WO2014023023A1

WO2014023023A1 - Multi-cpu packet processing method and system, switching unit and board

Info

Publication number: WO2014023023A1
Application number: PCT/CN2012/079959
Authority: WO
Inventors: 陈胜平; 雷霆; 龚小红
Original assignee: 华为技术有限公司
Priority date: 2012-08-10
Filing date: 2012-08-10
Publication date: 2014-02-13
Also published as: CN103765837B; CN103765837A

Abstract

Provided in an embodiment of the present invention are a multi-CPU packet processing method and system, switching unit and board, the method comprising: a switching unit receives a packet, and determines that the packet is a service packet; broadcasting a load detection request message via a service trunk port, such that each service CPU receiving the load detection request message returns a load detection response message via respective first ports, the load detection response message containing the current load ratio and management IP address of a corresponding service CPU; and transmitting the service packet to the service CPU having a minimum current load ratio according to the current load ratio and management IP address of the corresponding service CPU contained in each load detection response message. The present invention improves system resource utilization rate.

Description

The present invention relates to the field of communication and electronic device technologies, and in particular, to a multi-CPU message processing method and system, a switching unit, and a single board. Background technique

In the telecom equipment, in order to enhance the processing capability of the board, multiple processor CPUs are often set in the same board to form a multi-CPU board. For example, a microprocessor without internal interlocking water level (Microprocessor without Interconnected piped stages, MIPS) The Advanced Telecom Computing Architecture (ATCA) board generally integrates more than two CPUs. Since each CPU has independent memory and bus, it cannot use existing symmetric multiprocessing. Technology, because symmetric multiprocessing technology requires a shared memory subsystem and bus structure between CPUs.

The ATCA board of the existing MIPS architecture adopts asymmetric multi-processing technology. However, in the asymmetric multi-processing technology, external communication data is sent to each CPU through the interface device, and each CPU performs related service processing alone, and therefore, based on The asymmetric multi-processing technology that tasks and resources are processed by different processors cannot achieve load balancing between multiple CPUs and reduce the utilization of system resources. SUMMARY OF THE INVENTION The embodiments of the present invention provide a multi-CPU packet processing method and system, a switching unit, and a single board. The multiple CPUs of the asymmetric multi-processing technology can also implement load balancing between multiple CPUs and improve system resources. Utilization.

In a first aspect, an embodiment of the present invention provides a multi-CPU packet processing method, including: receiving a packet, and determining that the packet is a service packet;

And transmitting, by the service aggregation trunk port, a load detection request message, so that each service CPU that receives the load detection request message returns a load detection response message by using a first port, where the load detection response message includes a corresponding service CPU. Current load ratio and management IP address; And sending the service packet to the service CPU with the smallest current load ratio according to the current load ratio and the management IP address of the corresponding service CPU included in each load detection response message.

In a first possible implementation, the service packet is sent to the service CPU with the smallest current load ratio according to the current load ratio and the management IP address of the corresponding service CPU included in the load detection response message, specifically The implementation can also be:

Obtaining, according to the current load ratio of the corresponding service CPU and the management IP address included in the load detection response message, a management IP address corresponding to the service CPU having the smallest current load ratio; and according to the acquired management IP address, The first port of the service CPU with the smallest current load ratio sends the service packet to the service CPU corresponding to the acquired management IP address.

In a second possible implementation, the specific implementation of the method may also be:

Receiving a packet, and determining that the packet is a management control packet;

And sending, by the first management control port, the management control packet to a management CPU corresponding to the management port, so that the management CPU manages and controls the service CPUs.

In a third possible implementation, the specific implementation of the method may also be:

Receiving a packet, and determining that the packet is an authentication packet;

And sending, by the second management control port corresponding to the second port of the main service CPU, the authentication message to the main service CPU, where the main service CPU is one of the CPUs of the service CPUs.

In a second aspect, an embodiment of the present invention provides an exchange unit, including:

a determining module, configured to receive a packet, and determine that the packet is a service packet;

a broadcast module, configured to: after the determining module determines that the packet is a service packet, broadcast a load detection request message by using a service aggregation trunk port, so that each service CPU that receives the load detection request message passes respectively The respective first port returns a load detection response message, where the load detection response message includes a current load ratio and a management IP address of the corresponding service CPU, and a sending module, configured to detect, according to the load returned by each service CPU received by the broadcast module. The service packet is sent to the service CPU with the smallest current load ratio, which is the current load ratio of the corresponding service CPU and the management IP address.

In a third aspect, an embodiment of the present invention provides a switching unit, including: a processor, where the processor is specifically configured to: receive a packet, and determine that the packet is a service packet; And transmitting, by the service aggregation trunk port, a load detection request message, so that each service CPU that receives the load detection request message returns a load detection response message by using a first port, where the load detection response message includes a corresponding service CPU. Current load ratio and management IP address;

And sending the service packet to the service CPU with the smallest current load ratio according to the current load ratio of the corresponding service CPU and the management IP address included in the load detection response message.

