A cost-optimized dynamic group peering mechanism in P2P grid

Abstract

With the increasing number of computers on the Internet, there is a growing interest in harnessing the unused and inexpensive computational resources over the Internet. However, current approaches such as the Grid computing paradigm are not sufficient. We present our research work that uses extends of Peer-to-Peer (P2P) computing with a framework that allows Grid computing over the Internet. Our study describes important considerations to the design of P2P Grid computing system based on redundant group peer concept. Moreover, we introduce the detailed system architecture which is designed in order to provide some enhancement points with respect to Grid service reliability through the proposed scheme and algorithm. Currently, there are many system requirements to guarantee the scalability, fault-tolerance, and high performance in the rapidly evolving research fields such as Grid and P2P technologies. Due to the dynamic nature of the Grid computing and P2P in networks, the behavior of system might be very unpredictable in respect of reliability. As the first step, we make a probabilistic model for system reliability in redundant group peer based P2P Grid. With the considerations to system cost and adjusted redundancy level of group peer, we can obtain optimum level of group peer and reduce the instability of system (Redundancy Optimization). The estimation of system load is the second step to indicate current status of a particular system for example system utilization, availability, and instability. With analytical model for load distribution, we determine the SLA-constrained load scheduling policy (SCLSP) in order to fairly schedule and distribute offered load among group peers and resources. Thus we can reduce the execution time and loss rate of loads need to be handled by computing elements. Finally through the theoretical analysis and simulation, we found that SCLSP has quite good performance comparing with existing algorithm such as RLB (Random Load B...