492 ISSN 1858-1633 @2008 ICTS DIAGONAL DATA REPLICATION PROTOCOL FOR LARGE DYNAMIC NETWORK Rohaya Latip *, Hamidah Ibrahim †, Mohamed Othman*, Md Nasir Sulaiman†, Azizol Abdullah* *Department of Technology Communication and Network, †Department of Computer Science, Faculty of Computer Science and Information Technology, University Putra Malaysia, 43400 Serdang Selangor, Malaysia email : {rohaya, hamidah, mothman, nasir, azizol}@fsktm.upm.edu.my ABSTRACT Data replication can be used to improve data availability for a large network environment. In such a system, a mechanism is required to maintain the consistency of the replicated data. Diagonal data replication on grid (DRG) based on quorum approach is one of the solutions to improve data availability and maintain the data consistency even though the network size is increasing. However it is shown in the previous study, DRG still requires a bigger number of copies to be made available to construct a quorum. Thus it is not suitable for a large dynamic network. In this paper, we propose a protocol to enhance the DRG called Enhanced Diagonal Replication Grid (EDRG) where only one site from the diagonal sites organized in a logical grid structure is chosen to become the replica. Our results prove that our protocol, EDRG improves the performance of the data availability compared to DRG. Keywords: Data Management, Replication, Quorum, Replica Control Protocol, Availability 1 INTRODUCTION Networking Technologies are merging fast and now research such as [1,2] are focusing on grid where grid is a distributed network computing system, a virtual computer formed by a networked set of heterogeneous machines that agree to share their local resources with each other. A grid is a very large scale, generalized distributed network computing system that can scale to internet size environment with machines distributed across multiple organizations and administrative domains [3,4]. Ensuring efficient access to such a large network and widely distributed data is a challenge to those who design, maintain and manage the grid network. The availability of a data on a large network is an issue [5, 6, 7, 8] because geographically it is distributed and has different database management to share across the grid network whereas replicating data can become expensive if the number of operations such as read or write are high. Thus, to increase the data availability, a good strategy of data replication protocol must be introduced. In our work, we investigate the use of replication on large dynamic network to improve its ability to access data efficiently. Distributed computing manages thousands of computer systems and this has limited its memory and processing power. On the other hand, grid computing has some extra characteristics. It is concerned to efficient utilization of a pool of heterogeneous systems with optimal workload management utilizing an enterprise's entire computational resources (servers, networks, storage, and information) acting together to create one or more large pools of computing resources. There is no limitation of users or originations in grid computing. Even though grid sometime can be as minimum one node but for our algorithm the best number of nodes should be more the five nodes to implement the algorithm and suite the large dynamic network such as grid environment. The proposed protocol uses quorum approach because quorums improved the performance of fault tolerant and availability of replication protocols [9,10]. Quorums reduce the number of copies involved in reading or writing data. To address the availability, we replicate data in a logical structured of grid topology network and select the middle replica from a diagonal sites of the grid topology, which has been organized in quorums. This is because it is easy to access the database because of its middle location in the quorum. A simulation was developed using Java to evaluate the protocol. This paper is organized as follow, Section 2, discussed the previous protocol, DRG. In Section 3, we introduced our protocol EDRG. Section 4 discusses the simulation results and comparison. Brief conclusions and future works are discussed in Section 5.