Cisco Systems, Inc.
Forecasting SDN fabric saturation and machine learning-based flow admission control

Abstract:

In one embodiment, a device of a software defined wide area network (SD-WAN) predicts characteristics of a new traffic flow to be admitted to the SD-WAN, based on a set of initial packets of the flow. The device predicts an impact of admitting the flow to the SD-WAN, based in part on extrinsic or exogenous data regarding the SD-WAN. The device admits the flow to the SD-WAN, based on the predicted impact. The supervisory device uses reinforcement learning to adjust one or more call admission control (CAC) parameters of the SD-WAN, based on captured telemetry data regarding the admitted flow.