In grid computing, most of the SLA management systems focus on static SLAs, where the guarantee terms are defined as constants or bounds. None of them support dynamic SLAs based on analytical expressions. Therefore, to enforce dynamic SLAs and to improve services offered to applications, we propose a new monitoring method to control the resource utilization and to reconfigure them when necessary in order to match analytical expressions used to describe the service parameters. We are interested in the performance characteristics of grid computing at the network side, especially delay transfer, bandwidth and queue's lengths. The proposed method called flatness based trajectory tracking, considers time delay transfer by using delay predictor and provides value-added services for grid applications over Internet. Moreover, discrete-time simulation under GridSim simulator illustrates its feasibility and efficiency to provide guarantees for critical grid traffics with timely execution requirements and to match expressive data transfer SLAs.