In a fourth aspect, an embodiment of the present invention provides a multi-CPU packet processing system, configured to communicate with the outside of the system, including: each service CPU and a switching unit, where each service CPU includes at least two service CPUs. One of the service CPUs of each service CPU is a main service CPU; the exchange unit is respectively connected to the service CPUs, and specifically: the service aggregation trunk port of the switching unit is respectively the first of the service CPUs Ports are connected, and each of the second management control ports of the switching unit is respectively connected to the second port of each service CPU;

The switching unit is configured to receive a packet, and determine that the packet is a service packet. The service aggregation trunk port broadcasts a load detection request message, so that each service CPU that receives the load detection request message passes its own The first port returns a load detection response message, where the load detection response message includes a current load ratio of the corresponding service CPU and a management IP address; and a current load ratio and a management IP of the corresponding service CPU included in the load detection response message according to the load Address, the service message is sent to the service CPU with the smallest current load ratio;

Each of the service CPUs is configured to perform a corresponding service operation according to the received service packet.

The switching unit is configured to obtain a management IP address corresponding to the service CPU with the smallest current load ratio according to the current load ratio and the management IP address of the corresponding service CPU included in the load detection response message;

And sending, according to the obtained management IP address, the service packet to the service CPU corresponding to the acquired management IP address by using the first port of the service CPU with the smallest current load ratio.

In a first possible implementation, the system further includes: a management CPU, connected to the switching unit by using a first management control port;

The switching unit is further configured to receive a packet, and determine that the packet is a management control packet; and send, by using the first management port, the management control packet to the management CPU; The management CPU is configured to: if it is determined that the management control message includes a configuration command, send the configuration command to the switching unit by using the first management control port, so that the switching unit The configuration command is sent to the primary service CPU through a second management control port corresponding to the second port of the primary service CPU, so that the primary service CPU sends the configuration command to other services through the switching unit. a CPU, where the configuration command includes a logical network port and a service configuration parameter;

Each of the service CPUs is configured to perform a corresponding configuration operation according to the logical network port and the service configuration parameter included in the configuration command.

The logical network port and the service configuration parameter of each service CPU are the same;

The management CPU is further configured to: send the management control packet of the summary type to the switching unit by using the first management control port, if it is determined that the management control packet is a summary type of management control packet And the switching unit sends the management control packet of the summary type to each service CPU by using the second management control port, where the summary type of the management control packet includes a heartbeat detection packet or a query packet or a statistic At least one of the messages, so that each service CPU returns a management control response message to the management CPU through the switching unit;

The management CPU is further configured to: if the management control message is determined to be a non-aggregated type of management control message, send the non-aggregated type of management control message to the switch by using the first management control port. a unit, so that the switching unit sends the non-aggregated type of management control message to the primary service control port through a second management control port corresponding to the second port of the primary service CPU, so that the primary The service CPU returns a management control response message to the management CPU through the switching unit.

In a second possible implementation, the switching unit is further configured to receive a packet, where the packet is an authentication request authentication packet, and a second corresponding to the second port of the primary service CPU. And managing the control port, and sending the authentication packet to the primary service CPU.

In a fifth aspect, an embodiment of the present invention provides a board, where the board includes: the foregoing multi-CPU message processing system.

According to the foregoing technical solution, when receiving a service packet, the embodiment of the present invention sends a load detection request message to each service CPU through a preset service aggregation trunk port, so as to obtain a current load ratio of each service CPU and corresponding management. IP address, which is used to send service packets to the service CPU with the lowest load ratio through the service aggregation trunk port. The CPU is sent to the service CPUs, so that the CPUs of the service can improve the utilization of the system resources. Therefore, the technical solution of the embodiment can enable multiple CPUs based on the asymmetric multi-processing technology to implement multiple CPUs. Load balancing between.

Further, in the embodiment of the present invention, the packet is sent to the corresponding CPU through the preset port according to different types of the received packets, for example, the management control packet is sent to the management CPU through the first management control port. It is possible to realize unified management and control of each service CPU through the management CPU, thereby improving system reliability, scalability, and catastrophic resistance. For example, when one of the service CPUs fails, the management CPU can restart the service CPU. BRIEF DESCRIPTION OF THE DRAWINGS In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief description of the drawings used in the embodiments or the prior art description will be briefly described below. The drawings are some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive labor.

1 is a schematic flowchart of a packet processing method of a multi-CPU according to an embodiment of the present invention; FIG. 2 is a schematic structural diagram of a multi-CPU packet processing system applied to the embodiment shown in FIG. 1; A schematic structural diagram of an exchange unit provided by an embodiment;

4 is a schematic structural diagram of an exchange unit according to another embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a multi-CPU packet processing system according to another embodiment of the present invention; FIG. 6 is a signaling diagram between a management CPU and each service CPU in the system embodiment shown in FIG. The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. The embodiments are a part of the embodiments of the invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

FIG. 1 is a schematic flowchart of a packet processing method of a multi-CPU according to an embodiment of the present invention. As shown in FIG. 1 , when a switching unit receives a packet sent by any network entity outside the system, the The law specifically includes:

101. The switching unit receives the packet, and determines that the packet is a service packet.

In an optional implementation manner of the present invention, a virtual local area network (VLAN) configuration may be performed on a network port of the switching unit, and the specific configuration is as follows:

Configure the VLAN of the management port of the switching unit as 3000. The external port VLAN of the authentication Radius packet is 103, and the internal port VLAN of the authentication packet is 4003. The external port VLAN of the uplink service packet is 101. The upstream service is 3⁄4. The internal port VLAN of the downstream port is 4001; the external port VLAN of the downlink service is 102, and the internal port VLAN of the downlink service >3⁄4 is 4002; that is, the external port Radius_ext_vlan of the authentication message is 103;

The internal port of the authentication packet Radius_inner_vlan is 4003.

