Traffic management implementation method and apparatus, and network device

Patent metadata

Field	Value
Publication number	US-10129151-B2
Application number	US-201615258439-A
Country	US
Kind code	B2
Filing date	Sep 7, 2016
Priority date	Sep 5, 2014
Publication date	Nov 13, 2018
Grant date	Nov 13, 2018

Abstract

A traffic management (TM) implementation method and apparatus, and a network device, where the TM implementation apparatus is located aside a processor or a switching fabric chip, receives a packet management request sent by the processor or the switching fabric chip, where the packet management request includes a queue identifier, and the queue identifier is used to identify a flow queue in which the processor or the switching fabric chip stores a data packet, performs traffic management on the packet management request, and generates a packet management response according to a management result, where the packet management response includes a management indication and the queue identifier, sends the packet management response to the processor or the switching fabric chip such that the processor or the switching fabric chip processes, according to the management indication, the data packet in the flow queue corresponding to the queue identifier.

First claim

What is claimed is:

1. A traffic management implementation method, comprising:
receiving a packet management request from a processor or a switching fabric chip, the packet management request comprising a queue identifier that identifies a flow queue in which a data packet is stored;
determining, according to a congestion management policy and a buffer status of a scheduling queue, whether to discard the packet management request, the scheduling queue being used to store the packet management request;
discarding the packet management request when the packet management request is to be discarded;
generating a packet management response comprising a management indication and the queue identifier, the management indication being a discard processing indication;
storing the packet management request into the scheduling queue when the packet management request is not to be discarded; and
sending the packet management response to the processor or the switching fabric chip.

2. The method of claim 1 , further comprising:
performing dequeue scheduling on the scheduling queue according to a queue scheduling policy; and
generating the packet management response when the packet management request is dequeued from the scheduling queue,
wherein the management indication is a forward processing indication.

3. The method of claim 2 , the queue scheduling policy is performing scheduling by byte, the packet management request further comprises a packet length, and performing dequeue scheduling on the scheduling queue according to the queue scheduling policy comprises performing dequeue scheduling on the scheduling queue according to the queue scheduling policy and the packet length.

4. The method of claim 1 , wherein the congestion management policy is congestion tail drop, and determining, according to the congestion management policy and the buffer status of the scheduling queue, whether to discard the packet management request comprises:
determining, according to the congestion management policy, that the packet management request is to be discarded when a buffer of the scheduling queue is full; and
determining that the packet management request is not to be discarded when the buffer of the scheduling queue is not full.

5. The method of claim 1 , wherein the congestion management policy is congestion random drop, and determining, according to the congestion management policy and the buffer status of the scheduling queue, whether to discard the packet management request comprises:
acquiring, from the scheduling queue according to the congestion management policy, a packet that is ready to be randomly discarded when a buffer of the scheduling queue is full;
determining that the packet management request is to be discarded when the packet that is ready to be randomly discarded comprises the packet management request;
determining that the packet management request is not to be discarded when the packet that is ready to be randomly discarded does not comprise the packet management request; and
determining that the packet management request is not to be discarded when the buffer of the scheduling queue is not full.

6. A traffic management implementation method, comprising:
receiving a packet management request from a processor or a switching fabric chip, the packet management request comprising a packet length and a queue identifier that identifies a flow queue in which a data packet is stored;
performing traffic management on the packet management request;
generating a packet management response according to a management result, the packet management response comprising the packet length, a management indication and the queue identifier; and
sending the packet management response to the processor or the switching fabric chip.

7. A network device, comprising:
a traffic management (TM) implementation apparatus;
a switching fabric chip; and
a processor coupled to the TM implementation apparatus and the switching fabric chip, the processor comprising a first interface and a second interface, and the processor being configured to:
communicate with the TM implementation apparatus through the first interface;
communicate with the switching fabric chip through the second interface;
receive a data packet from the switching fabric chip via the second interface;
parse the data packet to acquire a flow identifier;
acquire a queue identifier of a flow queue according to the flow identifier;
store the data packet into the flow queue; and
send a packet management request to the TM implementation apparatus through the first interface, the packet management request comprising the queue identifier,

wherein the TM implementation apparatus is configured to:
receive the packet management request;
determine, according to a congestion management policy of the TM implementation apparatus and a buffer status of a scheduling queue of the TM implementation apparatus, whether to discard the packet management request, the scheduling queue being used to store the packet management request;
discard the packet management request when the packet management request is to be discarded;
generate a packet management response comprising a management indication and the queue identifier, the management indication being a discard processing indication;
store the packet management request into the scheduling queue when the packet management request is not to be discarded; and
send the packet management response to the processor, and

wherein the processor is further configured to:
discard, according to the discard processing indication, the data packet in the flow queue corresponding to the queue identifier;
receive the packet management response through the first interface; and
process, according to the management indication, the data packet in the flow queue corresponding to the queue identifier.

8. The network device of claim 7 , wherein the TM implementation apparatus is further configured to:
perform dequeue scheduling on the scheduling queue according to a queue scheduling policy of the TM implementation apparatus;
generate the packet management response when the packet management request is dequeued from the scheduling queue, the management indication being a forward processing indication, and
wherein the processor is further configured to forward, according to the forward processing indication, the data packet in the flow queue corresponding to the queue identifier.

9. The network device of claim 8 , wherein the queue scheduling policy is performing scheduling by byte, the packet management request further comprises a packet length, the TM implementation apparatus is further configured to perform dequeue scheduling on the scheduling queue according to the queue scheduling policy and the packet length, and the processor is further configured to acquire, according to the packet length, the data packet from the low queue corresponding to the queue identifier.

10. The network device of claim 7 , wherein the congestion management policy is congestion tail drop, and the TM implementation apparatus is further configured to:
determine, according to the congestion management policy, that the packet management request is to be discarded when a buffer of the scheduling queue corresponding to the queue identifier is full; and
determine that the packet management request is not to be discarded when the buffer of the scheduling queue corresponding to the queue identifier is not full.

11. The network device of claim 7 , wherein the congestion management policy is congestion random drop, and the TM implementation apparatus is further configured to:
acquire, according to the congestion management poli

Assignees

• Huawei Tech Co Ltd

Inventors

• Xiong Weibo
• Wu Binbin
• Wang Minghui
• Huo Guichao

Classifications

• H04L47/621

  Individual queue per connection or flow, e.g. per VC · CPC title

• H04L47/326

  with random discard, e.g. random early discard [RED] · CPC title

• H04L47/32 Primary

  by discarding or delaying data units, e.g. packets or frames · CPC title

• H04L43/062

  related to network traffic · CPC title

• H04L69/22

  Parsing or analysis of headers · CPC title