Network-processor accelerator

• US 20050074005A1
• Filed: 10/04/2004
• Published: 04/07/2005
• Est. Priority Date: 10/06/2003
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Okuno, Michitaka
• Granted Patent: US 7,440,457 B2
• Time in Patent Office: Days
• US Class Current: 370/389
• CPC Class Codes: H04L 45/00 Routing or path finding of packets in data switching networks routing or path finding in wireless networks H04W40/00
• CPC Class Codes: H04L 45/60 Router architectures
• CPC Class Codes: H04L 49/3009 Header conversion, routing tables or routing tags

Abstract

In order to perform cache processing for the received packets, a network apparatus is provided with a network-processor accelerator for caching the process result of a network processor. Accordingly, the network apparatus of the present invention ensuring higher packet-processing throughput can be realized by improving the packet-processing throughput without increase in the chip area, increase in the power consumption, and shortage in an external connected memory bandwidth.

Claims

1. A network apparatus comprising:

a network interface for transmitting and receiving packets to and from a network;

a network processors for transferring the received packets to a next transfer destination;

a table memory for storing a table referred for the transfer of packets;

a cache memory for storing result of process of the network processor; and

a network-processor accelerator for transferring the received packets by referring to the cache memory;

wherein the network-processor accelerator transfers the received packets having the destination addresses which are identical to that of the received packets already processed with the network processor.

3. The network apparatus according to claim 1, comprising:

a first network-processor accelerator including a first integrated cache memory;

a second network-processor accelerator including a second built-in cache memory;

a first network processor connected to the first network-processor accelerator;

a second network processor connected to the second network-processor accelerator;

a switch fabric; and

an input switch fabric interface and an output switch fabric interface connected to the switch fabric;

wherein the first network-processor accelerator is allocated between the network interface and the input switch fabric interface, while the second network-processor accelerator is allocated between the network interface and the output switch fabric interface.

4. The network apparatus according to claim 1,

wherein the network processor is integrated the network-processor accelerator.

5. The network apparatus according to claim 1,

wherein the network processor comprises the table memory.

6. The network apparatus according to claim 1,

wherein a plurality of other network-processor accelerators are connected between the network processor and the network-processor accelerator.

7. The network apparatus according to claim 1, further comprising:

a network-processor accelerator including a integrated cache memory;

a network processor connected to the accelerator;

a switch fabric;

an input switch fabric interface and an output switch fabric interface connected to the switch fabric;

wherein the network-processor accelerator is provided in common between the input switch fabric interface and the output switch fabric interface.

8. The network apparatus according to claim 3,

wherein the first network-processor accelerator and the second network-processor accelerator are constituted by the identical network-processor accelerator, and the first network processor and the second network processor are constituted by the identical network processor.

9. The network apparatus according to claim 1, further comprising:

a plurality of network processors; and

an interface unit for switching the connections between a plurality of network processors and the network-processor accelerator.

10. The network apparatus according to claim 1, further comprising:

a packet memory which is external connected to the network-processor accelerator or the network processor to temporarily store the received packets.

11. The network apparatus according to claim 10,

wherein the network-processor accelerator and the network processor are independently provided with the external connected packet memory.

12. The network apparatus according to claim 1,

wherein the relationship of NPin+NPout>

NAin is established among the number of wires NAin of the interface unit for receiving the packets of the network-processor accelerator, the number of wires NPin of the interface unit for receiving the packets of the network processor, and the number of wires NPout of the interface unit for sending the packets of the network processor.

13. The network apparatus according to claim 1,

wherein the relationship of NPin=NPout is established among the number of wires NAin of the interface unit for receiving the packets of the network-processor accelerator, the number of wires NPin of the interface unit for receiving the packets of the network processor, and the number of wires NPout of the interface unit for sending the packets of the network processor.

14. The network apparatus according to claim 1,

wherein the relationship of NPin+NPout=NAin is established between the number of wires NPin of the interface unit for receiving the packets of the network processor and the number of wires NPout of the interface unit for sending the packets of the network processor.

15. The network apparatus according to claim 1,

wherein the relationship of BWPin<

BWAin is established between the bandwidth BWAin of the interface unit for receiving the packets of the network-processor accelerator and the bandwidth BWPin of the interface unit for receiving the packets of the network processor.

