Technologies for balancing throughput across input ports of a multi-stage network switch

Patent metadata

Field	Value
Publication number	US-10911366-B2
Application number	US-201715639816-A
Country	US
Kind code	B2
Filing date	Jun 30, 2017
Priority date	Jun 30, 2017
Publication date	Feb 2, 2021
Grant date	Feb 2, 2021

Abstract

Technologies for balancing throughput across input ports include a network switch. The network switch is to generate, for an arbiter unit in a first stage of a hierarchy of stages of arbiter units, turn data indicative of a set of turns in which to transfer packet data from devices connected to input ports of the arbiter unit. The network switch is also to transfer, with the arbiter unit, the packet data from the devices in the set of turns. Additionally, the network switch is to determine weight data indicative of the number of turns represented in the set and provide the weight data from the arbiter unit in the first stage to another arbiter unit in a subsequent stage to cause the arbiter unit in the subsequent stage to allocate a number of turns for the transfer of the packet data from the arbiter unit in the first stage.

First claim

The invention claimed is:

1. A network switch for balancing throughput across input ports, the network switch comprising:
a plurality of arbiter units configured in a hierarchy of stages, wherein each arbiter unit includes one or more input ports and an output port;
circuitry to:
generate, for an arbiter unit in a first stage of the hierarchy, turn data indicative of a set of turns in which to transfer packet data from one or more devices connected to the one or more of the input ports of the arbiter unit to a subsequent stage;
transfer, with the arbiter unit, the packet data from the one or more devices in the set of turns to the subsequent stage;
determine weight data indicative of a number of turns represented in the set, wherein each turn corresponds to a packet transferred from a different device; and
provide the weight data from the arbiter unit in the first stage to another arbiter unit in the subsequent stage of the hierarchy to cause the arbiter unit in the subsequent stage to allocate the number of turns indicated in the weight data for the transfer of the packet data from the arbiter unit in the first stage to a downstream device.

2. The network switch of claim 1 , wherein to generate the turn data comprises to:
receive weight data from one or more of the devices; and
add a number of turns represented in the received weight data to the turn data.

3. The network switch of claim 1 , wherein to generate the turn data comprises to:
determine whether one or more of the devices did not provide weight data to the arbiter unit in the first stage; and
generate, in response to a determination that one or more of the devices did not provide weight data to the arbiter unit in the first stage, turn data indicative of one turn for each device that did not provide weight data.

4. The network switch of claim 1 , wherein to transfer the packet data from the one or more devices in the set of turns comprises to:
reduce a turn value associated with an input port for each packet received through the input port;
determine whether the turn value associated with the input port is equal to zero; and
advance, in response to a determination that the turn value associated with the input port is equal to zero, to a second input port of the arbiter unit to transfer one or more packets through the second input port.

5. The network switch of claim 1 , wherein to transfer the packet data from the one or more devices comprises to transfer packet data from one or more compute devices.

6. The network switch of claim 1 , wherein to transfer the packet data from the one or more devices comprises to transfer packet data from one or more other arbiter units.

7. The network switch of claim 1 , wherein to provide the weight data from the arbiter unit to another arbiter unit comprises to send a message to the other arbiter unit that includes the weight data.

8. The network switch of claim 1 , wherein to provide the weight data from the arbiter unit to another arbiter unit comprises to write the weight data to a memory location that is accessible to the other arbiter unit.

9. The network switch of claim 1 , wherein to provide the weight data from the arbiter unit to another arbiter unit comprises to write the weight data to a model specific register associated with the other arbiter unit.

10. The network switch of claim 1 , wherein the circuitry is further to:
generate, for a second arbiter unit in the first stage of the hierarchy, turn data indicative of a set of turns in which to transfer packet data from one or more devices connected to one or more of the input ports of the second arbiter unit;
transfer, with the second arbiter unit, the packet data from the one or more devices in the set of turns;
determine weight data indicative of the number of turns represented in the set, wherein each turn corresponds to a packet received from a different device; and
provide the weight data from the second arbiter unit in the first stage to the arbiter unit in the subsequent stage of the hierarchy to cause the arbiter unit in the subsequent stage of the hierarchy to allocate an equal number of turns for the transfer of the packet data from the second arbiter unit in the first stage.

11. The network switch of claim 1 , wherein the circuitry is further to transmit, with the arbiter unit in the subsequent stage of the hierarchy, the packet data to a device coupled to the output port of the arbiter unit in the subsequent stage.

12. The network switch of claim 1 , wherein to determine the weight data comprises to dynamically determine a running average of a weight on a per input port basis.

13. One or more non-transitory machine-readable storage media comprising a plurality of instructions stored thereon that, in response to being executed, cause a network switch to:
generate, for an arbiter unit in a first stage of a hierarchy, turn data indicative of a set of turns in which to transfer packet data from one or more devices connected to one or more input ports of the arbiter unit to a subsequent stage;
transfer, with the arbiter unit, the packet data from the one or more devices in the set of turns to the subsequent stage;
determine weight data indicative of a number of turns represented in the set, wherein each turn corresponds to a packet transferred from a different device; and
provide the weight data from the arbiter unit in the first stage to another arbiter unit in a subsequent stage of the hierarchy to cause the arbiter unit in the subsequent stage to allocate the number of turns indicated in the weight data for the transfer of the packet data from the arbiter unit in the first stage to a downstream device.

14. The one or more non-transitory machine-readable storage media of claim 13 , wherein to generate the turn data comprises to:
receive weight data from one or more of the devices; and
add a number of turns represented in the received weight data to the turn data.

15. The one or more non-transitory machine-readable storage media of claim 13 , wherein to generate the turn data comprises to:
determine whether one or more of the devices did not provide weight data to the arbiter unit in the first stage; and
generate, in response to a determination that one or more of the devices did not provide weight data to the arbiter unit in the first stage, turn data indicative of one turn for each device that did not provide weight data.

16. The one or more non-transitory machine-readable storage media of claim 13 , wherein to transfer the packet data from the one or more devices in the set of turns comprises to:
reduce a turn value associated with an input port for each packet received through the input port;
determine whether the turn value associated with the input port is equal to zero; and
advance, in response to a determination that the turn value associated with the input port is equal to zero, to a second input port of the arbiter unit to transfer one or more packets through the second input port.

17. The one or more non-transitory machine-readable storage media of claim 13 , wherein to transfer the packet data from the one or more devices comprises to transfer packet data from one or more compute devices.

18. The one or more non-transitory machine-readable storage media of claim 13 , wherein to transfer the packet data from the one or more devices comprises to transfer packet data from one or more other arbiter units.

19. The one or more non-transitory machine-readable storage media of claim 13 , wherein to provide the weight data from the a

Assignees

• Intel Corp

Inventors

• Diesing Scott S
• Parker Michael A
• Cheng Albert S
• Ni Nan

Classifications

• H04L49/112

  Switch control, e.g. arbitration · CPC title

• H04L49/111

  Switch interfaces, e.g. port details · CPC title

• H04L47/60 Primary

  implementing hierarchical scheduling · CPC title

• H04L47/125 Primary

  by balancing the load, e.g. traffic engineering · CPC title

• H04L49/30 Primary

  Peripheral units, e.g. input or output ports · CPC title