The uplink port of the uplink service ^艮文 is Uplink_ext_vlan is 101;

The internal port Uplink_inner_vlan of the uplink service packet is 4001.

The external port downlink_ext_vlan of the downlink service packet is 102;

The internal port downlink_inner_vlan of the downlink service packet is 4002.

The management port manager—vlan is 3000.

It should be noted that the above-mentioned switching unit provides two 10GE NICs for each service CPU. For the separation of the data layer and the control plane, one of the 10G ports sends and receives service packets (that is, data packets), and another 10G port transmits and receives. Manage control messages or authentication messages. The specific implementation may be as follows: Figure 2 is a schematic structural diagram of a multi-CPU packet processing system applied in the embodiment shown in Figure 1. As shown in Figure 2, the xe1 and xe8 interfaces of the switching unit are used as service aggregation trunk ports (sal interface). Connects to the first port of each service CPU (the logical interface ID is spil) and allows only service packets to pass. The xe2 interface and the xe7 interface of the switching unit are configured as the second management control port, and are respectively connected to the second port of each service CPU (the logical interface identifier is spiO), and only the management control packet or the authentication packet is allowed to pass.

Further, the VLANs of the interfaces ge1, ge2, ge4, xe2, and xe7 in the switching unit are 3000; the VLANs of the xe3, xe4, and xe2 interfaces are 103 and 4003; and the VLANs of the xe3, xe4, and sal interfaces are 101, 4001, and 102. 4002, ie

Vlan3000 ge1, ge2; ge4; xe2, xe7;

Vlan103 xe3,xe4; xe2;

Vlan4003 xe3,xe4; xe2; Vlan101 xe3,xe4; sal;

Vlan4001 xe3,xe4; sal;

Vlan102 xe3,xe4; sal;

Vlan4002 xe3,xe4; sal;

Trunk sal xe1 ,xe8;

In other words, the ge1, ge2, ge4, xe2, and xe7 interfaces can allow management packets to pass. The xe3 and xe4 interfaces can allow authentication packets or uplink or downlink service packets to pass. The xe2 interface can allow authentication packets. The sa1 interface can allow uplink service packets or downlink service packets to pass.

For example, if the incoming physical interface of the packet is gel and the VLAN is 3000, the received packet is determined to be a management control packet.

For example, if the incoming physical interface of the packet is xe3 and the VLAN is 103 or 4003, the received packet is determined to be an authentication packet.

For example, if the incoming physical interface of the packet is xe3 and the VLAN is 101 or 102 or 4001 or 4002, the received packet is determined to be a service packet.

102. The service load aggregation request message is broadcasted by the service aggregation trunk port, so that each service CPU that receives the load detection request message returns a load detection response message by using a first port, where the load detection response message includes a corresponding service. The current load ratio of the CPU and the management IP address.

In an optional implementation of the present invention, each service CPU communicates with the outside through a respective first port and a second port, wherein the logical interface identifier of the first port of each service CPU is spi1, respectively, and is exchanged. As shown in Figure 2, the first port of the first service CPU is connected to the Xe1 interface of the switching unit, and the first port of the second service CPU is connected to the Xe8 interface of the switching unit, and only the service report is allowed. The logical port of the second port of each service CPU is identified as spiO, the second port of the first service CPU is connected to the Xe2 interface of the switching unit, and the second port of the second service CPU is connected to the Xe7 interface of the switching unit, Allows management control packets or authentication packets to pass.

In an alternative embodiment of the present invention, a virtual local area network (LAN) can be configured for the network port of each service CPU. Table 1 shows the virtual local area network configuration of each service CPU. The specific configuration is shown in Table 1. : Z - 0 2Z 01'9 : V6:6zl:08:00 !ds 01

Zl- 0 I Z 2Z 01 9 : V6:6zl:08:00 !ds 01

Zl- 0 I 10 2Z 01'99 : V6:6zl:08:00 !ds OP

Zl- 0 I zo 2Z 01 99 : V6:6zl:08:00 !ds OP

Zl- 0 0 8U I 2Z

:66:6zl:08:00 !ds 0|B-pUBlS 01

Zl- 0 0 8U z 2Z

:66:6zl:08:00 !ds 0|B-pUBlS

Zl- 0 0 8U ε 2Z

:66:6zl:08:00 !ds 0|B-pUBlS 01

Zl- 0 0 8U 10 2Z '99 :66:6zl:08:00 !ds 0|B-pUBlS OP

Zl- 0 0 8U zo 2Z '99 :66:6zl:08:00 !ds 0|B-pUBlS OP

Zl- 0 0 8U εο 2Z 'Z9 :66:6zl:08:00 !ds 0|B-pUBlS OP

Zl- 0 0 00 9.1 8 : 66: 6zl: 08:00 !ds οε

P!~J!

