WO2008128464A1

WO2008128464A1 - Device and method for realizing transferring single stream by multiple network processing units

Info

Publication number: WO2008128464A1
Application number: PCT/CN2008/070733
Authority: WO
Inventors: Ke Chen; Zheng Li; Zhenhua Xu; Wenyang Lei
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2007-04-24
Filing date: 2008-04-17
Publication date: 2008-10-30
Also published as: CN101043460B; CN101043460A

Abstract

A device for realizing transferring single stream by multiple network processing units comprises: multiple network processing units, a data receiving management unit and a data transmitting management unit; the data receiving management unit allocates the received data to multiple network processing units to be processed; the data transmitting management unit combines all the data processed by multiple network processing units into single-stream data for transmission. Also a method for realizing transferring single stream by multiple network processing units is provided. This invention uses multiple NPs to improve the ability of single-stream transferring and to save the cost on the development and investment of high-performance NP.

Description

Device and method for realizing single stream forwarding of multi-network processing unit

The present invention relates to the field of communications, and in particular, to an apparatus and method for implementing single stream forwarding of a multi-network processing unit.

Background technique

With the rapid development of broadband networks and the increasing number of new applications, even the data communication equipment with the highest processing capacity in the prior art is under pressure from higher traffic. The existing forwarding processing functions are mainly implemented by devices such as a CP (Control Processor), an NP (Network Processor), and a TM (Traffic Management)/Fabric (Switching Network). The NP implements the core forwarding processing function of the data communication device, which can be implemented by a NP unit with a bidirectional processing function, as shown in FIG. 1; the user side service completes the physical layer processing of the physical line signal through the inbound interface unit, and performs parallel conversion and go Frame processing, partial Layer 2 overhead processing, etc., and then the service is sent to the NP unit for processing. If the protocol 4 text needs to be processed by the CP, the NP and the CP interact, and finally enter the network side through the traffic management/switching network unit; The network side service reaches the NP unit through the traffic management/switching network unit, and then the physical interface processing, the serial-to-parallel conversion, the framing processing, and the partial Layer 2 overhead processing of the physical line signal are performed by the outbound interface unit to reach the user side.

The NP function can also be implemented by two NP units with one-way processing function, as shown in FIG. 2: The service uplink direction completes physical layer processing, parallel-to-serial conversion, de-frame processing, and partial Layer 2 overhead of physical line signals through the inbound interface unit. After processing the functions, the user enters the uplink NP processing unit and enters the network side through the traffic management/switching network unit; the network side service enters the downlink NP processing unit through the traffic management/switching network unit, and then the physical interface signal is completed by the outgoing interface unit. Physical layer processing, serial-to-parallel conversion, framing processing, partial Layer 2 overhead processing, etc., are sent to the user side.

In the process of implementing the present invention, the inventors found that the existing methods have at least the following shortcomings. Point:

If the service traffic increases, the NP of the existing processing capability must not meet the requirements. If the existing forwarding processing architecture is adopted, the high-performance NP is the only option for solving large traffic. In order to improve the forwarding processing capability of data communication equipment, although high-performance NP can be developed to replace the existing low-processing capability NP, it will bring huge investment in research and development and waste of existing resources. In addition, if the next business volume is increased, it will lead to another high cost investment in hardware upgrades, so that the new high-performance NP will not be able to meet the rapid growth of business demand, which will eventually lead to network operator CapEx. (Capital Expense, basic capital expenditure). Summary of the invention

The embodiments of the present invention provide a method and a device for implementing high-performance single-flow forwarding processing by using multiple NPs, so as to reduce the cost of system upgrade caused by an increase in network traffic.

An embodiment of the present invention provides a device for implementing single-flow forwarding of a multi-network processing unit, including multiple network processing units, a data receiving management unit, and a data transmission management unit. The data receiving management unit allocates received data to the device. Said a plurality of network processing units for processing;

The data transmission management unit combines the data processed by the plurality of network processing units into a single stream data transmission.

The embodiment of the invention further provides a method for implementing single stream forwarding of a multi-network processing unit, including:

Allocating received data to multiple network processing units for processing;

The data processed by the plurality of network processing units is combined into a single stream data transmission.

