Abstract: In one embodiment, a device deploys short timescale prediction model and a long timescale prediction model to one or more hosts in a network, whereby the short timescale prediction model predicts failure conditions for an online application that are attributable to the network on a timescale that is shorter than that of the long timescale prediction model. The device configures a trigger that causes the long timescale prediction model to activate predictions by the short timescale prediction model. The device evaluates performance of the short timescale prediction model. The device adjusts the trigger, when the performance of the short timescale prediction model is unacceptable.

Abstract: In one embodiment, a device forms clusters of network paths via which traffic for an online application is conveyed by applying clustering to time series of quality of experience metrics for the online application. The device calculates, for a particular cluster of network paths among the clusters of network paths, correlation metrics between path features of those network paths of that cluster. The device selects, based on the correlation metrics, a particular path feature from among the path features as a root cause of poor quality of experience metrics for the online application along the network paths of the particular cluster of network paths. The device provides, to a user interface, an indication of the particular path feature as being the root cause of poor quality of experience metrics for the online application.

Abstract: In one embodiment, a device receives, via a user interface, an indication of what is considered noise within a time series of a path performance metric. The device selects, based on the indication, a particular denoising filter to be applied to telemetry data obtained from one or more network paths regarding the path performance metric. The device forms model training data by applying the particular denoising filter to telemetry data obtained from one or more network paths regarding the path performance metric. The device trains, using the model training data, a prediction model to predict when a given network path will experience a failure condition.

Abstract: In one embodiment, a device computes states of a network path associated with an online application by representing time series of telemetry data regarding the network path as discrete values. The device generates state trajectories from the states of the network path computed by the device. The device selects one or more sub-sequences of the state trajectories based on prediction performance metrics that represent how well the one or more sub-sequences are able to predict a failure condition of the network path. The device causes a networking entity to use the one or more sub-sequences of the state trajectories to perform predictive routing for the network path.

Abstract: In one embodiment, a device obtains quality of experience metrics for an online application. The device obtains network metrics for one or more network paths over which traffic for the online application was routed. The device identifies a lack of correlation between the quality of experience metrics for the online application and the network metrics for the one or more network paths over which traffic for the online application was routed. The device disables, based on the lack of correlation, explicit probing of the one or more network paths.

Abstract: In one embodiment, a device obtains telemetry data for a path in a network that is used to convey traffic associated with an online application. The device identifies, based on the telemetry data, oscillations of the path between a first state in which the path provides acceptable user experience for the online application and a second state in which the path does not provide acceptable user experience for the online application. The device determines a stability metric that quantifies the oscillations of the path. The device provides an indication of the oscillations of the path, based in part on the stability metric.

Abstract: In one embodiment, a network element in a network receives one or more machine learning models configured to make an inference about the network. The network element requests, according to a predefined peering plan, telemetry attribute data from one or more peer network elements specified by the peering plan. The network element receives the telemetry attribute data from the one or more peer network elements. The network element makes, using the one or more machine learning models, an inference about the network based in part on the received telemetry attribute data.

Abstract: In one embodiment, a method comprises causing, by a network controller device, a first access point (AP) device to initiate a reverse sounding operation comprising wireles sly requesting a mobile constrained network device to transmit a null data packet (NDP) at a first transmission interval, wirelessly receiving the NDP at the first transmission interval, and generating a reception report describing reception of the NDP and including beamforming information; causing, by the network controller device, a second AP device to generate a corresponding reception report describing a corresponding wireless detection of the NDP at the first transmission interval; and causing, by the network controller device, the mobile constrained network device to connect to a selected one of the first AP device or the second AP device for an identified data flow based on the respective reception reports from the first and second AP devices.

Abstract: In one embodiment, a device clusters traffic characteristics of traffic associated with a plurality of online applications into one or more clusters. The device determines representative traffic characteristics for a particular cluster in the one or more clusters. The device generates, based on the representative traffic characteristics, a probing strategy for the plurality of online applications associated with the particular cluster. The device causes path probes to be sent along one or more network paths in accordance with the probing strategy.

