Packet scheduling using hierarchical scheduling process with priority propagation

• US 10,447,608 B2
• Filed: 11/14/2014
• Issued: 10/15/2019
• Est. Priority Date: 11/14/2014
• Status: Active Grant

• Current Assignee: Marvell Asia Pte Limited (Marvell Technology Group Limited)
• Original Assignee: Marvell Semiconductor (Marvell Technology Group Limited)
• Inventors: Daniel, Tsahi, Panchagnula, Vamsi
• Primary Examiner(s): Rutkowski, Jeffrey M
• Assistant Examiner(s): Ma, Basil
• Application Number:
  US14/542,393
• Publication Number:
  US 20160142331A1
• Time in Patent Office: 1,796 Days
• CPC Class Codes: H04L 47/60 implementing hierarchical scheduling
• CPC Class Codes: H04L 47/6215 Individual queue per QOS, rate or priority

Abstract

System and method of data routing according to a hierarchical scheduling process. Incoming data traffic is allocated to various queues of a buffer. A scheduling tree has a top level for queues, a bottom level for egress ports, and a plurality of intermediate levels corresponding to different granularities with respect to service categories. Each queue is assigned to a particular node in each intermediate level of the scheduling tree. The scheduling tree traverses through multiple scheduling stages from the bottom to the top level to select a winner node in each level based on a variety of fairness and differentiating variables. A queue associated with the winner nodes in various levels is selected for outgoing transmission at the selected egress port. Priority information is dynamically propagated from upper nodes to lower nodes such that a subsequent scheduling process uses the updated priority information.

Claims

1. A method of routing data traffic over a communication network, said method comprising:

mapping incoming data traffic into data groups, wherein a respective data group is assigned to nodes of different levels in a scheduling tree, and wherein a priority is associated with said respective data group;

changing said priority of said respective data group responsive to a new data packet mapped to said respective data group or a data packet leaving said respective data group;

responsive to said priority of said respective data group being changed, dynamically updating priorities of said nodes of said different levels;

traversing said scheduling tree from a bottom level to a top level to select a data group for transmission through an egress port of a plurality of egress ports of a data routing device, wherein said bottom level comprises said plurality of egress ports of said data routing device, wherein said traversing comprises;

selecting said egress port from said plurality of egress ports;

responsive to said selecting said egress port, selecting an upper node according to a scheduling process, wherein said scheduling process is based on;

state information of a lower node that is arranged in a lower level in said scheduling tree than said upper node, wherein said lower node has been selected according to a prior scheduling process during said traversing said scheduling tree, and

updated priorities of nodes in a same level with said upper node, wherein said upper node is arranged in a subtree of said lower node in said scheduling tree; and

identifying a data group, wherein said identified data group is associated with all nodes that are selected during said traversing said scheduling tree; and

sending data in said identified data group to said selected egress port for said transmission.

2. The method as described in claim 1, wherein said scheduling process comprises a Deficit Weighted Round Robin (DWRR) process, and wherein said scheduling process is further based on variables representing strict priority enable, weight, and token bucket.

3. The method as described in claim 1, wherein said dynamically updating said priorities comprises propagating an updated priority of said respective data group to said nodes in said different levels.

4. The method as described in claim 3, wherein said propagating comprises determining an updated priority for a first node that is associated with said respective data group by comparing updated priorities of a plurality of data groups that are associated with said first node.

5. The method as described in claim 1 further comprising buffering said incoming data traffic, wherein said data groups correspond to queues in a buffer.

6. The method as described in claim 1, wherein said nodes of said different levels correspond to network service categories in different granularity levels.

7. The method as described in claim 1, wherein selecting said egress port is performed in accordance with a pipeline scheduler process.