~Λμά

UBIA

OVIAI Ι !Λ

9dA丄 dl |SB|/\|

1? Leaf dl dl i

6S66.0/ZT0ZN3/X3d εζοε 请ΐοζ OAV 10.71.1 255.255.255.1 40 float spi 00:80:F9:99: 67.10 28 03 0 0E:12

10.71.6 255.255.255.1 10 float spi 00:80:F9:99: 7.10 28 3 0 0E:12

In an optional implementation manner of the present invention, a virtual local area network is configured on a network port of each service CPU, and the specific implementation may be:

Typedef struct

{

Uint32_t ip; IP IP configured for VLAN uint32_t net_mask; 〃 network mask

Uint16_t vlan_id; //vlan-id

u i nt8_t phyjfjd; 〃 corresponding physical port. uint8_t vmac[MAC_ADDR_LEN]; 〃 virtual MAC

Uint8_t ip_type; 〃0-single IP 1-Mng IP 2-floating IP atomic_t refcnt;

} lbu_interface_ip_t.

It should be noted that the logical interface identifier phyjfjd indicates which physical port sends and receives the packet corresponding to the logical interface identifier. The value of Phyjfjd can be, for example, 0-spiO, 1-spi1.

In an optional embodiment of the present invention, the configuration rules of phyjfjd are as follows:

The logical interface identifier of the processing authentication message phyjfjd is configured as spiO;

Configure the logical interface identifier phyjfjd of the management control packet (such as configuration, query, and detection packet) as spiO;

Configure the logical interface identifier phyjfjd of the floating interface that configures the upstream service packets to be spit. Configure the logical interface identifier phyjfjd of the floating interface that configures the downlink service packets as the spi. Configure the logical interface identifier phyjfjd of the single-interface interface that configures the uplink service packets as spiO. The logical interface identifier phyjfjd of the stand-alone interface of the downlink service packet is configured as spi0. It should be noted that, in this embodiment, the logical network port and the service configuration parameter configured by each service CPU are the same. In specific implementation, as shown in FIG. 2, for example, a management program is created on the management CPU, and the management CPU passes. The hypervisor configures and maintains the management IP of each service CPU (IP address in Table 1), and the management CPU communicates with each service CPU through the switching unit. Management CPU can Transmitting the configuration command to the first service CPU by using the second control port corresponding to the second port of the first service CPU (the primary service CPU) in the switching unit, so that the first service CPU sends the configuration command synchronously The second service CPU (other service CPU) includes the logical network port and the service configuration parameter. The second service CPU performs corresponding configuration operations according to the logical network port and service configuration parameters included in the configuration command.

In an optional implementation manner of the present invention, in order to implement load balancing between the service CPUs, the switching unit may broadcast a load detection request message to each service CPU through a preset trunk port to obtain the current load of each service CPU. Ratio

Each service CPU obtains a corresponding current load ratio according to the received load detection request message, and respectively returns a load detection response message to each switching unit through a corresponding first port (logical interface identifier spil), and each load detection response message It includes the current load ratio and management IP address of the corresponding service CPU.

103. Send the service packet to the service CPU with the smallest current load ratio according to the current load ratio of the corresponding service CPU and the management IP address included in each load detection response message.

In an optional implementation manner of the present invention, the switching unit obtains the management IP corresponding to the service CPU with the smallest current load ratio according to the current load ratio and the management IP address of the corresponding service CPU included in the load detection response message returned by each service CPU. An address is sent to the service corresponding to the acquired management IP address by the first port of the service CPU having the smallest ratio of the current load according to the acquired management IP address. In a specific implementation, as shown in FIG. 2, assuming that the current load ratio of the second service CPU is smaller than the current load ratio of the first service CPU, the switching unit may pass the service packet through the Xe8 interface of the switching unit and the second service CPU. A port (logical interface identifier spil) sends a service packet to the second service CPU.

In an optional implementation manner of the present invention, the switching unit receives the packet, and if it is determined that the packet is a management control packet, the preset first management control port, as shown in the ge4 interface of FIG. 2, The management control message is sent to the corresponding management CPU, so that the management CPU manages and controls each service CPU according to the received management control message.

The management CPU manages and controls each service CPU according to the received management control message. The specific implementation can be:

If the management CPU determines that the management control message is a summary type of management control message, the management CPU sends the summary type management control message to the switch through the first management control interface (ge4 interface). The unit, the switching unit sends a summary type of management control message to each service CPU through a preset second management control port, such as the xe2 and xe7 interfaces shown in FIG. 2, where each service CPU passes its respective corresponding The second port (the logical interface identifier is spiO) sends a management control response message to the switching unit, so that the switching unit sends the management control response message sent by each service CPU to the management CPU through the first management control interface (ge4 interface).

The management type of the summary type includes, but is not limited to, a heartbeat detection packet, a query packet, or a statistics packet. For example, the management CPU detects the status of each service CPU through a heartbeat detection packet. When one of the service CPUs is abnormal. The management CPU can restart the service CPU.

If the management CPU determines that the management control packet is a non-aggregated type of management control packet, the management CPU sends the non-aggregated type of management control packet to the switching unit through the first management control interface (ge4 interface), and the switching unit passes And the main service CPU (the first service CPU shown in FIG. 2) is connected to the second management control port (xe2) to send the non-aggregated type of management control message to the main service CPU; the main service CPU passes its own second port ( The logical interface identifier is spiO), and sends a management control response message to the switching unit, so that the switching unit sends the management control response message sent by the primary service CPU to the management CPU through the first management control interface (ge4 interface).