Abstract: In one embodiment, a device forms a plurality of clusters of network paths used to convey traffic for an online application by applying clustering to telemetry data for those network paths. The device determines a predictability metric for a particular cluster in the plurality of clusters. The device provides an indication of the predictability metric for the particular cluster for display. The device enables, based in part on the predictability metric, predictive routing for the network paths in the particular cluster.

Abstract: In one embodiment, a device obtains telemetry data for a path in a network that is used to convey traffic associated with an online application. The device identifies, based on the telemetry data, oscillations of the path between a first state in which the path provides acceptable user experience for the online application and a second state in which the path does not provide acceptable user experience for the online application. The device determines a stability metric that quantifies the oscillations of the path. The device provides an indication of the oscillations of the path, based in part on the stability metric.

Abstract: In one embodiment, a supervisory device in a network notifies, via an access point of the network, a node as to an ability of the network to support virtual access points. The supervisory device receives, in response to notifying the node, information from the node regarding characteristics of the node. The supervisory device selects, based on the characteristics of the node, a plurality of access points in the network to form a virtual access point with which the node may communicate. The supervisory device configures the plurality of access points to function as the virtual access point, wherein the node communicates with the network via the virtual access point.

Abstract: In one embodiment, a device identifies a plurality of paths between a pair of network addresses, wherein one of the pair of network addresses is associated with an online application. The device obtains telemetry data from the plurality of paths for the online application. The device computes, based on the telemetry data, fate sharing metrics for the plurality of paths. The device controls routing of application traffic between the pair of network addresses, based on the fate sharing metrics for the plurality of paths.

Abstract: In one embodiment, a device receives path telemetry data for one or more network paths in a network over which traffic for an online application is conveyed. The device computes time series dynamics for the path telemetry data. The device determines a mapping of the time series dynamics to application experience metrics for the online application. The device routes traffic associated with the online application based on the mapping.

Abstract: In one embodiment, a device makes a prediction regarding service level agreement violations by a network transport available between a site and an online application. The device associates a risk metric with the prediction, based in part on a type of the network transport. The device computes a longevity metric for the prediction that indicates an expected validity period for the prediction. The device cause traffic to be routed between the site and the online application using the network transport, based on the prediction and its associated risk metric and its longevity metric.

Abstract: In one embodiment, a device computes an efficiency metric regarding ingestion of telemetry data from a particular portion of a network by a predictive routing engine used to make predictive routing decisions for that portion of the network. The device makes a comparison between the efficiency metric and one or more control rules. The device determines, based on the comparison, whether ingestion of the telemetry data from the particular portion of the network by the predictive routing engine should be disabled. The device causes the predictive routing engine to stop ingesting telemetry data from the particular portion of the network.

Abstract: In one embodiment, a device computes states of a network path associated with an online application by representing time series of telemetry data regarding the network path as discrete values. The device makes, using a machine learning model, a prediction that a quality of experience metric for the online application will be degraded, based on a particular transition pattern of the states being observed for the network path. The device determines one or more performance metrics for the machine learning model with respect to the network path. The device provides an indication of the particular transition pattern of the states for display, based in part on the one or more performance metrics for the machine learning model with respect to the network path.

Abstract: In one embodiment, a device obtains path probe data between one or more end-user sites and an online application. The device makes, based on the path probe data, a prediction as to whether a direct Internet access path or a backhaul path would offer better application performance for the online application. The device generates, based on the prediction, a split tunnel policy for a particular end-user site. The device causes a particular end-user site to connect to the online application in accordance with the split tunnel policy.

Abstract: In one embodiment, a device obtains data indicative of quality of experience for an online application. The device predicts, based on the data, path performances of network paths between an endpoint and the online application for different traffic loads. The device selects traffic loads for the network paths between the endpoint and the online application, based on the path performances predicted by the device. The device causes application traffic to be load balanced across the network paths between the endpoint and the online application, in accordance with those traffic loads selected by the device.

Abstract: In one embodiment, a device obtains metadata for routing decisions made by a predictive routing service for a plurality of network deployments. The device identifies a network topology for a network deployment that does not use the predictive routing service. The device estimates, based on the metadata for routing decisions made by the predictive routing service, performance metrics for the predictive routing service were it to be used to make routing decisions for the network topology. The device sends, to a user interface, report data indicative of the performance metrics estimated for the predictive routing service were it to be used to make routing decisions for the network topology.