In the embodiment of the present invention, the received data is separately sent to a plurality of network processing units for processing; and the data processed by the plurality of network processing units is reordered, the single stream data is synthesized, and the multi-NP single stream is realized. The function of transmitting, improving the forwarding capability of the device based on the use of multiple existing NPs, does not require the development of a new high-performance NP, thus saving R&D investment and development costs. DRAWINGS

1 is a system for implementing NP function in a bidirectional NP unit in the prior art; FIG. 2 is a system for implementing NP function in two unidirectional NP units in the prior art; FIG. 3 is a multi-network processing unit according to an embodiment of the present invention; Flow chart of single stream forwarding method;

4 is a structural diagram of a device for implementing single-stream forwarding of a multi-network processing unit according to an embodiment of the present invention;

5 is a structural diagram of a data receiving management unit in an embodiment of the present invention;

6 is a structural diagram of a data transmission management unit in an embodiment of the present invention;

7 is a structural diagram of a backup unit of a network processing unit backup control unit in a data receiving management unit according to an embodiment of the present invention;

FIG. 8 is a structural diagram of a network processing unit backup control subunit in a data transmission management unit according to an embodiment of the present invention. detailed description

The embodiment of the invention provides a method for implementing single-flow forwarding of a multi-network processing unit. As shown in FIG. 3, the method includes the following steps:

Step s301: The single stream data received from the user side or the network side is allocated to a plurality of network processing units for processing.

The processing specifically includes: sorting the received data, and assigning the sorting identifier, and then transmitting the data of the assigned sorting identifier to the plurality of network processing units. Due to the differences in the processing flow of the NP forwarding engine itself, the difference in access to the memory when the table entries are searched, and the differences in the PCB (Print Circuit Board) traces, the delay of the single-stream data is delayed. . Therefore, the sequence number byte is added before the data message before entering the NP, and then sent to the NP for processing.

In order to save network resources and improve the utilization efficiency of the network processing unit, load balancing, congestion control, and the like can also be performed on the received data. In addition, a backup network processing unit may be added to the system. If the trigger condition is met, for example, the network processing unit is faulty, the data is switched from the working network processing unit to the backup network processing unit. Yuan.

In step s302, the data processed by the plurality of network processing units is combined into a single stream data transmission.

The processing specifically includes: reordering the data processed by the plurality of network processing units according to the sorting identifier, and synthesizing the single stream data to be sent; the sorting and the ending of the serial number are completed by adding the relative direction logic to the sequence number.

In the above embodiment, when the total data traffic of the system is greater than the sum of the NP processing capabilities, or the NP fails, the traffic cannot be processed at the line rate, and when the data is congested, the congestion control strategy and algorithm need to be used. For example, using the FQ (Fair Queuing) algorithm, n NPs are n queues, and the timing query data is sent to the cache state before each NP. If the cache corresponding to one or all NPs overflows, it is considered that A problem occurs in the NP or all NPs, and the traffic originally allocated to the NP is discarded until the corresponding cache has no overflow.

In the above embodiment, the data traffic is allocated to multiple NPs in a reasonable and balanced manner according to the processing capabilities of each service and each NP, and the maximum efficiency utilization NP processing resource can use the load balancing algorithm. The load balancing algorithm has a polling equalization, a weighted polling equalization, a random equalization, a weighted random equalization, and a processing capability equalization. In this embodiment, each NP has the same performance and hardware and software configuration, so the polling equalization plus processing capability is used. Balanced integration is more appropriate. Specifically, the data packet entered by the interface is polled and assigned to each NP, such as the data packet P1 enters NP1, the data packet P2 enters NP2, .... the data packet Pn enters NPn, so that the data packet allocation is based on the NP processing capability. The packet processing rate is used for equalization. At the same time, considering that the length of the data packet changes, in the load balancing, the buffer remaining space on the NP sending side should also be considered to adjust the allocation. For example, when the data packet Pn+1 is entered into NP1 and the data packet Pn+2 is selected, NP2 is selected. ~ NPn corresponds to the largest cache remaining space, to achieve the purpose of optimal load balancing.

In the above embodiment, since the data bus bandwidth at the total traffic on the user side and the TM/Fabric side is the largest, and the data bus bandwidth on each NP side is much smaller, therefore, different bus standards are often selected on both sides. If there is such a difference, The interface conversion function can be used to seamlessly connect data, such as SERDES with large traffic on the user side or the network side. (Serializer/Deserializer, Serializer/Deserializer) bus, converted to SPI (Serial Peripheral Interface) 4.2, XAUI (Xialix Assistant Unit Interface), Xilinx Assistant Unit Interface, etc. interface.

In addition, the above congestion control function is usually added to the total traffic entrance instead of the NP entry. This is because: The router IP congestion control algorithm is for different connections, discards according to certain policies, and balances all traffic. Assigned to multiple NPs, so the same connection will be dispersed into n streams. If congestion control is placed in each NP, the weighted congestion control algorithm will not be implemented accurately, such as WFQ ( Weighted Fair Queuing). In the embodiment of the present invention, the message sequence number is used to ensure the strict sequence of the message, and the message sequence number is added at the total traffic, if at each NP. Discarding packets will result in missing and discontinuous message sequence numbers.

