End-System Performance-Adaptive Peak Link Utilization Transport Over Dedicated Lambda Grids

Next generation network applications will require high, dynamically-stable transport throughputs at the application level, in order to support such tasks as computational monitoring and steering on supercomputers, massive wide-area data transfers between storage systems, and remote instrument control. However, the current shared IP infrastructure is unsuited for providing the unimpeded stable bandwidth that is needed. Also, the widely used transport protocol TCP does not scale well enough to provide the needed throughput or stable dynamics, particularly at the application level. The problem is that traditional transport methods have not been optimized for such transfers. Moreover, circuit-switched connections push congestion to the endpoints congestion that most rate-based transport protocols are ill-suited to handle. This project will design, analyze, implement, and test an end-system, performance-adaptive, peak-link utilization transport system for achieving high throughputs and stable dynamics over complex, dedicated, lambda grid network connections. Commercial Applications and other Benefits as described by the awardee: The underlying principles of the technology should impact the design of transport software modules with controlled dynamics over complex channels. The integrated transport package would be made available to the broad research community to analyze and optimize data transfer in a wide range of applications including earth science, high energy and nuclear physics, astrophysics, climate modeling, nanoscale materials science, and genomics