On How to Provision Quality of Service (QoS) for Large Dataset Transfers
For large-sized scientific dataset transfers, scientists typically invest in high-end computing systems that can source and sink data to/from their disk systems at high speeds. These transfers are referred to as flows as they dominate other flows. They also cause increased burstiness, which in turn impacts delay-sensitive real-time audio/video flows. In prior work, the authors proposed an overall architecture for an intra-domain traffic engineering system called Hybrid Network Traffic Engineering System (HNTES) that performs two tasks: analyzes NetFlow reports offline to identify flows, and configures the ingress routers for future -flow redirection to traffic-engineered QoS-controlled paths.