In addition, the core router is on a very important network node, and reliability is an important indicator. At present, physical link backup and important board-level backup are widely used to ensure high reliability. As the most important interface in the router, the NP forwarding processing module should also implement module-level n+1 backup. . In the embodiment of the present invention, the n+1 backup of the NP module can be easily implemented by using the network processing unit backup control function. For example, normal single-stream traffic requires n NPs to process, and NP backup units are added for backup. The fault detection mechanism of the NP is judged by the overflow status of the independent FIFO (First In First Out) sent to each NP. If the overflow state is detected at a certain time, the system starts the backup function and sends the backup function to the NP. The traffic is forwarded to the backup NP.

In the above embodiment, the message sorting and reordering function is a necessary function, and functions such as congestion control, load balancing, interface conversion, and backup control are optional functions, and may be provided at the same time, or only one or several items may be provided. Features.

The above method provided by the embodiment of the present invention can conveniently provide the n+1 backup capability of the NP module, improve the reliability of the system, save the R&D investment and development cost of the high-performance NP, and prolong the service life of the existing NP.

An embodiment of the present invention provides a device for implementing single-stream forwarding of a multi-network processing unit, including n-2 slices of NPs, which are completely peer-to-peer with each other, and simultaneously increase and decrease before multiple NPs. A special logic function is added, which can be implemented by an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuits). Specifically, as shown in FIG. 4, the method includes: a plurality of network processing units (which can be divided into an uplink/downlink network processing unit), a data receiving management unit 100, and a data sending management unit 200, wherein the data receiving management unit 100 and the data sending management unit 200 Connect to the switching network 300 with traffic management capabilities. The data reception management unit 100 distributes the single stream data received from the user side to a plurality of network processing units, and the data transmission management unit 200 combines the data processed by the plurality of network processing units into single stream data and transmits the data to the switching network 300; or data. The reception management unit 100 distributes the data received from the switching network 300 to a plurality of network processing units, and the data transmission management unit 200 combines the data processed by the plurality of network processing units into single stream data and transmits them to the user side. The plurality of network processing units, the data receiving processing unit 100, and the data transmission processing unit 200 are connected to the control processing unit, and are controlled and managed by the control processing unit.

Referring to FIG. 5, the data receiving management unit 100 further includes: a message sorting subunit 130, sorting the received data, and transmitting the data of the assigned sorting identifier to the plurality of network processing units respectively; the network processing unit backup controller The unit 140 is connected to the message sorting subunit 130, and is configured to switch the data processed by the network processing unit to the backup by the network processing unit when the triggering condition is met, that is, when the working network processing unit fails or the data carried exceeds the load. a network processing unit; load balancing sub-unit 120, load-balances the received data and distributes the received data to a plurality of network processing units through the message sorting sub-unit 130; the congestion control sub-unit 110, for entering the load balancing sub-unit The data before 120 is subjected to congestion control; the interface conversion sub-unit 150 performs interface conversion on the received data. The connection relationship of each subunit in this embodiment is only an application example, and is not limited to that shown in FIG. 5. For example, when there is no load balancing subunit 120, the message ordering subunit 130 is directly connected to the congestion control subunit 110 and the like.

Referring to FIG. 6, the data transmission management unit 200 further includes a message sorting sub-unit 230, which reorders the messages processed by the plurality of network processing units according to the sorting identifier, and synthesizes the single stream data transmission; the network processing unit backup control subunit 220, connected to the message de-sorting sub-unit 230, configured to: when the trigger condition is met, that is, the working network processing unit When the faulty or bearer data exceeds the load, the data processed by the network processing unit is switched by the network processing unit to the backup network processing unit; the interface conversion subunit 210 performs interface conversion on the data processed by the plurality of network processing units.

In the addition of the backup NP unit, the traffic of each NP is simultaneously connected to the NP backup, but the switch control is required, and the opening and closing of the switch is determined by the NP fault detection result, the network processing unit backup control subunit 140 in FIG. 5 and the The network processing unit backup control sub-unit 220 implements the principle as shown in FIG. 7 and FIG. 8, respectively, and the working NP1 to the working NPn are respectively connected to the backup NP through the switch. The switching time of the working NP and the backup NP is mainly determined by the fault detection time, which is a microsecond level. Of course, the network processing unit backup control subunit may also exist only in the data reception management unit 100 or the data transmission management unit 200.