In an optional implementation manner of the present invention, the switching unit receives the message, and if it is determined that the message is an authentication request authentication message, passes through the switching service unit and the main service CPU (the first service CPU shown in FIG. 2) The second management control port (xe2) is connected, and the authentication message is sent to the main service CPU.

When receiving the service packet, the embodiment of the present invention sends a load detection request message to each service CPU through the preset service aggregation trunk port, which is used to obtain the current load ratio of each service CPU and the corresponding management IP address, and report the service. The traffic is transmitted to the service CPU with the lowest load ratio through the service aggregation trunk port, so that the service packets can be sent to the service CPUs in a balanced manner, so that the service CPUs do not need to compete for system resources, which can improve the utilization of system resources. The technical solution of this embodiment can enable multiple CPUs based on asymmetric multi-processing technology to implement load balancing between multiple CPUs.

Further, in the embodiment of the present invention, the packet is sent to the corresponding CPU through the preset port according to different types of the received packets, for example, the management control packet is sent to the management CPU through the first management control port. It is possible to realize unified management and control of each service CPU through the management CPU, thereby improving system reliability, scalability, and catastrophic resistance. For example, when one of the service CPUs fails, the management CPU can restart the service CPU. FIG. 3 is a schematic structural diagram of a switching unit according to another embodiment of the present invention. As shown in FIG. 3, the method includes:

a determining module 31, configured to receive a packet, and determine that the packet is a service packet;

The broadcast module 32 is configured to: after the determining module determines that the packet is a service packet, broadcast a load detection request message by using a service aggregation trunk port, so that each service CPU that receives the load detection request message respectively Returning, by the respective first port, a load detection response message, where the load detection response message includes a current load ratio and a management IP address of the corresponding service CPU, and a sending module 33, configured to return according to each service CPU received by the broadcast module. The current load ratio of the corresponding service CPU and the management IP address included in the load detection response message are sent to the service CPU with the lowest current load ratio.

In an optional implementation manner of the present invention, the sending module 33 is configured to obtain a current current load ratio and a management IP address of the corresponding service CPU included in the load detection response message returned by each service CPU received by the broadcast module. a management IP address corresponding to the service CPU with the smallest load ratio; sending, according to the acquired management IP address, the service packet to the management of the acquisition through the first port of the service CPU having the smallest ratio of the current load The service CPU corresponding to the IP address.

In an optional implementation manner of the present invention, the determining module 31 is further configured to determine that the information is a management control message;

Correspondingly, the sending module 33 is further configured to send, by using the first management control port, the management control packet to the corresponding management CPU, on the basis that the determining module determines that the packet is a management control packet. The management CPU is configured to manage and control the service CPUs.

In an optional implementation manner of the present invention, the determining module 31 is further configured to determine that the authentication file is an authentication request authentication document;

Correspondingly, the sending module 33 is further configured to: after the determining module determines that the packet is an authentication packet, by using the second management control port corresponding to the second port of the main service CPU, The rights message is sent to the main service CPU, and the main service CPU is one of the CPUs of the service CPUs.

It should be noted that, in order to implement the function of the switching unit in this embodiment, a virtual local area network (VLAN) configuration needs to be performed on each physical port of the switching unit and each physical port of each service CPU. Related description in the illustrated embodiment, not Let me repeat.

When the switching unit of the embodiment of the present invention receives the service packet, it can be aggregated by using a preset service.

The Trunk port sends a load detection request message to each service CPU to obtain the current load ratio of the service CPU and the corresponding management IP address, and send the service packet to the service CPU with the smallest load ratio through the service aggregation trunk port. The method for transmitting the service packets to the service CPUs is balanced, so that the service CPUs do not need to compete for the system resources. Therefore, the utilization of the system resources can be improved. Therefore, the technical solution of the embodiment can be based on the asymmetric multi-processing technology. Multiple CPUs can also achieve load balancing between multiple CPUs.

Further, the switching unit of the embodiment of the present invention may send the packet to the corresponding CPU through a preset port according to different types of the received packets, for example, sending the management control packet to the first management control port. Management CPU can realize unified management and control of each service CPU through the management CPU, thereby improving system reliability, scalability, and catastrophic resistance. For example, when one of the service CPUs fails, the management CPU can restart the service CPU.

FIG. 4 is a schematic structural diagram of a switching unit according to another embodiment of the present invention. As shown in FIG. 4, the method includes: a processor, when the processor is running, performing the following steps:

Receiving a packet, and determining that the packet is a service packet;

And transmitting, by the service aggregation trunk port, a load detection request message, so that each service CPU that receives the load detection request message returns a load detection response message by using a first port, where the load detection response message includes a corresponding service CPU. Current load ratio and management IP address;

In the first possible implementation, when the processor is running, it can also execute:

In the second possible implementation, when the processor is running, it can also execute:

Receiving a packet, and determining that the packet is a management control packet; And sending, by the first management control port, the management control packet to the corresponding management CPU, so that the management CPU manages and controls the service CPUs.

