Replication In Data Intensive Scientific Applications

Abstract: Data Intensive scientific computing involves organizing, moving, visualizing, and analyzing massive amounts of data from around the world, as well as employing large-scale computation. There are systems that try to unify job and data management, but these are two different tasks to face in a Grid environment. Existing systems have tried to achieve only either job scheduling or data replication and does not provide fault tolerance. Data replication and job scheduling are two different but complementary functions in Data Grids: one to minimize the total file access cost (thus total job execution time of all sites), and the other to minimize the MakeSpan (the maximum job completion time among all sites). The two main challenges: first, how to formulate a problem that incorporates not only data replication but also job scheduling, and which addresses both total access cost and maximum access cost; and second, how to find an efficient algorithm that, if it cannot find optimal solutions of minimizing total/maximum access cost, gives near-optimal solution for both objectives in the proposed system. A novel method to achieve maximum fault tolerance in the Grid environment system by using a fault recovery pool is proposed.

Keywords: MakeSpan, Optimal MakeSpan, Nominal Distribution, Data Grid, GridSimI