8. An apparatus for routing data over a communication network, said apparatus comprising:

ingress ports configured to receive data streams;

a buffer unit coupled to said ingress ports and configured to store said data streams;

egress ports coupled to said buffer unit and configured to transmit said data streams;

first logic coupled to said ingress ports and said buffer unit, and configured to map incoming data traffic into data groups, wherein a respective data group is assigned to nodes of different levels in a scheduling tree, and wherein a priority is associated with said respective data group; and

an arbiter coupled to said first logic and configured to;

change said priority of said respective data group responsive to a new data packet mapped to said respective data group or a data packet leaving said respective data group;

responsive to said priority of said respective data group being changed, dynamically update priorities of said nodes of said different levels;

traverse said scheduling tree from a bottom level to a top level to select a data group for transmission through an egress port of a plurality of egress ports, wherein said bottom level comprises said plurality of egress ports, wherein to traverse said scheduling tree, said arbiter is configured to;

select said egress port from said plurality of egress ports;

responsive to said selection of said egress port, select an upper node according to a scheduling process, wherein said scheduling process is based on;

state information of a lower node that is arranged in a lower level in said scheduling tree than said upper node, wherein said lower node has been selected according to a prior scheduling process during said traversal of said scheduling tree, and

updated priorities of nodes in a same level with said upper node, wherein said upper node is in a subtree of said lower node in said scheduling tree; and

identify a data group, wherein said identified data group is associated with all nodes that are selected during said traversal of said scheduling tree; and

send data in said identified data group to said selected egress port for said transmission.

9. The apparatus as described in claim 8, wherein said scheduling process comprises a Deficit Weighted Round Robin (DWRR) process, and is further based on variables representing strict priority enable, weight, and token bucket.

10. The apparatus as described in claim 8, wherein said scheduling tree is user customizable.

11. The apparatus as described in claim 8, wherein said arbiter is further configured to propagate an updated priority of said respective data group to said nodes of said different levels.

12. The apparatus as described in claim 8, wherein said arbiter is further configured to determine an updated priority for a first node of said nodes by comparing updated priorities of a plurality of data groups that are associated with said first node.

13. The apparatus as described in claim 8, wherein said first logic and said arbiter are integrated in an Application-Specific Integrated Circuit (ASIC).

14. A method of scheduling data traffic at a network switching device, said method comprising:

mapping incoming data traffic into data groups in a buffer, wherein a priority is associated with a respective data group of said data groups, wherein said data groups are classified based on a hierarchy, wherein;

a bottom level of said hierarchy comprises a plurality of egress ports of said network switching device; and

an intermediate level of said hierarchy comprises a plurality of categories, and wherein a respective category in said intermediate level has a subordinate category;

updating said priority of said respective data group responsive to a new data packet mapped to said respective data group or a data packet leaves said respective data group;

responsive to said priority of said respective data group being updated, propagating an updated priority of said respective data group to categories associated with said respective data group across said hierarchy;

traversing said hierarchy from said bottom level a top level thereof to select a data group for transmission through an egress port of said plurality of egress ports, wherein said traversing comprises;

selecting said egress port from said plurality of egress ports;

responsive to said selecting said egress port, selecting a category for a respective level according to a scheduling process, wherein said scheduling process is based on;

state information of a selected category in another level that is arranged in a lower level than said respective level, wherein said selected category in said another level has been selected according to a prior scheduling process during said traversing said hierarchy, and

updated priorities associated with categories in said respective level, wherein said selected category for said respective level is a subordinate of said selected category in said another level; and

identifying a data group associated with all categories in various levels of said hierarchy; and

sending data in said identified data group to said selected egress port for said transmission.

15. The method as described in claim 14, wherein said selecting said egress port is performed in accordance with a pipeline scheduling process.

16. The method as described in claim 14, wherein said propagating comprises computing an updated priority for a first category by selecting a highest priority among subordinate categories of said first category, and wherein said first category is associated with said respective data group.

17. The method as described in claim 16, wherein said scheduling process is further based on variables representing strict priority enable, weight, and credit bucket shaper.

18. The method as described in claim 17, wherein said scheduling process comprises a Deficit Weighted Round Robin (DWRR) process, and wherein said categories of said various levels of said hierarchy correspond to network user groups in different granularity levels.