Security-Assured Resource Allocation for Trusted Grid Computing*

A new trust-based model is developed for optimizing resource allocation in a distributed Grid computing environment. Highly shared resources in a Grid create the insecurity and dependability concerns that hinder distributed applications. We model the Grid resource allocation process as a multi-objective integer-programming problem. Trusted Grid computing power is maximized over multiple resources. At the same time, the aggregate cost of Grid services is minimized to yield a high performance/cost ratio. Our new Security-Assured Resource Allocation (SARA) scheme enables dynamic resources under the security constraints. The Grid resources are assured with a distributed security infrastructure. First, we present the mathematical model for SARA optimization process. Then we use the Simplex method and branch-and-bound method to obtain optimal or suboptimal solutions. We illustrate the ideas with a working example of a pool of 11 host machines from 2 resource sites. For k resource sites with N machine hosts, our SARA algorithm has a search complexity O(2N/k), which is much lower than O(2) in exhaustive search for the optimal solutions. Our model can be applied to secure many Grid applications in scientific supercomputing or in cyberspace information services.