Resource reservation in grid computing environments: Design issues

Grid computing is an environment that enforces utilization of distributively owned and geographically dispersed resources for supporting various applications with variety of services. In this environment, one of the key considerations is resource management with respect to quality of service (QoS). QoS constraints can be specified using benefit functions. The execution of network-based applications often in need of end-to-end quality of service to prove the performance impact. The end-to-end QoS can be achieved and guaranteed through proper configuration, reservation and allocation of corresponding resources. For the discussion on reservation in grid computing environment we are considering the following parameters; resource utiliz ation rate and request rejection rate. Present reservation mechanisms support the improvement of resource utilization rate and reduction of request rejection rate on both high and low reservation scenarios. Also one of the serious problems is that current supercomputer scheduling systems do not provide mechanisms that allow several scheduling systems to provide simultaneous access to resources. This is called co-allocation problem. In order to solve this problem, advance reservation of resources is used. In contrast, advance reservation technique will result in low resource utilization rate and high rejection rate when the reservation rate is high. In order to mitigate this negative effect, a relaxed resource advance reservation policy (RARP) [1] has been introduced. But it is successful only at high reservation rate. In order to improve the utilization rate we have to concentrate more on the prior information from the resources and the trust factor. Resources will give static and dynamic information about it. If the environment is supporting the adaptation of trust oriented algorithms and techniques, leads the environment to perform better in both high and low reservation rate.