System and method for facilitating data-driven intelligent network with per-flow credit-based flow control
Data-driven intelligent networking systems and methods are provided. The system can accommodate dynamic traffic with fast, effective per-flow credit-based flow control. The system can maintain state information of individual packet flows, which can be set up or released dynamically based on injected data. Each flow can be provided with a flow-specific input queue upon arriving at a switch. Packets of a respective flow can be acknowledged after reaching the egress point of the network, and the acknowledgement packets can be sent back to the ingress point of the flow along the same data path. As a result, each switch can obtain state information of each flow and perform flow control on a per-flow basis.
1. A method, comprising:
monitoring an input queue for a flow of packets received by a physical ingress port of a switch, wherein packets in the flow of packets comprise a source identifier indicating the source of the flow of packets and exclude a flow identifier;
upon receiving the flow of packets, assigning, by the switch, the flow identifier to the packets in the flow, the flow identifier being unique to the physical ingress port, and the flow identifier being local to the physical ingress port of the switch;
determining that an amount of data of the flow stored in the input queue exceeds a threshold;
setting an indicator which indicates that flow control is in effect with respect to an upstream switch; and
generating a flow control message for the upstream switch, wherein the flow control message comprises the flow identifier and a headroom value that indicates an amount of data of the flow the upstream switch is allowed to forward toward an input port of a downstream switch, thereby controlling a rate at which acknowledgement (ACK) packets of the flow are forwarded by the upstream switch to the downstream switch, the ACK packets having been generated upon receipt of packets of the flow at the flow's destination,
when a packet is delivered to a destination egress switch, the destination egress switch enabled to generate and send back an acknowledgement (ACK) in an upstream direction along a same data path to the switch being an ingress edge device.
2. The method of claim 1 , further comprising:
forwarding a second packet of the flow from the input queue to an output port;
determining, subsequent to forwarding the second packet, that the amount of data of the flow stored in the input queue is less than a target watermark value; and
setting the headroom value in the flow control message to be greater than a size of the forwarded packet.
3. The method of claim 1 , further comprising:
forwarding a second packet of the flow from the input queue to a physical egress port;
determining, subsequent to forwarding the second packet, that the amount of data of the flow stored in the input queue is greater than a target watermark value; and
setting the headroom value in the flow control message to be less than a size of the forwarded packet.
4. The method of claim 1 , further comprising:
determining that the data of the flow stored in the input queue is greater than a high watermark value; and
setting the headroom value to indicate that the upstream switch is precluded from forwarding any data of the flow toward the physical ingress port.
5. The method of claim 1 , further comprising:
determining that the amount of data of the flow stored in the input queue is less than a drop watermark value;
clearing the indicator; and
generating a flow control message for the upstream switch to discontinue flow control, thereby allowing the upstream switch to transmit data of the flow without restriction.
6. The method of claim 1 , wherein the packets in the flow are assigned the flow identifier that corresponds with a dedicated, stream-specific input buffer that is configured to individually control the packets in the flow that are associated with the flow identifier.
7. The method of claim 1 , wherein the packets in the flow are acknowledged after reaching an egress point of a network where the switch is located.
8. The method of claim 1 , further comprising:
upon determining that the input queue excludes packets in the flow of packets, releasing the flow identifier for reuse of the flow identifier for a second flow of packets.
9. A switch, comprising:
an input queue monitor to monitor an input queue for a flow of packets corresponding to a physical ingress port of the switch, wherein packets in the flow of packets comprise a source identifier indicating the source of the flow of packets and exclude a flow identifier;
an input queue management logic block to:
upon receiving the flow of packets, assigning, by the switch, the flow identifier to the packets in the flow, the flow identifier being unique to the physical ingress port, and the flow identifier being local to the physical ingress port of the switch;
determine that an amount of data of the flow stored in the input queue exceeds a threshold; and
a flow control logic block to:
set an indicator which indicates that flow control is in effect with respect to an upstream switch; and
generate a flow control message for the upstream switch, wherein the flow control message comprises the flow identifier and a headroom value that indicates an amount of data, including acknowledgement (ACK) packets of the flow the upstream switch is allowed to forward toward an input port of a downstream switch, thereby controlling a rate at which the flow is forwarded by the upstream switch to the downstream switch, the ACK packets having been generated upon receipt of packets of the flow at the flow's destination,
when a packet is delivered to a destination egress switch, the destination egress switch enabled to generate and send back an acknowledgement (ACK) in an upstream direction along a same data path to the switch being an ingress edge device.
