A Multi-Agent Based Intelligent Query Processing System for Hadoop with Foundation for Intelligent Physical Agents-OS Using Cooperating Agent in Cloud Environment

摘要

The cloud provides a flexible environment for customers to host and process their information. Many applications dealing with massive data and complex queries are routed to the cloud. The response time depends on query optimization and resource scheduling. There are several techniques to optimize the query and resource scheduling but the information about the query is not stored i.e., the frequent queries will recursively do the same set of procedures. Hence, to reduce the time consumed by redundant query processing procedure we introduce agents in Hadoop framework (JobTracker). As JobTracker is a single point of failure i.e., if this goes down, all running jobs are halted. Therefore, we introduce agents in place of JobTracker to intellectually process the query and assign the task to TaskTrackers. Here we use FIPA-OS (Foundation for Intelligent Physical Agents) toolkit enabling rapid development interacting agent based system in the JADE (Java Agent Development Environment) platform.