In a third possible implementation, when the processor is running, it can also execute:

Receiving a packet, and determining that the packet is an authentication request authentication packet;

The Trunk port sends a load detection request message to each service CPU, which is used to obtain the current load ratio of the service CPU and the corresponding management IP address, and send the service packet to the service CPU with the smallest load ratio through the service aggregation trunk port. The service packet can be sent to the service CPUs in a balanced manner, so that the service CPUs do not need to compete for system resources. Therefore, the utilization of the system resources can be improved. Therefore, the technical solution of the embodiment can be based on the asymmetric multi-processing technology. Multiple CPUs can also achieve load balancing between multiple CPUs.

FIG. 5 is a schematic structural diagram of a multi-CPU packet processing system according to another embodiment of the present invention. As shown in FIG. 5, the system is configured to communicate with the outside of the system, and specifically includes: a service CPU 51 and an exchange unit 52, and services. The number of CPUs 51 is at least two, one of which is a service CPU and the other is a slave CPU;

It should be noted that, assuming that the number of service CPUs is two or more, in this embodiment, one service CPU in each service CPU is used as a main service CPU, and other service CPUs are slave services.

CPU.

In an optional implementation manner of the present invention, the switching unit 52 is connected to each service CPU 51, specifically:

The switching unit 52 of the embodiment provides two interfaces for each service CPU, and one interface (referred to as a service port) is used for receiving the data service layer and the management control layer. Sending a service packet, and another interface (the second management control port) sends and receives a management control packet or an authentication packet.

In this embodiment, the service ports of the switching unit are aggregated together as the service aggregation _trunk port, and are respectively connected to the first port of each service CPU, and only service packets are allowed to pass; Each of the second management control ports of the switching unit is connected to the second port of each service CPU, and only the management control packet or the authentication packet is allowed to pass.

The switching unit 52 is configured to receive the packet, and determine that the packet is a service packet. The service aggregation Trunk port broadcasts a load detection request message, so that each service CP U that receives the load detection request message passes its own The first port returns a load detection response message, where the load detection response message includes a current load ratio of the corresponding service CPU and a management IP address; and a current load ratio and a management IP of the corresponding service CPU included in the load detection response message according to the load Address, the service message is sent to the service CPU with the smallest current load ratio;

Each service CPU 51 is configured to perform corresponding service operations according to the received service message.

It is to be noted that, the switching unit 52 is configured to obtain, according to the current load ratio and the management IP address of the corresponding service CPU included in the load detection response message, a management IP address corresponding to the service CPU with the smallest current load ratio; The obtained management IP address is sent to the service CPU corresponding to the acquired management IP address by using the first port of the service CPU with the smallest current load ratio.

In an optional implementation manner of the present invention, the system further includes: a management CPU 53 connected to the switching unit 52; specifically:

The switching unit 52 of the present embodiment provides a connection between the first management control port and the management CPU 53 in order to implement unified management of the CPUs of the management CPUs.

The switching unit 53 is further configured to receive the packet, and determine that the packet is a management control packet; and send, by using the first management port, the management control packet to the management CPU;

The management CPU 53 is configured to: if it is determined that the configuration control message includes a configuration command, send the configuration command to the switching unit by using the first management control port, so that the switching unit sends the configuration command And sending, by the switching service unit, the configuration command to the other service CPU by using the second management control port corresponding to the second port of the primary service CPU, to send the configuration command to the service CPU. The configuration command includes a logical network port and a service. Configuration parameter

Each service CPU 51 is configured to perform a corresponding configuration operation according to the logical network port and the service configuration parameter included in the configuration command.

It should be noted that the logical network ports and service configuration parameters of the service CPUs are the same.

The management CPU 53 is further configured to send, by the first management control port, the management control packet of the summary type to the switching unit, if the management control packet is determined to be a summary type of the management control packet, And causing, by the switching unit, the management control packet of the summary type to be sent to each service CPU by using the second management control port, where the summary type of the management control packet includes a heartbeat detection packet, a query packet, or a statistics packet. At least one of the items, so that each service CPU returns a management control response message to the management CPU through the switching unit;

The management CPU 53 is further configured to: if the management control packet is determined to be a non-aggregated type of management control packet, send the non-summary type management control packet to the switching unit by using the first management control port. So that the switching unit sends the non-aggregated type of management control packet to the primary service control port through a second management control port corresponding to the second port of the primary service CPU, so that the primary service is The CPU returns a management control response message to the management CPU through the switching unit.

It should be noted that, in this embodiment, the networking configuration, the service configuration, and the running data (such as the session retention table, the address resolution protocol (ARP) table, and the service server state) of the service CPUs are the same. The difference is that the management IP of each service CPU is different.

The management CPU manages the CPUs of each service CPU through the management IP of each service CPU. For example, the management CPU is responsible for providing a unified external interface for each service CPU, and the management CPU manages the configuration, statistics, and query of each service CPU. The management CPU passes the heartbeat detection. The status of each service CPU; and the ability to load and restart each service CPU.

In an optional implementation manner of the present invention, the switching unit 52 is further configured to receive a packet, where the packet is an authentication request authentication packet, and a second corresponding to the second port of the primary service CPU. And managing the control port, and sending the authentication packet to the primary service CPU.

