Techniques to utilize queues for network interface devices

• US 8,098,676 B2
• Filed: 08/12/2004
• Issued: 01/17/2012
• Est. Priority Date: 08/12/2004
• Status: Active Grant

• Current Assignee: Tahoe Research Limited (f/k/a Learndale Limited) (Vector Capital Corporation)
• Original Assignee: Intel Corporation
• Inventors: Connor, Patrick L.
• Primary Examiner(s): Sheikh, Ayaz
• Assistant Examiner(s): Moutaouakil, Mounir
• Application Number:
  US10/917,728
• Publication Number:
  US 20060034310A1
• Time in Patent Office: 2,714 Days
• Field of Search: 370/401, 370/419, 709/250, 711158-159
• US Class Current: 370/419
• CPC Class Codes: H04L 49/90 Buffering arrangements
• CPC Class Codes: H04L 49/9063 Intermediate storage in different physical parts of a node or terminal

Abstract

Techniques to allocate packets for processing among multiple processor(s). In an embodiment, capability to receive packets from a plurality of network interfaces may be provided, where each of the network interfaces is associated with at least one receive queue. the plurality of network interfaces may use any of the receive queues. Packet processing may be performed among a plurality of processors. Other embodiments are also disclosed.

Claims

1. A method comprising:

receiving, at a host system having receive queues, packets from a plurality of network interfaces, each of the network interfaces associated with at least one receive queue from the receive queues, wherein the plurality of network interfaces are capable to use any of the receive queues;

providing availability of a plurality of processors for packet processing; and

providing allocation of packet processing among the plurality of processors based on a table which associates one or more of the receive queues with a processor of the plurality of processors based on a hash value, wherein the hash value corresponding to the processor is to be determined based on connection-specific information in a header of each of the packets, wherein the processor queues are to store the packets from the receive queues, wherein the table is to associate hash values with the receive queues based on an unload analysis, wherein the unload analysis is to consider which processor from the plurality of processors is least busy based on;

fullness of the one or more receive queues associated with the processor and processor utilization.

2. The method of claim 1, wherein the providing allocation comprises:

associating at least one TCP processing block for execution by each processor among the plurality of processors;

associating at least one queue with each TCP processing block; and

allocating each of the packets for processing by a selected TCP processing block.

3. The method of claim 2, wherein the allocating each of the packets for processing by a selected TCP processing block comprises:

determining a hash value for each packet based on processing of a header associated with each packet; and

selecting a queue based on the hash value.

4. The method of claim 1, further comprising:

providing load balancing of use of output receive queues by the plurality of network interfaces.

5. The method of claim 1, further comprising:

providing traffic balancing of packets capable of being received by the plurality of network interfaces.

6. The method of claim 5, wherein the providing traffic balancing comprises:

generating a request to a node capable of transmitting packets, wherein the request includes a request to transmit a packet to a specified network interface.

7. The method of claim 1, further comprising:

providing interaction with the plurality of network interfaces as a single network interface with a plurality of receive queues.

8. The method of claim 1, wherein the connection-specific information is to comprise N tuple information and wherein the N tuple information is to consist of:

a packet source IP address, a destination IP address, a source port, a destination port, and a protocol to be used.

9. An apparatus comprising:

a memory to receive packets, wherein the packets are provided by a plurality of network interfaces;

a plurality of processors; and

an intermediate driver to;

provide capability to the plurality of network interfaces to use receive queues associated with any of the network interfaces; and

provide allocation of packet processing among the plurality of processors based on a table which associates one or more of the receive queues with a processor of the plurality of processors based on a hash value, wherein the hash value corresponding to the processor is to be determined based on connection-specific information in a header of each of the packets, wherein the processor queues are to store the packets from the receive queues, wherein the table is to associate hash values with the receive queues based on an unload analysis, wherein the unload analysis is to consider which processor from the plurality of processors is least busy based on;

fullness of the one or more receive queues associated with the processor and processor utilization.

