Performance Analysis of Agent Based Framework

Abstract

Cloud computing is the delivery of on-demand computing resources – everything from applications to datacenters over the internet on pay per use basis. Service composition in cloud is an important mechanism usually done through several scheduling policies or by dynamic resource allocation techniques. Services delivered to the users lack Quality of Service during this composition phase. Hence, an Agent-based cloud computing paradigm for resource allocation is proposed that consist of designing and developing software agents for cloud service discovery, finding appropriate service and service composition. Here, two types of agents are created – consumer agent that process the user requirements and service provider agent that clusters the output from consumer agent. Finally, each user is mapped to each producer. Experimental results shows efficient execution of user requests using agent with 100% success execution rate in a parallel manner than that of using Hadoop.