Lawrence Berkeley National Laboratory

Scientific workflows often involve large data transfers, which increasingly require completion-time guarantees. To support these time-sensitive flows, the Energy Science Network (ESnet) has implemented on-demand circuits with packet priority, allowing the circuit to be utilized by other traffic when the deadline-sensitive flow is inactive. In this paper, we explore a deterministic networking framework designed to support large scientific data transfers with completion guarantees. We consider an ideal network where all nodes are time-synchronized and utilize Cyclic Queueing and Forwarding (CQF) to achieve reliable low-latency data transfers. Specifically, the CQF cycle time is configured to ensure that all data transfers between neighboring nodes are completed within the cycle time. The number of packets transferable between two neighboring nodes depends on the cycle time, propagation delay, and link bandwidth. We conduct simulations to compare the performance of the deterministic networking framework with two circuit-based schemes: one utilizing fixed bandwidth allocation for all requests and another employing dynamic bandwidth reservation, which adjusts the allocated bandwidth based on the available bandwidth along the path. Our results show that the deterministic network architecture achieves performance comparable to the dynamic bandwidth reservation scheme. We believe that a more optimized version of the time-sensitive networking protocol, exploiting multi-path routing, could offer better completion guarantees than traditional network reservation options, while enhancing the overall network bandwidth utilization.