Congestion control in a network

• US 20060104298A1
• Filed: 04/25/2005
• Published: 05/18/2006
• Est. Priority Date: 11/15/2004
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Gupta, Tanmay, McAlpine, Gary L., Wadekar, Manoj
• Granted Patent: US 7,733,770 B2
• Time in Patent Office: Days
• US Class Current: 370/412
• CPC Class Codes: H04L 47/10 Flow control; Congestion control
• CPC Class Codes: H04L 47/115 using a dedicated packet
• CPC Class Codes: H04L 47/263 Rate modification at the source after receiving feedback
• CPC Class Codes: H04L 49/90 Buffering arrangements
• CPC Class Codes: H04L 49/9078 using an external memory or storage device

Abstract

A source node receives related frames of data to be transmitted from the source node to a destination node and places the frames in a queue. The queue is associated with a communication path over which the related frames are to be transmitted to the destination node. An interface that couples the source node to the communication path receives an indication directed to the source node of traffic congestion in the communication path. In response, the source node controls the rate at which the related frames of data are transferred from the queue to a transmit buffer accessible to the interface, in order to relieve the traffic congestion in the communication path.

Claims

1. In a source node, a method for transmitting data therefrom, comprising:

receiving in a queue related frames of data to be transmitted from the source node to a destination node;

associating the queue with one of a plurality of communication paths over which the related frames may be transmitted to the destination node;

receiving at an interface that couples the source node to the communication path an indication directed to the source node of traffic congestion in the communication path; and

controlling a rate at which the related frames of data are transferred from the queue to a transmit buffer accessible to the interface responsive to the indication of traffic congestion.

2. The method of claim 1, wherein the plurality of communication paths comprises a switched interconnect, the source node is an ingress node located at an edge of the switched interconnect, and the destination node is an egress node located at the edge of the switched interconnect.

3. The method of claim 1, wherein the related frames of data are selected from a group consisting of a flow, group of flows, class of service, quality of service, type, connection, virtual circuit, and virtual local area network.

4. The method of claim 2, wherein the one communication path comprises a physical path from the ingress node through the switched interconnect to the egress node, and wherein the plurality of communication paths comprise a number of physical paths from the ingress node to a respective number of reachable egress nodes.

5. The method of claim 2, wherein associating the queue with one of a plurality of communication paths over which the related frames may be transmitted to the destination node comprises associating the queue with an address of an egress node.

6. The method of claim 2, wherein receiving an indication directed to the source node of traffic congestion in the communication path comprises a receiving an indication from a switch in the switched interconnect that detects traffic congestion caused at least in part by the transmission of related frames of data from the source node.

7. The method of claim 1, wherein controlling the rate at which the related frames of data are transferred from the queue to a transmit buffer of the interface further is responsive to timing information provided by the interface.

8. The method of claim 7, further comprising discarding a related frame of data from the queue responsive to the indication of traffic congestion and timing information.

9. A source node, comprising:

a computing module;

an interface to couple the computing module to a switched interconnect, the switched interconnect to provide a plurality of communication paths to a plurality of destination nodes, the interface including a path rate control interface, the path rate control interface to receive from the communication paths indications directed to the source node of traffic congestion in the communication paths;

a path rate control module coupled to the path rate control interface to receive therefrom the indications of traffic congestion;

a plurality of queues each to receive related frames of data from the computing module that may be transmitted from the source node;

a path association module to associate each queue with one of the plurality of communication paths over which the related frames may be transmitted to a destination node; and

a dequeue rate control module coupled to the plurality of queues, path association module, and path rate control module, to control the rate at which to dequeue related frames of data from each queue to one of a plurality of transmit buffers associated with the interface responsive to receipt of the indications of traffic congestion.

10. The source node of claim 9, wherein the dequeue rate control module and path rate control module are each implemented as a software driver.

11. The source node of claim 9, wherein the related frames of data received by a queue represent a flow of data traffic.

12. The source node of claim 9, wherein the dequeue rate control module to dequeue related frames from a plurality of queues to one transmit buffer according to an arbitration scheme.

13. The source node of claim 9, wherein the switched interconnect to operate in accordance with the Institute for Electrical and Electronics Engineers (IEEE) 802.3 standard for a Carrier Sense Multiple Access/Collision Detection (CSMA/CD) protocol, including Ethernet, operating in full-duplex mode over Fast Ethernet, Gigabit Ethernet, and 10 Gigabit Ethernet.

14. An ingress component, comprising:

an interface to couple the ingress component to a packet switched network including a plurality of unique paths to a respective plurality of egress components, the interface including a path rate control interface to receive indications of congestion in the paths;

a path rate control module coupled to the path rate control interface to receive therefrom the indications of traffic congestion;

a plurality of flow queues each to receive a flow of data frames that may be transmitted from the ingress component;

a path association module to associate each flow queue with one of the plurality of paths; and

a control module coupled to the flows queues, path association module, and path rate control module, to control the rate at which to transfer frames from the flow queues to a plurality of transmit buffers associated with the interface responsive to the indications of traffic congestion.

15. The ingress component of claim 14, wherein the ingress component is an end station, and the interface is a network interface controller to couple the end station to an internetwork of end stations.

16. The ingress component of claim 14, wherein the ingress component in a network layer router, and the interface is a network interface controller to couple the router to an internetwork of end stations and/or other routers.

17. The ingress component of claim 15, wherein the packet switched network is a layer 2 subnetwork.

18. An article of manufacture, comprising:

a medium providing instructions that when executed by a node, cause the node to;

associate a queue with one of a plurality of communication paths over which related frames of data in the queue may be transmitted to the other node;

receive from the communication paths an indication directed to the node of traffic congestion in the communication path; and

control a rate at which the related frames of data are transferred from the queue to a transmit buffer, from which to transmit the related frames to the other node, responsive to the indication of traffic congestion.

19. The article of manufacture of claim 18, wherein the medium further provides instructions that when executed by the node, cause the node to associate the queue with a destination address of a switch in a communication path via which the other node is reachable.

20. The article of manufacture of claim 18, wherein the medium further provides instructions that when executed by the node, cause the node to discard a related frame of data from the queue responsive to the indication of traffic congestion.