SWITCHING NODE WITH LOAD BALANCING OF BURSTS OF PACKETS

• US 20130201826A1
• Filed: 08/25/2010
• Published: 08/08/2013
• Est. Priority Date: 07/09/2010
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Testa, Patrizia, Germoni, Angelo, Listani, Marco
• Granted Patent: US 9,276,870 B2
• Time in Patent Office: Days
• US Class Current: 370/230
• CPC Class Codes: H04J 14/0212 using optical switches or wavelength selective switches [WSS]
• CPC Class Codes: H04J 14/0282 WDM tree architectures
• CPC Class Codes: H04J 14/0283 WDM ring architectures
• CPC Class Codes: H04J 14/0286 WDM hierarchical architectures
• CPC Class Codes: H04L 47/10 Flow control; Congestion control
• CPC Class Codes: H04L 47/125 by balancing the load, e.g. traffic engineering
• CPC Class Codes: H04L 47/43 Assembling or disassembling of packets, e.g. segmentation and reassembly [SAR]
• CPC Class Codes: H04L 47/70 Admission control; Resource allocation
• CPC Class Codes: H04Q 2213/13106 Microprocessor, CPU
• CPC Class Codes: H04Q 2213/13164 Traffic (registration, measurement,...)
• CPC Class Codes: H04Q 2213/13166 Fault prevention
• CPC Class Codes: H04Q 2213/1331 Delay elements, shift registers
• CPC Class Codes: H04Q 2213/13389 LAN, internet

Abstract

A packet switching node is coupled by links to other nodes of a network, and receives and assembles packets belonging to a specified packet flow, into bursts of packets with a burst control packet indicating a sequence of the burst in the flow. The node determines whether to distribute the flow across several links. If so, the bursts are then forwarded for switching to the output ports of the selected links. Distributing the flow over multiple links can enable more flexible and efficient filling of allocated bandwidth on links, as traffic increases. To reduce the risk of losing the order of packets the sequence of the bursts is indicated for use in reordering at intermediate nodes during transmission through the network.

Claims

1. A method of operating a packet switching node coupled by links to other nodes, as an ingress node, the method having the steps of:

receiving packets belonging to a specified packet flow, to be sent on to a destination node;

assembling the received packets of that flow into bursts of packets, each burst having a burst control packet, the burst control packet indicating a sequence of the burst in the flow, to enable the sequence to be maintained after transmission;

determining whether to distribute the flow;

selecting, if the flow is to be distributed, at least two of the links to use for sending on the bursts of the flow towards the destination node;

distributing the bursts of the flow between the selected links by forwarding a first of the bursts for switching to a first output port, for transmission over a first of the selected links; and

forwarding a second of the bursts for switching to a second output port, for transmission over a second of the selected links.

2. The method of claim 1, the received packets being Ethernet packets, and the burst control packet being an Ethernet packet.

3. The method of claim 1, the node having at least two buffers for queueing the bursts before switching, each of the buffers being associated with at least one of the output ports, the method having the step of queueing the bursts in whichever of said buffers corresponds to their selected output port.

4. The method of claim 1, the node having a channel allocation table indicating a bandwidth available for each node on each link, and the distributing step comprising determining what proportion of a total bandwidth needed for the flow, is to be distributed to each of the selected links, based on an occupation level for each link.

5. The method of claim 1, having the step of controlling when the bursts are forwarded by sending a switch request for a given burst to an output controller for its selected output port, and forwarding the burst for switching when the output controller sends an acknowledgement to allow the switching.

6. The method of claim 5, the sending of the acknowledgement being dependent on a preceding burst in the sequence having been acknowledged.

7. The method of claim 1, having the step of controlling when the bursts are forwarded according to an indication of the sequence of the bursts of a given flow.

8. The method of claim 1, having the step of controlling when the bursts are forwarded, according to a rate limit for the flow for the link.

9. The method of claim 1, the flow having a specified class of service, and the step of controlling when the bursts are forwarded being made according to the class of service of the flow.

10. A method of controlling switching in a packet switching node, the node having a local input port for receiving packets belonging to a specified packet flow, to be sent on to a destination node, a burst assembler for assembling the received packets of that flow into bursts of packets, each burst having a burst control packet, the burst control packet indicating a sequence of the burst in the flow, to enable the sequence to be maintained after transmission, and a switch coupled to the local input port and to output ports, the method having the steps of:

determining whether to distribute the flow;

selecting, if the flow is to be distributed, at least two of the links to use for sending on the bursts of the flow towards the destination node; and

distributing the bursts of the flow between the selected links by;

forwarding a first of the bursts for switching to a first output port, for transmission over a first of the selected links; and

forwarding a second of the bursts for switching to a second output port, for transmission over a second of the selected links according to the indicated sequence for the flow.

