Distributed dynamic load balancing in network systems
A source switching device in a switching system receives information measured by a target switching device in the switching system. The information is indicative of an amount of data received in a given amount of time by the target switching device via each of two or more first links coupled to the target switching device. The source switching device determines, based at least in part on the information received from the target device, a path, from among multiple paths from the source switching device to the target switching device, for transmission of a packet flow directed to the target switching device. The source switching device transmits, via the determined path for transmission of the packet flow to the target device, one or more packets belonging to the packet flow.
1. A method for balancing traffic load in a network system that includes at least a first leaf switch device communicatively coupled to one or more second leaf switch devices via one or more spine switch devices, the method comprising:
for each of a plurality of paths through the network system, determining, at the first leaf switch device, a respective traffic share metric for a particular packet flow being transmitted by the first leaf switch device during a first time interval, the respective traffic share metric indicating a comparison of an amount of data in the particular packet flow during the first time interval versus an amount of data on the respective path during the first time interval;
determining, by the first leaf switch device, a path from among multiple paths for transmitting the particular packet flow during a second time interval using the traffic share metrics for the plurality of paths; and
transmitting, by the first leaf switch device, one or more packets belonging to the particular packet flow via the determined path during the second time interval.
2. The method for balancing traffic load of claim 1 , wherein the multiple paths include a first path and one or more second paths, and wherein the method further comprises:
for each of the one or more second paths, determining, by the first leaf switch device, a respective comparison metric that indicates a respective comparison of a first traffic share metric corresponding to the first path with a respective second traffic share metric corresponding to the second path;
wherein determining the path from among the multiple paths comprises determining the path using at least the one or more comparison metrics.
3. The method for balancing traffic load of claim 2 , wherein:
determining, for each of the one or more second paths, the respective comparison metric comprises determining, for each of the one or more second paths and using the traffic share metrics corresponding to the first path and the one or more second paths, a respective gain metric that indicates a respective potential gain in a share of traffic corresponding to the particular packet flow if the particular packet flow is redirected (i) from the first path to (ii) a respective second path; and
wherein determining the path from among the multiple paths comprises determining the path using at least the one or more gain metrics.
4. The method for balancing traffic load of claim 1 , wherein the plurality of paths includes a first path and one or more second paths, wherein the particular packet flow is actually transmitted by the first leaf switch device via the first path during the first time interval, and wherein determining, for each of the plurality of paths, the respective traffic share metric comprises:
determining a first traffic share metric corresponding to the first path using received traffic quantity information that is indicative of a first amount of data received at a first port of the one or more second leaf switch devices during the first time interval, the first port corresponding to the first path, the first amount of data corresponding to a first plurality of packet flows received at the first port; and
determining one or more respective second traffic share metrics corresponding to the one or more second paths using received respective traffic quantity information that are indicative of one or more respective second amounts of data received at one or more second ports of the one or more second leaf switches during the first time interval, the one or more second ports respectively corresponding to the one or more second paths, the one or more second amounts of data corresponding to one or more respective second pluralities of packet flows respectively received at the one or more second ports.
5. The method for balancing traffic load of claim 4 , further comprising:
receiving, at the first leaf switch device, one or more control messages from the one or more second leaf switch devices, the one or more control messages including i) the traffic quantity information that is indicative of the first amount of data received at the first port, and ii) the respective traffic quantity information that are indicative of the one or more respective second amounts of data received at the one or more second ports.
6. The method for balancing traffic load of claim 5 , wherein receiving the one or more control messages comprises receiving the one or more control messages via the one or more spine switch devices.
7. The method for balancing traffic load of claim 4 , further comprising:
determining, at the first leaf switch device, one or more respective comparison metrics that indicates respective comparisons of the first traffic share metric with the one or more second traffic share metrics; and
wherein determining the path from among the multiple paths comprises determining the path using at least the one or more comparison metrics.
8. The method for balancing traffic load of claim 1 , further comprising:
measuring, at the first leaf switch device, transmitted traffic quantity information that corresponds to the first leaf switch device transmitting the particular traffic flow during the first time interval;
wherein determining, for each of the plurality of paths, the respective traffic share metric for the particular packet flow comprises determining, for each of the plurality of paths, the respective traffic share metric for the particular packet flow using the transmitted traffic quantity information corresponding to the first leaf switch device transmitting the particular traffic flow.