10. The apparatus of claim 9, wherein to provide allocation, the intermediate driver is to:

associate at least one TCP processing block for execution by each processor among the plurality of processors;

associate at least one queue with each TCP processing block; and

associate each of the packets for processing by a selected TCP processing block.

11. The apparatus of claim 10, wherein to associate each of the packets for processing by a selected TCP processing block, the intermediate driver is to:

determine a hash value for each packet based on processing of a header associated with each packet; and

select a queue based on the hash value.

12. The apparatus of claim 9, wherein to provide capability to the plurality of network interfaces to use receive queues associated with any of the network interfaces, the intermediate driver is to:

provide load balancing of use of output receive queues by the plurality of network interfaces.

13. The apparatus of claim 9, wherein the intermediate driver is further to:

provide traffic balancing of packets capable of being received by the plurality of network interfaces.

14. The apparatus of claim 13, wherein to provide traffic balancing of packets capable of being received by the plurality of network interfaces, the intermediate driver is to:

generate a request to a node capable of transmitting packets, wherein the request includes a request to transmit a packet to a specified network interface.

15. The apparatus of claim 9, wherein the connection-specific information is to comprise N tuple information and wherein the N tuple information is to consist of:

a packet source IP address, a destination IP address, a source port, a destination port, and a protocol to be used.

16. A computer-readable medium comprising instructions stored thereon which when executed by a machine cause the machine to:

provide availability of a plurality of processors for packet processing;

provide capability to a plurality of network interfaces to use receive queues associated with any of the network interfaces; and

provide allocation of packet processing among the plurality of processors based on a table which associates one or more of the receive queues with a processor of the plurality of processors based on a hash value, wherein the hash value corresponding to the processor is to be determined based on connection-specific information in a header of each of the packets, wherein the processor queues are to store the packets from the receive queues, wherein the table is to associate hash values with the receive queues based on an unload analysis, wherein the unload analysis is to consider which processor from the plurality of processors is least busy based on;

fullness of the one or more receive queues associated with the processor and processor utilization.

17. The computer-readable medium of claim 16, wherein to provide allocation of packet processing among the plurality of processors, the instructions are to:

associate at least one TCP processing block for execution by each processor among the plurality of processors;

associate at least one queue with each TCP processing block; and

associate each of the packets for processing by a selected TCP processing block.

18. The computer-readable medium of claim 17, wherein to associate each of the packets for processing by a selected TCP processing block, the instructions are to:

determine a hash value for each packet based on processing of a header associated with each packet; and

select a queue based on the hash value.

19. The computer-readable medium of claim 16, wherein the connection-specific information is to comprise N tuple information and wherein the N tuple information is to consist of:

a packet source IP address, a destination IP address, a source port, a destination port, and a protocol to be used.

20. A system comprising:

a computing platform including a plurality of processors and a memory device;

a plurality of network interfaces, each capable to transfer a packet into a payload buffer stored in the memory device;

a bus capable of providing intercommunication between the computing platform and the plurality of network interfaces; and

code that is capable of causing any of the processors to be operable to;

provide capability to the plurality of network interfaces to use any receive queue associated with any of the plurality of network interfaces; and

provide allocation of packet processing among the plurality of processors

based on a table which associates one or more of the receive queues with a processor of the plurality of processors based on a hash value, wherein the hash value corresponding to the processor is to be determined based on connection-specific information in a header of each of the packets, wherein the processor queues are to store the packets from the receive queues, wherein the table is to associate hash values with the receive queues based on an unload analysis, wherein the unload analysis is to consider which processor from the plurality of processors is least busy based on;

fullness of the one or more receive queues associated with the processor and processor utilization.

21. The system of claim 20, wherein the bus is compatible with PCI express.

22. The system of claim 20, wherein the bus is compatible with PCI.

23. The system of claim 20, wherein the computing platform further includes a bus interface communicatively coupled to the bus.

24. The system of claim 20, further including a second memory device communicatively coupled to the bus.

25. The system of claim 20, wherein the connection-specific information is to comprise N tuple information and wherein the N tuple information is to consist of:

a packet source IP address, a destination IP address, a source port, a destination port, and a protocol to be used.