11. A method of operating a packet switching node coupled by links to other nodes, as an egress node, the method having the steps of:

receiving at line input ports, bursts of packets belonging to a specified packet flow, sent over different links from another node, each burst having a burst control packet, the burst control packet indicating a sequence of the burst in the flow;

selecting a local output port for packets of the flow;

switching the bursts to the selected local output port; and

ordering the packets of the bursts of a same flow, according to the indicated sequence.

12. A method of operating a packet switching node coupled by links to other nodes, as an intermediate node, the method having the steps of:

receiving at line input ports, bursts of packets belonging to a specified packet flow, sent from another node, each burst having a burst control packet, the burst control packet indicating a sequence of the burst in the flow;

determining if the flow is distributed;

if not distributed, determining whether to distribute it over different links to different adjacent nodes;

if distributed, determining whether to recombine it;

if distributed and not to be recombined, selecting at least two of the links to use for sending on the bursts of the flow towards a destination node, and distributing the flow between the selected links by;

forwarding a first of the bursts for switching to a first output port, for transmission over a first of the selected links; and

forwarding a second of the bursts for switching to a second output port, for transmission over a second of the selected links according to the indicated sequence for the flow.

13. (canceled)

14. A traffic controller for controlling a packet switching node coupled by links to other nodes, the node having a local input port for receiving packets belonging to a specified packet flow, to be sent on to a destination node, a burst assembler for assembling the received packets of that flow into bursts of packets, each burst having a burst control packet, the burst control packet indicating a sequence of the burst in the flow, to enable the sequence to be maintained after transmission, and a switch coupled to the local input port and to output ports, the traffic controller being arranged to perform the following:

selecting at least two of the links to use for sending on the bursts of this flow towards the destination node; and

being coupled to the burst assembler to distribute the flow between the selected links by forwarding a first of the bursts from a first burst assembler queue to the switch for switching to a first output port, for transmission over a first of the selected links, and by forwarding a second of the bursts from a second burst assembler queue to the switch for switching to a second output port, for transmission over a second of the selected links.

15. The traffic controller of claim 14, wherein the node having at least two buffers for queueing the bursts before switching, each of the buffers being associated with at least one of the output ports, the traffic controller being coupled to control said buffers to queue the bursts in whichever of said buffers corresponds to their selected output port.

16. The traffic controller of claim 14, having an output controller associated with a group of output ports, and having a forwarding controller coupled to the output controller to control when the bursts are forwarded by sending a switch request for one of the bursts to the output controller associated with the group of output ports selected for the respective burst, and the forwarding controller being arranged to forward the respective burst for switching when it receives an acknowledgement of the request from the output controller to allow the switching.

17. The traffic controller of claim 14, the flow having a specified class of service, and the burst assembler having buffers for queueing the bursts according to a class of service of the flow.

18. A packet switching node coupled by links to other nodes, the node having a local input port for receiving packets belonging to a specified packet flow, to be sent on to a destination node, a burst assembler for assembling the received packets of that flow into bursts of packets, each burst having a burst control packet, the burst control packet indicating a sequence of the bursts in the flow, to enable the sequence to be maintained after transmission, and a switch coupled to the local input port and to output ports, the node also having:

a traffic controller being arranged to perform the following;

selecting at least two of the links to use for sending on the bursts of this flow towards the destination node; and

being coupled to the burst assembler to distribute the flow between the selected links by forwarding a first of the bursts from a first burst assembler queue to the switch for switching to a first output port, for transmission over a first of the selected links, and by forwarding a second of the bursts from a second burst assembler queue to the switch for switching to a second output port, for transmission over a second of the selected links.

19. A packet switching node of claim 18, wherein the node having at least two buffers for queueing the bursts before switching, each of the buffers being associated with at least one of the output ports, the traffic controller being coupled to control said buffers to queue the bursts in whichever of said buffers corresponds to their selected output port.

20. A packet switching node of claim 18, wherein the traffic controller has an output controller associated with a group of output ports, and having a forwarding controller coupled to the output controller to control when the bursts are forwarded by sending a switch request for one of the bursts to the output controller associated with the group of output ports selected for the respective burst, and the forwarding controller being arranged to forward the respective burst for switching when it receives an acknowledgement of the request from the output controller to allow the switching.

21. A packet switching node of claim 18, wherein the flow has a specified class of service, and the burst assembler having buffers for queueing the bursts according to a class of service of the flow.