10. The switch of claim 9 , wherein the flow control logic block is further to:
determine, subsequent to a second packet of the flow being forwarded from the input queue to an output port, that the amount of data of the flow stored in the input queue is less than a target watermark value; and
set the headroom value in the flow control message to be greater than a size of the forwarded packet.
11. The switch of claim 9 , wherein the flow control logic block is further to:
determine, subsequent to a second packet of the flow being forwarded from the input queue to an output port, that the amount of data of the flow stored in the input queue is greater than a target watermark value; and
set the headroom value in the flow control message to be less than a size of the forwarded packet.
12. The switch of claim 9 , wherein the flow control logic block is further to:
determine that the data of the flow stored in the input queue is greater than a high watermark value; and
set the headroom value to indicate that the upstream switch is precluded from forwarding any data of the flow toward the input port.
13. The switch of claim 9 , wherein the flow control logic block is further to:
determine that the amount of data of the flow stored in the input queue is less than a drop watermark value;
clear the indicator; and
generate a flow control message for the upstream switch to discontinue flow control, thereby allowing the upstream switch to transmit data of the flow without restriction.
14. A non-transitory computer-readable storage medium storing a plurality of instructions executable by a processor, the plurality of instructions when executed by the processor cause the processor to:
monitor an input queue for a flow of packets received by a physical ingress port of a switch, wherein packets in the flow of packets comprise a source identifier indicating the source of the flow of packets and exclude a flow identifier;
upon receiving the flow of packets, assign the flow identifier to the packets in the flow, the flow identifier being unique to the physical ingress port, and the flow identifier being local to the physical ingress port of the switch;
determine that an amount of data of the flow stored in the input queue exceeds a threshold;
set an indicator which indicates that flow control is in effect with respect to an upstream switch; and
generate a flow control message for the upstream switch, wherein the flow control message comprises the flow identifier and a headroom value that indicates an amount of data of the flow the upstream switch is allowed to forward toward an input port of a downstream switch, thereby controlling a rate at which acknowledgement (ACK) packets of the flow are forwarded by the upstream switch to the downstream switch, the ACK packets having been generated upon receipt of packets of the flow at the flow's destination,
when a packet is delivered to a destination egress switch, the destination egress switch enabled to generate and send back an acknowledgement (ACK) in an upstream direction along a same data path to the switch being an ingress edge device.
15. The non-transitory computer-readable storage medium of claim 14 , the processor is further configured to:
forward a second packet of the flow from the input queue to an output port;
determine, subsequent to forwarding the second packet, that the amount of data of the flow stored in the input queue is less than a target watermark value; and
set the headroom value in the flow control message to be greater than a size of the forwarded packet.
16. The non-transitory computer-readable storage medium of claim 14 , the processor is further configured to:
forward a second packet of the flow from the input queue to a physical egress port;
determine, subsequent to forwarding the second packet, that the amount of data of the flow stored in the input queue is greater than a target watermark value; and
set the headroom value in the flow control message to be less than a size of the forwarded packet.
17. The non-transitory computer-readable storage medium of claim 14 , the processor is further configured to:
determine that the data of the flow stored in the input queue is greater than a high watermark value; and
set the headroom value to indicate that the upstream switch is precluded from forwarding any data of the flow toward the physical ingress port.
18. The non-transitory computer-readable storage medium of claim 14 , the processor is further configured to:
determine that the amount of data of the flow stored in the input queue is less than a drop watermark value;
clear the indicator; and
generate a flow control message for the upstream switch to discontinue flow control, thereby allowing the upstream switch to transmit data of the flow without restriction.
19. The non-transitory computer-readable storage medium of claim 14 , wherein the packets in the flow are assigned the flow identifier that corresponds with a dedicated, stream-specific input buffer that is configured to individually control the packets in the flow that are associated with the flow identifier.
20. The non-transitory computer-readable storage medium of claim 14 , wherein the packets in the flow are acknowledged after reaching an egress point of a network where the switch is located.
21. The non-transitory computer-readable storage medium of claim 14 , the processor is further configured to:
upon determining that the input queue excludes packets in the flow of packets, release the flow identifier for reuse of the flow identifier for a second flow of packets.