Grid computing is becoming the most demanding platform for solving large-scale scientific problems. Grid scheduling is the core component of a Grid infrastructure. Grid scheduling plays a key role in the efficient and effective execution of Grid jobs. In this context, understanding the characteristics of real Grid workloads is a critical step for improving the quality of an existing Grid scheduler, and in guiding the design of new scheduling solutions. Towards this goal, in this paper we present our developed web based simulator for the statistical analysis of Grid workload traces. Our web based simulator provides a comprehensive characterization of the real workload traces. Metrics that we characterize include system utilization, job arrival rate and inter-arrival time, job size (degree of parallelism), job runtime, data correlation and Fourier analysis. Our paper provides a realistic basis for experiments in resource management and evaluations of different job scheduling algorithms in Grid computing. © 2011 Springer-Verlag.