Knowledge based Replica Management in Data Grid Computation

Abstract

As the number of scientific disciplines has increased, the large data collections are emerging as important community resources. In domains as high energy physics, and computational genomic, the volume of interesting data is already measured in terabytes and will soon total peta bytes. The Research communities need to access and analyze this data using complex computational and access techniques. No data management infrastructure provides solution against the complex computational analysis of huge and geographically distributed data sets. Most of queries usually search analyzed data from terabytes of distributed data repository, over wide area networks. Replicas, and other advanced techniques collectively maximize the use of scarce storage, networking, and computing resources. Existing data grid replication technique no doubt provide us with availability of required data sets, but in order to create replica it has to bear an overhead of huge computation for required data sets. Large network traffic cause the performance unsatisfactory. Our goal in this effort is to provide users with replication infrastructure in Grid that uses Knowledgebase having learning capability so as to reduce the computation for creating dataset on each user request.