It should be noted that, in this embodiment, a management program can be created on the management CPU, and the management CPU maintains the management IP (IP address in Table 1) of each service CPU through the management program, thereby realizing management and control of each service CPU. Figure 6 is a management CPU in the system embodiment shown in Figure 5. The signaling diagram between the CPU and each service is as shown in Figure 6.

In an optional implementation manner of the present invention, when the management CPU receives the configuration command sent by the system, the configuration command is sent to the main service CPU. In specific implementation, the management CPU sends the configuration command to the first management control port. The switching unit sends the configuration command to the primary service CPU through a second management control port connected to the second port of the primary service CPU.

After the main service CPU processes the configuration command, the real-time synchronization sends the configuration command to each slave CPU and waits for the configuration command response returned from each service CPU. It should be noted that, when the synchronization configuration of the service LBU is unsuccessful, the main service CPU resynchronizes the configuration command to the slave service CPU; after all the service CPUs are configured, the main service CPU returns a configuration command response to the management CPU. In a specific implementation, the main service CPU sends the configuration command response to the switching unit through its second port, and the switching unit sends the configuration command response to the management through the first management control port.

The CPU so that the management CPU can return the received configuration command response to the outside of the system.

It should be noted that, in this embodiment, when the main CPU receives the save configuration command sent by the management CPU, the main CPU may save the configuration command in the memory to, for example, a flash memory, and when the main service CPU restarts, it can read from the flash. Take the configuration command and store the configuration command in memory. When restarting from the service CPU, the main service CPU can be requested to send a configuration command to itself.

In an optional implementation manner of the present invention, when the management CPU 13 receives the summary type of management control message sent by the system, for example, the summary type query command 1, the query command 1 is sent to the main service CPU. The management CPU sends the query command 1 to the switching unit through the first management control port, and the switching unit sends the query command 1 to the main service CPU and the slave service CPU through the second management control port, and the main service CPU and the slave service CPU execute the query. After the command is executed, the response of the query command 1 is returned to the management CPU. In the specific implementation, the primary service CPU and the secondary service CPU send the response of the query command 1 to the switching unit through its second port, and the switching unit queries the command 1 The response is sent to the management CPU through the first management control port, and the management CPU can return the response of the received query command 1 to the outside of the system.

In an optional implementation manner of the present invention, when the management CPU receives the non-summary type management control text sent by the system, for example, the non-summary type query command 2, the query command 2 is sent to the main service CPU, and the specific implementation time is The management CPU sends the query command 2 to the switching unit through the first management control port, and the switching unit sends the query command 2 to the main service CPU through the second management control port corresponding to the second port of the main service CPU, and the main service CPU executes End query command 2 Then, the response of the query command 2 is returned to the management CPU. In specific implementation, the main service CPU sends the response of the query command 2 to the switching unit through its second port, and the switching unit passes the response of the query command 2 through the first management control port. Sended to the management CPU, the management CPU can return the response of the received query command 2 to the outside of the system.

When the system receives the service packet, the system can be aggregated by using a preset service.

Another embodiment of the present invention provides a board, including a multi-CPU packet processing system according to the embodiment shown in FIG. 5. The specific implementation principle and technical effects are as shown in the related embodiment in the embodiment shown in FIG. 5.

A person skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the system, the device and the unit described above can refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.

In the several embodiments provided by the present application, it should be understood that the disclosed system, device and The method can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

The above-described integrated unit implemented in the form of a software functional unit can be stored in a computer readable storage medium. The above software functional units are stored in a storage medium and include a number of instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to perform some of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a USB flash drive, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes. Medium.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will understand that The technical solutions described in the foregoing embodiments are modified, or some of the technical features are equivalently replaced. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

claims

1. A multi-CPU message processing method, characterized by including:

The switching unit receives the message and determines that the message is a service message;

The load detection request message is broadcast through the service aggregation Trunk port, so that each service CPU that receives the load detection request message returns a load detection response message through their respective first port, and the load detection response message includes the corresponding service CPU's Current load ratio and management IP address;

According to the current load ratio and management IP address of the corresponding service CPU contained in each load detection response message, the service message is sent to the service CPU with the smallest current load ratio.

2. The method according to claim 1, characterized in that, according to the current load ratio and management IP address of the corresponding service CPU contained in each load detection response message, the service message is sent to the server with the smallest current load ratio. The business CPU specifically includes:

According to the current load ratio and management IP address of the corresponding service CPU contained in each load detection response message, obtain the management IP address corresponding to the service CPU with the smallest current load ratio; according to the obtained management IP address, through the The first port of the service CPU with the smallest current load ratio sends the service message to the service corresponding to the obtained management IP address.

CPU.

3. The method according to claim 1, further comprising:

Receive the message and determine that the message is a management control message;

The management control message is sent to the management CPU corresponding to the management port through the first management control port, so that the management CPU manages and controls each service CPU.

4. The method according to claim 1, further comprising:

Receive the message and determine that the message is an authentication message;

The authentication message is sent to the main service CPU through the second management control port corresponding to the second port of the main service CPU, and the main service CPU is one of the CPUs of the service CPUs.

