Predictive BGP peering
In one embodiment, a device determines a mapping between a network destination and Border Gateway Protocol (BGP) peers located across a plurality of autonomous systems for which the network destination is reachable. The device causes, based on the mapping, performance of probing tests along a plurality of paths to the network destination and via the BGP peers, to obtain path performance measurements for the plurality of paths. The device uses a prediction model to generate predicted performance metrics for the plurality of paths based on the path performance measurements. The device configures, based on the predicted performance metrics for the plurality of paths, the BGP peers with BGP peering policies to convey application traffic associated with the network destination via particular path from among the plurality of paths.
1 . A method comprising:
determining, by a device, a mapping between a network destination and Border Gateway Protocol peers located across a plurality of autonomous systems for which the network destination is reachable;
causing, by the device and based on the mapping, performance of probing tests along a plurality of paths to the network destination and via the Border Gateway Protocol peers, to obtain path performance measurements for the plurality of paths;
using, by the device, a prediction model to generate predicted performance metrics and corresponding uncertainty estimates for the plurality of paths based on the path performance measurements; and
configuring, by the device and based on the predicted performance metrics and the corresponding uncertainty estimates for the plurality of paths, the Border Gateway Protocol peers with Border Gateway Protocol peering policies to convey application traffic associated with the network destination via a particular path from among the plurality of paths, wherein the particular path is selected based on a comparison of a predicted performance metric improvement associated with conveying the application traffic via the particular path, considering a corresponding uncertainty estimate for the predicted performance metric improvement, relative to another path.
2 . The method as in claim 1 , wherein the network destination is associated with an online application.
3 . The method as in claim 2 , wherein the predicted performance metrics indicate that a current path to the network destination will violate a service level agreement (SLA) for the online application.
4 . The method as in claim 2 , further comprising:
obtaining quality of experience data from the online application, wherein the prediction model generates the predicted performance metrics based in part on the quality of experience data.
5 . The method as in claim 1 , wherein the Border Gateway Protocol peering policies adjust a Border Gateway Protocol local preference attribute or route weight attribute, to cause the Border Gateway Protocol peers to favor the particular path.
6 . The method as in claim 1 , wherein the Border Gateway Protocol peering policies adjust a multi-exit discriminator or path length attribute, to cause the Border Gateway Protocol peers to favor the particular path.
7 . The method as in claim 1 , wherein the network destination comprises an Internet Protocol (IP) prefix.
8 . The method as in claim 1 , further comprising:
reverting the Border Gateway Protocol peering policies, based on a determination that the predicted performance metrics for the particular path were incorrect.
9 . The method as in claim 1 , further comprising:
providing information regarding the mapping, the predicted performance metrics, and the particular path to a user interface.
10 . The method as in claim 1 , wherein the device determines the mapping based on Border Gateway Protocol routing information advertised by the Border Gateway Protocol peers.
11 . An apparatus, comprising:
one or more network interfaces;
a processor coupled to the one or more network interfaces and configured to execute one or more processes; and
a memory configured to store a process that is executable by the processor, the process when executed configured to:
determine a mapping between a network destination and Border Gateway Protocol peers located across a plurality of autonomous systems for which the network destination is reachable;
cause, based on the mapping, a performance of probing tests along a plurality of paths to the network destination and via the Border Gateway Protocol peers, to obtain path performance measurements for the plurality of paths;
use a prediction model to generate predicted performance metrics and corresponding uncertainty estimates for the plurality of paths based on the path performance measurements; and
configure, based on the predicted performance metrics and the corresponding uncertainty estimates for the plurality of paths, the Border Gateway Protocol peers with Border Gateway Protocol peering policies to convey application traffic associated with the network destination via a particular path from among the plurality of paths, wherein the particular path is selected based on a comparison of a predicted performance metric improvement associated with conveying the application traffic via the particular path, considering a corresponding uncertainty estimate for the predicted performance metric improvement, relative to another path.
12 . The apparatus as in claim 11 , wherein the network destination is associated with an online application.
13 . The apparatus as in claim 12 , wherein the predicted performance metrics indicate that a current path to the network destination will violate a service level agreement (SLA) for the online application.
14 . The apparatus as in claim 12 , wherein the process when executed is further configured to:
obtain quality of experience data from the online application, wherein the prediction model generates the predicted performance metrics based in part on the quality of experience data.
15 . The apparatus as in claim 11 , wherein the Border Gateway Protocol peering policies adjust a Border Gateway Protocol local preference attribute or route weight attribute, to cause the Border Gateway Protocol peers to favor the particular path.
16 . The apparatus as in claim 11 , wherein the Border Gateway Protocol peering policies adjust a multi-exit discriminator or path length attribute, to cause the Border Gateway Protocol peers to favor the particular path.
17 . The apparatus as in claim 11 , wherein the network destination comprises an Internet Protocol (IP) prefix.
18 . The apparatus as in claim 11 , wherein the process when executed is further configured to:
revert the Border Gateway Protocol peering policies, based on a determination that the predicted performance metrics for the particular path were incorrect.
19 . The apparatus as in claim 11 , wherein the process when executed is further configured to:
provide information regarding the mapping, the predicted performance metrics, and the particular path to a user interface.
20 . A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising:
determining, by the device, a mapping between a network destination and Border Gateway Protocol peers located across a plurality of autonomous systems for which the network destination is reachable;
causing, by the device and based on the mapping, performance of probing tests along a plurality of paths to the network destination and via the Border Gateway Protocol peers, to obtain path performance measurements for the plurality of paths;
using, by the device, a prediction model to generate predicted performance metrics and corresponding uncertainty estimates for the plurality of paths based on the path performance measurements; and
configuring, by the device and based on the predicted performance metrics and the corresponding uncertainty estimates for the plurality of paths, the Border Gateway Protocol peers with Border Gateway Protocol peering policies to convey application traffic associated with the network destination via a particular path from among the plurality of paths, wherein the particular path is selected based on a comparison of a predicted performance metric improvement associated with conveying the application traffic via the particular path, considering a corresponding uncertainty estimate for the predicted performance metric improvement, relative to another path.