A Hybrid Model for Load aware Trust management in grid.(Report)

Abstrct: Problem statement: Trust management in a distributed dynamic environment like grid has been very vital since the allocation of appropriate resources to meet user request influences the success of the system. While considering a set of resources which are loaded invariably, balancing the load also contributes for the efficient resource utilization. Approach: This study proposes a trust calculation model, which considers weighted parameters like direct trust, reputation trust, load average information and network efficiency. Network efficiency also varies with the granularity of jobs, hence the experiment were conducted with an inclusion of this criteria. Results and Conclusion: The proposed Hybrid Model for Load aware Trust management system (HMLT) allocates the job based on the trust values. The resources with higher trust value get more jobs. As a result the performance of the proposed system found to be faster and has a better utilization of grid resources when studied in a grid environment.

Key words: Trust management, network efficiency, load index, grid environment, separate services, grid resources, trust calculation, behavioral trust

INTRODUCTION

Grid is a distributed computing technology that supports aggregation of distributed computational resources that span beyond organizational boundaries. The coordinated use of resources meets the requirements of advanced science and engineering. Grid can be distinguished from conventional distributed computing by its focus on large scale resource sharing, high performance and solving compute/data intensive applications. Grid supports researchers and scientists from diverse organizations to share information, instruments, data, compute and storage resources dynamically in a flexible and secure manner(Foster et al., 2001) thereby forming a 'Virtual Organization' (VO) to solve challenging applications. The resources in grid are dynamic and are organized as a number of virtual organizations. The real complexity in scheduling a user job lies in identifying the suitable resource with the expected capability of the user job execution need. The process of match making in the domain of VOs becomes much critical due to the dynamic availability of resources in a grid. This led to the introduction of the familiar methods of adding the notion of trust with grid resources. Trust in the context of grid is classified broadly into security trust and behavioral trust; the former refers to the means of available protections for securing resources and the latter is concerned with the expected performance of the resource. Trust in traditional terms cannot be measured with any other domain where it is applied. In grid and web, the trust models work with trust calculation and trust is equated to an integer. In the proposed model, the behavioral trust of grid resources was considered. Behavioral trust is usually calculated by combining direct trust which is evaluated with the direct contacts of resources with the reputation trust that are received from other well known contacts' recommendation. Further, the trust calculations are normally made taking into consideration the decay of the calculated trust over a time since the environment being dynamic and the trust value changing often.

Trust calculation techniques developed so far have not taken into account the load balancing, job size and network efficiency which are the significant parameters that influence performance. Hence, the proposed model discusses a novel trust calculation paradigm in grid considering parameters such as job granularity, network efficiency and past behavior of resources. These parameters are measured through separate services, which together can be exposed as a single web service …