Adaptive routing with endpoint feedback
Systems, switches, network endpoints, and methods are provided. In one example, a system is described that includes a latency measurement circuit to measure traffic on a network from an endpoint sender to an endpoint receiver across multiple paths. The system also includes a packet marking circuit to provide a routing mark for a packet destined for the endpoint receiver according to a network traffic measurement provided by the latency measurement circuit, where the routing mark provides an indication that supports routing for the packet to reach the endpoint receiver via a chosen path or subset of paths among the multiple paths.
1 . A network endpoint, comprising:
a network interface that provides connectivity to a network;
a latency measurement circuit to measure traffic on the network from a sender to a receiver across multiple paths; and
a packet marking circuit to provide a routing mark for a packet destined for the receiver according to a network traffic measurement provided by the latency measurement circuit, wherein the routing mark provides an indication that supports routing for the packet to reach the receiver via a chosen path or subset of paths among the multiple paths, and wherein the routing mark provides an indication of a sub-group of switch ports to avoid when routing the packet to the receiver.
2 . The network endpoint of claim 1 , wherein the routing mark defines the chosen path or subset of paths for the packet to follow or another path in the multiple paths for the packet to avoid as the packet travels to the receiver, wherein the chosen path or subset of paths traverses one or more switches.
3 . The network endpoint of claim 2 , wherein the chosen path or subset of paths traverses a plurality of switches.
4 . The network endpoint of claim 3 , wherein the plurality of switches comprise at least one of a Top-of-Rack (ToR) switch and a spine switch.
5 . The network endpoint of claim 2 , wherein the chosen path or subset of paths is positively defined by the routing mark.
6 . The network endpoint of claim 2 , wherein the chosen path or subset of paths is defined by identifying the another in the multiple paths that are to be avoided.
7 . The network endpoint of claim 1 , wherein the latency measurement circuit generates and transmits a plurality of probe packets that travel to the receiver via the multiple paths, wherein the receiver returns the plurality of probe packets to the network endpoint, and wherein the latency measurement circuit analyzes a travel duration associated with each of the plurality of probe packets to measure the traffic on the network.
8 . The network endpoint of claim 7 , wherein a first probe packet among the plurality of probe packets traversed a first network path among the multiple paths and required a first duration to traverse the first network path, wherein a second probe packet among the plurality of probe packets traversed a second network path among the multiple paths and required a second duration to traverse the second network path, wherein the second duration is greater than the first duration, and wherein the second network path is identified as a path among the multiple paths to avoid in response to the latency measurement circuit determining that the second duration is greater than the first duration by a predetermined duration.
9 . The network endpoint of claim 7 , wherein a first probe packet among the plurality of probe packets traversed a first network path among the multiple paths and required a first duration to traverse the first network path, wherein the latency measurement circuit determines that the first duration exceeds an average duration required for the plurality of probe packets to reach the receiver by a predetermined duration, and wherein the first network path is identified as a path among the multiple paths to avoid in response to the latency measurement circuit determining that the first duration exceeds the average duration required for the plurality of probe packets to reach the receiver by a predetermined duration.
10 . The network endpoint of claim 9 , wherein the predetermined duration comprises at least one standard deviation in excess of the average duration.
11 . The network endpoint of claim 9 , wherein the latency measurement circuit determines that a second probe packet among the plurality of probe packets traversed the first network path in a duration less than the first duration and, in response thereto, notified the packet marking circuit that the first network path is no longer required to be avoided.
12 . The network endpoint of claim 1 , wherein the packet is part of a Remote Direct Memory Access (RDMA) packet flow.
13 . A system, comprising:
a latency measurement circuit to measure traffic on a network from an endpoint sender to an endpoint receiver across multiple paths; and
a packet marking circuit to provide a routing mark for a packet destined for the endpoint receiver according to a network traffic measurement provided by the latency measurement circuit, wherein the routing mark provides an indication that supports routing for the packet to reach the endpoint receiver via a chosen path or subset of paths among the multiple paths, and wherein the routing mark provides an indication of a sub-group of switch ports to avoid when routing the packet to the endpoint receiver.
14 . The system of claim 13 , further comprising a routing circuit to route the packet within the network according to the routing mark and wherein the latency measurement circuit generates and transmits a plurality of probe packets that travel to the endpoint receiver via the multiple paths, wherein the endpoint receiver returns the plurality of probe packets to the endpoint sender, and wherein the latency measurement circuit analyzes a travel duration associated with each of the plurality of probe packets to measure the traffic on the network.
15 . The system of claim 14 , wherein a first probe packet among the plurality of probe packets traversed a first network path among the multiple paths and required a first duration to traverse the first network path, wherein a second probe packet among the plurality of probe packets traversed a second network path among the multiple paths and required a second duration to traverse the second network path, wherein the second duration is greater than the first duration, and wherein the second network path is identified as a path among the multiple paths to avoid in response to the latency measurement circuit determining that the second duration is greater than the first duration by a predetermined duration.
16 . The system of claim 14 , wherein a first probe packet among the plurality of probe packets traversed a first network path among the multiple paths and required a first duration to traverse the first network path, wherein the latency measurement circuit determines that the first duration exceeds an average duration required for the plurality of probe packets to reach the endpoint receiver by a predetermined duration, and wherein the first network path is identified as a path among the multiple paths to avoid in response to the latency measurement circuit determining that the first duration exceeds the average duration required for the plurality of probe packets to reach the endpoint receiver by a predetermined duration.
17 . The system of claim 16 , wherein the predetermined duration comprises at least one standard deviation in excess of the average duration.
18 . The system of claim 16 , wherein the latency measurement circuit determines that a second probe packet among the plurality of probe packets traversed the first network path in a duration less than the first duration and, in response thereto, notified the packet marking circuit that the first network path is no longer required to be avoided.
19 . The system of claim 13 , wherein the chosen path or subset of paths traverses a plurality of switches.
20 . The system of claim 13 , wherein the packet is part of a Remote Direct Memory Access (RDMA) packet flow.