Trust oriented resource allocation using bidding
Sonu Mariam Paulose, Rajkumar Venkatesan, Krishnamoorthi Ramalakshmi · 2011
Conjunction of massive amount of idle computers or resources that may be loosely coupled, heterogeneous and geographically dispersed to reach a common goal leads to a virtual computing platform for sharing resources across the world. Resource management, application development and usage models in these environments has some dilemma in undertaking resources due to resource providers with multiple administrative domains having their own policies and terms. A ditch in the Grid computing environment is how to coordinate the distributed resources amongst a dynamic set of individuals and organizations where the requesters and providers are allowed to join and leave Grid environment at any time. Bidding model prevents single point of failure and server overload problems of match making model while minimizing turnaround time by using some set of deterministic and probabilistic selection heuristics where resource requesters and resource providers were given the privilege to take autonomous decisions regarding resource selection. Autonomous decision making is enabled via peer-to-peer decentralized scheduling frame work. In decentralized environment, lack of global information is a key challenge to facilitate optimum decision making which can lead to greedy selection of the best provider. Therefore some probabilistic selection is used to reduce the fairness deviation among processors while minimizing the turnaround time. Currently just various level of information about providers has been concentrated to minimize the turnaround time. Simply concentrating on various level of information may also leads to failure due to rejection factor resulted by number of failures occurred at provider. However by merging trust oriented mechanisms along with various level of information, rejection factor can also be minimized along with minimization of the turnaround time.