International Journal on Future Revolution in Computer Science & Communication Engineering ISSN: 2454-4248 Volume: 4 Issue: 3 37 –40 _______________________________________________________________________________________________ 37 IJFRCSCE | March 2018, Available @ http://www.ijfrcsce.org _______________________________________________________________________________________ Secured and Adaptive Load Balancing with Backup Approach for Computational Grids Dr. B.Jayanthi, Mr.S.Vijayakumar Department of Computer Science(P.G.) Kongu Arts and Science College(Autonomous) Erode, Tamilnadu, India sjaihere@gmail.com,sakthiveluvijayakumar@gmail.com Mr. M. Chandru Department of Computer Science Kongu Arts and Science College(Autonomous) Erode, Tamilnadu, India emchandru@gmail.com Abstract—Load Balancing is one of the big issues in Grid Computing.This work aims to develop a secured load balancing algorithm which reduces the download time, network overhead and improve the packet delivery ratio of the resources. This work enhances the PWSLB algorithm for load balancing, fault tolerant scheduling and security. The experimental results show an average of 0.2 to 8 % increase in Packet delivery Ratio and 0.080 to 0.1 % of network overhead reduction at 0.1324 milliseconds reduction in Download time. Finally this work Reduces, the download time, network overhead of tasks and also increases the packet delivery ratio Keywords- Grid computing, Load balancing, Fault tolerant scheduling, security __________________________________________________*****_________________________________________________ I. INTRODUCTION Grid computing is a collection of computer resources from multiple locations to reach a common goal. The grid can be thought of as a distributed system with non-interactive workloads that involve a large number of files. Grid computing has an emerged as the next generation of parallel and distributed computing methodology that aggregates dispersed heterogeneous resources for solving various kinds of large scale applications in science, engineering and commerce [9]. Load Balancing is one of the big issues in Grid Computing [1]. Load balancing Algorithm types and three policies are Information policy, Triggering Policy, and Selection Policy in Grid Environment are discussed in [10], [6]. In general load balancing algorithms can be classified as centralized or decentralized, and static or dynamic. In the centralized approach one node in the system acts as a scheduler and makes all the load balancing decisions. Information is sent from the other nodes to this node. In the decentralized approach [8], all nodes in the system are involved in the load balancing decisions. Many fault-tolerant schemes have been proposed for grid systems [2], [3], and [7]. Backup overloading to reduce replication cost of independent jobs introduced in [5]. SHA-3 preserves the online nature of SHA-2. That is, the algorithm process comparatively small blocks (512 or 1024) at a time instead of requiring the entire message to be buffered in memory before processing it [4]. II. MATERIALS AND METHODS A. Piggybacking In this study, piggybacking technique is introduced for load balancing. Each resource maintains the load information of other resources by using the state object. The state object helps a resource to estimate the load and efficiency of other resources at any time without message transfer. Each item in state object of neighbor or partner resource has a property list such as load, efficiency, time. Load denotes the load information of neighbor or partner resource, efficiency denotes the efficiency value of neighbor or partner resource, time denotes the neighbor’s or partner’s local time. When the load information or efficiency value is reported, each resource collects and maintains the load information of only its neighbor’s and partner’s. In order to minimize the overhead of information collection, load information exchange is done by piggybacking. Specifically, when resource transfers a packet to neighbor or partner resource for processing, resource appends the load information and efficiency values of itself, its neighbors, its partners to the packer and sent to neighbor or partner resource by piggybacking. Neighbor or partner resource updates the corresponding load information and efficiency values of its state object by comparing the timestamps if the resource contained in the packet belongs to its neighbors or partners. Similarly, neighbor or partner resource inserts the current load information and efficiency values of itself, its neighbors and its partners in the acknowledgement to resource. So resource can update its state objects. An advantage of piggybacking strategy reduces the message overhead and can takes small amount of network bandwidth. In this way, the load information packet should be simple and small sized as possible. B. Boundary Schedules Fault tolerant scheduling is an imperative step for large scale computational grid systems, as often geographically distributed nodes co-operate to execute a job [3]. Primary- backup approach is a commonly used for fault tolerance wherein each packet has a primary copy and backup copy on