The device provided by the embodiment of the present invention can conveniently provide the n+1 backup capability of the NP module, improve the reliability of the system, save the R&D investment and development cost of the high-performance NP, and prolong the service life of the existing NP.

Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is a better implementation. the way. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium, including a plurality of instructions for making a A computer device (which may be a personal computer, server, or network device, etc.) performs the methods described in various embodiments of the present invention. The above is only a few specific embodiments of the present invention, but the present invention is not limited thereto, and any changes that can be made by those skilled in the art should fall within the protection scope of the present invention.

The above is only the process and method embodiments of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention. Any modifications, equivalents, improvements, etc., which are made within the spirit and principles of the embodiments of the present invention, are It should be included in the scope of protection of the embodiments of the present invention.

Claims

Rights request

A device for implementing single-stream forwarding of a multi-network processing unit, comprising: a plurality of network processing units, a data receiving management unit, and a data transmission management unit; wherein the data receiving management unit allocates the received data to the Said a plurality of network processing units for processing;

2. The apparatus for implementing single stream forwarding of a multi-network processing unit according to claim 1, wherein:

The data receiving management unit further includes a message sorting subunit, performing message sorting on the received data, and transmitting data of the assigned sorting identifier to the plurality of network processing units respectively;

The data transmission management unit further includes a message de-sorting sub-unit, reordering the data processed by the plurality of network processing units according to the sorting identifier, and synthesizing the single-stream data transmission.

The device for implementing single-flow forwarding of a multi-network processing unit according to claim 2, wherein the device further comprises a backup network processing unit, configured to be processed by the network processing unit before processing the handover when the trigger condition is met The data receiving management unit further includes a network processing unit backup control subunit connected to the message sorting subunit; the data sending management unit further includes a network connected to the packet descrambling subunit Processing unit backup control subunit; the network processing unit backup control subunit, when the trigger condition is satisfied, causing data processed by the network processing unit to be switched by the network processing unit to the backup network processing unit.

The apparatus for implementing single-stream forwarding of a multi-network processing unit according to claim 2, wherein the data receiving management unit further comprises a load balancing sub-unit, load-balances the received data, and sorts by the message. The subunit equalizes the received data to the plurality of network processing units.

5. The apparatus for implementing single stream forwarding of a multi-network processing unit according to claim 4, The data receiving management unit further includes a congestion control sub-unit that performs congestion control on data before entering the load balancing sub-unit.

The device for implementing the single-flow forwarding of the multi-network processing unit according to claim 2, wherein the data receiving management unit further comprises an interface conversion sub-unit, performing interface conversion on the received data; The management unit further includes an interface conversion subunit that performs interface conversion on the processed data of the plurality of network processing units.

The apparatus for implementing single-stream forwarding of a multi-network processing unit according to any one of claims 1 to 6, wherein the data receiving unit receives user-side single-stream data, and the data sending unit sends the data to the network side. Single stream data; or the data receiving unit receives network side single stream data, and the data sending unit sends single stream data to the user side.

8. A method for implementing single stream forwarding of a multi-network processing unit, characterized in that:

Allocating received data to multiple network processing units for processing;

9. The method for implementing single stream forwarding of a multi-network processing unit according to claim 8, wherein:

The distributing the received data to the plurality of network processing units for processing specifically includes: performing message sorting on the received data, and transmitting the data of the assigned sorting identifier to the plurality of network processing units for processing;

The combining the data processed by the plurality of network processing units into the single stream data transmission specifically includes: reordering the data processed by the plurality of network processing units according to the sorting identifier, and synthesizing the single stream data.

The method for implementing the single-stream forwarding of the multi-network processing unit according to claim 9, wherein the distributing the received data to the plurality of network processing units for processing further comprises: when the trigger condition is met, The data processed by the network processing unit is switched by the network processing unit to the backup network processing unit.

The method for implementing single-stream forwarding of a multi-network processing unit according to claim 9, wherein before the allocating the received data to the plurality of network processing units, the method further comprises: performing load balancing on the received data. .

The method for implementing single-stream forwarding of a multi-network processing unit according to claim 9, wherein before the allocating the received data to the plurality of network processing units, the method further comprises: performing congestion control on the received data. .

The method for implementing the single-flow forwarding of the multi-network processing unit according to claim 9, wherein the distributing the received data to the plurality of network processing units further comprises: performing interface conversion on the received data; Before the combining the data processed by the plurality of network processing units into the single stream data, the method further comprises: performing interface conversion on the data processed by the plurality of network processing units.