5. A switching unit, characterized by including:

Determining module, used to receive the message and determine that the message is a service message;

A broadcast module, configured to broadcast a load detection request message through the service aggregation Trunk port on the basis that the determination module determines that the message is a service message, so that the load detection request can be received. Each service CPU requesting the message returns a load detection response message through its respective first port. The load detection response message includes the current load ratio and management IP address of the corresponding service CPU; a sending module, used to receive according to the broadcast module The load detection response message returned by each service CPU contains the current load ratio and management IP address of the corresponding service CPU, and the service message is sent to the service CPU with the smallest current load ratio.

6. The switching unit according to claim 5, characterized in that the sending module is specifically configured to determine the current load ratio and the load detection response of the corresponding service CPU contained in the load detection response message returned by each service CPU received by the broadcast module. Management IP address, obtain the management IP address corresponding to the service CPU with the smallest current load ratio; according to the obtained management IP address, send the service message to the first port of the service CPU with the smallest current load ratio. The service CPU corresponding to the obtained management IP address.

7. The switching unit according to claim 5, characterized in that the determining module is also used to determine that the message is a management control message;

The sending module is also configured to send the management control message to the corresponding management CPU through the first management control port based on the determination module determining that the message is a management control message, so that the The management CPU manages and controls each service CPU.

8. The switching unit according to claim 5, characterized in that the determination module is also used to determine that the message is an authentication request authentication message;

The sending module is also configured to send the authentication message through the second management control port corresponding to the second port of the main service CPU after the determining module determines that the message is an authentication message. The message is sent to the main service CPU, and the main service CPU is one of the CPUs of the service CPUs.

9. A switching unit, characterized in that it includes: a processor, the processor is specifically configured to: receive a message and determine that the message is a service message;

10. The switching unit according to claim 9, the processor is further configured to: obtain the minimum current load ratio based on the current load ratio and management IP address of the corresponding service CPU contained in each load detection response message. The management IP address corresponding to the service CPU;

According to the obtained management IP address, the service packet is sent to the service CPU corresponding to the obtained management IP address through the first port of the service CPU with the smallest current load ratio.

11. The switching unit according to claim 9, the processor is further configured to: receive a message and determine that the message is a management control message;

The management control message is sent to the corresponding management CPU through the first management control port, so that the management CPU manages and controls each service CPU.

12. The switching unit according to claim 9, the processor is further configured to: receive a message and determine that the message is an authentication request authentication message;

13. A multi-CPU message processing system, used to communicate with the outside of the system, characterized in that it includes: each service CPU and a switching unit, each of the service CPUs includes at least two service CPUs, and each of the service CPUs includes at least two service CPUs. One of the service CPUs is the main service CPU;

The switching unit is connected to each service CPU respectively, specifically: the service aggregation Trunk port of the switching unit is connected to the first port of each service CPU, and each second management control port of the switching unit is respectively Connect to the second port of each service CPU;

The switching unit is configured to receive a message and determine that the message is a service message; broadcast a load detection request message through the service aggregation Trunk port, so that each service CPU receiving the load detection request message passes their respective The first port returns a load detection response message, which includes the current load ratio and management IP address of the corresponding service CPU; according to the current load ratio and management IP address of the corresponding service CPU included in each load detection response message. Address, send the service message to the service CPU with the smallest current load ratio;

Each of the service CPUs is used to perform corresponding service operations according to the received service messages.

14. The system according to claim 13, characterized in that, the switching unit, specifically For obtaining the management IP address corresponding to the service CPU with the smallest current load ratio according to the current load ratio and management IP address of the corresponding service CPU contained in each load detection response message; according to the obtained management IP address, through The first port of the service CPU with the smallest current load ratio sends the service packet to the service CPU corresponding to the obtained management IP address.

15. The system according to claim 13, further comprising: a management CPU connected to the switching unit through a first management control port;

The switching unit is also configured to receive a message and determine that the message is a management control message; and send the management control message to the management CPU through the first management port;

The management CPU is configured to, if it is determined that the management control message includes a configuration command, send the configuration command to the switching unit through the first management control port, so that the switching unit will The configuration command is sent to the main service CPU through the second management control port corresponding to the second port of the main service CPU, so that the main service CPU synchronously sends the configuration command to other services through the switching unit CPU, the configuration command includes logical network port and service configuration parameters;

Each of the service CPUs is used to perform corresponding configuration operations according to the logical network port and service configuration parameters included in the configuration command;

The logical network ports and service configuration parameters configured by each service CPU are the same;

The management CPU is also configured to send the summary type management control message to the switching unit through the first management control port if it is determined that the management control message is a summary type management control message. , so that the switching unit sends the summary type management control message to each service CPU respectively through the second management control port, and the summary type management control message includes a heartbeat detection message or a query message or statistics At least one item in the message, so that each service CPU returns a management control response message to the management CPU through the switching unit;

The management CPU is also configured to, if it is determined that the management control message is a non-summary type management control message, send the non-summary type management control message to the switch through the first management control port. unit, so that the switching unit sends the non-summary type management control message to the main service CPU through the second management control port corresponding to the second port of the main service CPU, so that the main service CPU The service CPU returns a management control response message to the management CPU through the switching unit.

16. The system according to claim 13, characterized in that the switching unit is also used to receive a message and determine that the message is an authentication request authentication message; through the second communication with the main service CPU The second management control port corresponding to the port sends the authentication message to the main service CPU.

17. A single board, characterized in that the single board includes: at least one multi-CPU text processing system according to any one of claims 13-17.