16. The network apparatus according to claim 1,

wherein the relationship of BWPin=BWPout is established between the bandwidth BWPin of the interface unit for receiving the packets of the network processor and the bandwidth BWPout of the interface unit for sending the network processor.

2. A network apparatus comprising:

a network interface for transmitting or receiving packets to a network;

a first processor for transferring the received packets to a next transfer destination;

a table memory for storing a table referred for the transfer process;

a cache memory for storing result of process of the first processor; and

a second processor for transferring the received packets by referring to the cache memory;

wherein the received packets are transferred by referring to the result of process stored in the cache memory when the destination addresses thereof are identical to that of the packets processed and stored in the cache memory.

17. A network-processor accelerator, comprising:

an input interface unit for receiving packets;

a packet memory for temporarily storing the received packets;

a packet analysis unit for obtaining the extracted information by extracting the necessary domain from the received packets;

a cache memory for storing result of processes required for the transfer process of the received packets;

a means for determining whether the result of process to be applied to the received packets exists or not by referring to the cache memory; and

an output interface for transmitting the V packets.

18. The network-processor accelerator according to claim 17, further comprising:

a means for processing a cache miss, that is generated if the result of process conducted for the received packets does not exist in the cache memory, when the cache miss is generated; and

an interface for NP for connection to the network processor.

19. The network-processor accelerator according to claim 18,

wherein the means for processing cache-miss comprising:

integrated processors for transferring the cache miss; and

an interface for referring to a routing table required by the integrated processors for the transfer process.

20. The network-processor accelerator according to claim 19,

wherein if the integrated processors cannot perform the transfer process of the received packets, the received packets are transferred to the interface for NP.

21. The network-processor accelerator according to claim 19, comprising the integrated network processor.

22. The network-processor accelerator according to claim 18, comprising in common a means for processing the cache miss in order to independently provide the other packet analysis means, cache-related means, transmitting packet generation means or the like for both input and output sides.

23. A network processor including a means for caching the result of packet process, comprising:

a primary table for management of packets when there is no relevant information in the cache; and

a secondary table for management of second and subsequent packets that are containing the relevant information not existing in the cache and are expected to provide the process result which is identical to that of the packets recorded in the primary table;

wherein the secondary tables are prepared in the number equal to the number of entries of the primary table for management of cache-miss packets and the number of entries of the respective secondary tables is all set to the constant value.

24. A network processor including a means for caching the result of packet process, comprising:

a primary table for management of packets when the relevant information does not exist in the cache; and

a secondary table for management of second and subsequent packets that are containing the relevant information not existing in the cache and are expected to provide the process result which is identical to that of the packets recorded in the primary table;

wherein the secondary table is placed under the management of a linked-list style.

25. A network processor including a means for caching result of packet process, further comprising:

a primary table for management of packets when there is no relevant information in the cache; and

a secondary table for management of the second and subsequent packets that are containing the relevant information not existing in the cache and are expected to provide the process result which is identical to that of the packets recorded in the primary table;

wherein the secondary table is placed under the management by utilizing the CAM or fully associative memory.

26. A network-processor accelerator, comprising:

an interface unit for temporarily storing the received packets;

a packet analysis unit for obtaining the extracted information by extracting the necessary domain from the received packets;

a cache memory for storing the process result required for transfer process of the received packets;

an address generation means for referring to the cache from the extracted information;

a means for determining whether the relevant information exists or not in the cache as the result of reference to the cache from the extracted information;

a means for reassembling the packets from the relevant information when the relevant information exists in the cache;

a primary table for management of the packets when the relevant information does not exist in the cache;

a secondary table for management of the second and subsequent packets that are containing the relevant information not existing in the cache and are expected to provide the process result which is identical to that of the packets recorded in the primary table;

an interface unit for transmitting the packets and necessary information to an external connected network processor for implementing, with a programmable manner, the processes of cache-miss packet recorded to the primary table;

an interface unit for receiving the processed packets and added information from the external connected network processor;

a means for recording the process result information of the processed packets to the cache;

a means for applying the process result to the relevant packets recorded to the secondary table;

a means for releasing the entries of the primary and secondary tables having completed the processes;

a means for generating the transmitting packets from the processed packets or the process result information; and

an interface unit for transmitting the processed packets.