9. The method for balancing traffic load of claim 1 , wherein determining, for each of the plurality of paths, the respective traffic share metric comprises:
determining, for each of the plurality of paths, the respective traffic share metric using a data rate of the particular packet flow.
10. The method for balancing traffic load of claim 1 , wherein transmitting the one or more packets via the determined path comprises transmitting the one or more packets to one of the second leaf target devices via one of the spine switch devices included in the determined path.
11. A first leaf switch device for operation in a network system, the network system further including one or more second leaf switch devices and one or more spine switch devices that communicatively couple the first leaf switch device with the one or more second leaf switch devices, the first leaf switch device comprising:
a plurality of ports to couple the first leaf switch device to respective links in the network system;
a packet processor device coupled to the plurality of ports, the packet processor configured to forward packets to the plurality of ports for transmission via the respective links, wherein the packet processor device comprises a load balancer device configured to:
for each of a plurality of paths through the network system, determine a respective traffic share metric for a particular packet flow being transmitted by the first leaf switch device during a first time interval, the respective traffic share metric indicating a comparison of an amount of data in the particular packet flow during the first time interval versus an amount of data on the respective path during the first time interval,
determine a path from among multiple paths for transmitting the particular packet flow during a second time interval using the traffic share metrics for the plurality of paths, and
forward one or more packets belonging to the particular packet flow via the determined path during the second time interval.
12. The first leaf switch device of claim 11 , wherein the multiple paths include a first path and one or more second paths, and wherein the load balancer device is further configured to:
for each of the one or more second paths, determine a respective comparison metric that indicates a respective comparison of a first traffic share metric corresponding to the first path with a respective second traffic share metric corresponding to the second path; and
determine the path using at least the one or more comparison metrics.
13. The first leaf switch device of claim 12 , wherein the load balancer device is further configured to:
determine, for each of the one or more second paths and using the traffic share metrics corresponding to the first path and the one or more second paths, a respective gain metric that indicates a respective potential gain in a share of traffic corresponding to the particular packet flow if the particular packet flow is redirected (i) from the first path to (ii) a respective second path; and
determine the path using at least the one or more gain metrics.
14. The first leaf switch device of claim 11 , wherein the plurality of paths includes a first path and one or more second paths, wherein the particular packet flow is actually transmitted by the first leaf switch device via the first path during the first time interval, and wherein the load balancer device is further configured to:
determine a first traffic share metric corresponding to the first path using received traffic quantity information that is indicative of a first amount of data received at a first port of the one or more second leaf switch devices during the first time interval, the first port corresponding to the first path, the first amount of data corresponding to a first plurality of packet flows received at the first port; and
determine one or more respective second traffic share metrics corresponding to the one or more second paths using received respective traffic quantity information that are indicative of one or more respective second amounts of data received at one or more second ports of the one or more second leaf switches during the first time interval, the one or more second ports respectively corresponding to the one or more second paths, the one or more second amounts of data corresponding to one or more respective second pluralities of packet flows respectively received at the one or more second ports.
15. The first leaf switch device of claim 14 , wherein the load balancer device is further configured to:
receive one or more control messages from the one or more second leaf switch devices, the one or more control messages including i) the traffic quantity information that is indicative of the first amount of data received at the first port, and ii) the respective traffic quantity information that are indicative of the one or more respective second amounts of data received at the one or more second ports.
16. The first leaf switch device of claim 15 , wherein the load balancer device is configured to receive the one or more control messages via the one or more spine switch devices.
17. The first leaf switch device of claim 14 , wherein the load balancer device is further configured to:
determine one or more respective comparison metrics that indicates respective comparisons of the first traffic share metric with the one or more second traffic share metrics; and
determine the path using at least the one or more comparison metrics.
18. The first leaf switch device of claim 11 , wherein:
the packet processor is further configured to measure transmitted traffic quantity information that corresponds to the first leaf switch device transmitting the particular traffic flow during the first time interval; and
the load balancer device is further configured to determine, for each of the plurality of paths, the respective traffic share metric for the particular packet flow using the transmitted traffic quantity information corresponding to the first leaf switch device transmitting the particular traffic flow.
19. The first leaf switch device of claim 11 , wherein the load balancer device is further configured to:
determine the respective traffic share metric using a data rate of the particular packet flow.
20. The first leaf switch device of claim 11 , wherein the packet processor is configured to forward the one or more packets to one of the second leaf target devices via one of the spine switch devices included in the determined path.