WO2009055982A1

WO2009055982A1 - An atca blade server multi-frame concatenation system

Info

Publication number: WO2009055982A1
Application number: PCT/CN2007/003694
Authority: WO
Inventors: Juwei Bian; Xiankui Xiong
Original assignee: Zte Corporation
Priority date: 2007-10-31
Filing date: 2007-12-19
Publication date: 2009-05-07
Also published as: CN101150413B; CN101150413A

Abstract

An ATCA blade server main frame and an ATCA blade server sub-frame whose exchanging node boards are connected with a switch are provided. The blade server main frame is composed of a load balancing node board, which is connected with the switch for managing the load balancing of the blade node board of the main frame and sub-frame. The load balancing node board provides a uniform medium access control addresses and a virtual Internet protocol addresses.

Description

ATCA blade server multi-frame cascading system

Technical field

The present invention relates to the field of communication technologies, and in particular, to an ATCA (Advanced Telecommunication Computing Architecture) blade multi-frame cascading system.

Background technique

The ATCA blade server is a frame-based, high-performance blade server built using the ATCA (Advanced Telecommunication Computing Architecture) specification. Among them, the blade server or blade server is a HAHD (High Availability High Density) low-cost server platform designed for special application industries and high-density computer environments. The main structure of the blade server is a large main body frame, and a plurality of inserts of a shape such as a blade can be inserted inside, and thus the card is also simply referred to as a blade, wherein each blade is actually a system motherboard, similar to a A separate server that can boot its own operating system from a local hard drive. Each blade can run its own system, serving a different set of users, without correlation. ATCA is a standard architecture system specification for next-generation carrier-grade communication equipment developed and promulgated by PICMG (Peripheral Component Interconnect Industrial Computer Manufacturers Group), which was evolved from the previous PICMG 2.X. Including, the core specification PICMG3.0 and a series of auxiliary specifications PICMG3.X, where the core specification defines the mechanical design, system management, power distribution, heat dissipation, backplane interconnection, etc. in the ATCA series of specifications; The transmission methods that are interconnected in the core specification are defined. The ATCA specification represents a fully open, modular industry standard, which enables telecom equipment manufacturers to use interoperable, mature commercial hardware and software components from third-party vendors.

The ATCA blade server chassis typically includes multiple switch node boards and multiple high-performance blade node boards. The switch node board supports Layer 2 switching, and the blade node board is used to implement high-density calculation. It can provide up to 12 blade gusset boards in the existing ATCA blade server box. When the computing power of the 12 blade node boards cannot meet the specific application requirements, more cascades are needed. The ATCA blade server chassis is constructed as an ATCA blade server multi-chassis cascade system. At present, there is no related ATCA blade server multi-frame cascading system construction scheme. In summary, the technical solutions of the existing ATCA blade server multi-frame cascading system are obviously inconvenient and defective in practical use, so it is necessary to improve.

Summary of the invention

The technical problem to be solved by the present invention is to provide an ATCA blade server multi-chassis-level connection system, which can implement cascading of multiple ATCA blade server chassis.

In order to solve the above technical problem, the present invention provides an ATCA blade server multi-chassis cascading system, each ATCA blade server chassis includes a plurality of switching node boards and blade node boards, and the cascading system includes the ATCA blade server Box and switch;

A switch node board of each of the ATCA blade server chassis is interconnected with the switch to implement an ATCA blade server multi-chassis cascade.

Further, the cascading system may further have the following features: the ATCA blade server chassis is configured as an ATCA blade server host frame and an ATCA blade server chassis, and the ATCA blade server host frame further includes:

The load balancing node board is interconnected with the switch for performing load balancing processing on the ATCA blade server host frame and the blade node board of the ATCA blade server sub-frame, the load balancing process including the ATCA blade server host frame and the ATCA blade server The blade node board of the chassis is shielded, scheduled, and health monitored.

Further, the foregoing cascading system may further have the following features: the load balancing node board provides a unified medium access control address and a virtual internet protocol address for external users to access the multi-frame cascading system.

Further, the foregoing cascading system may further have the following features: each blade node board of the ATCA blade server host frame and the ATCA blade server sub-frame has a private Internet Protocol address, and the Internet Protocol address is only a load balancing node. The board is visible to implement a load balancing node board to shield the ATCA blade server host frame and the blade node board of the ATCA blade server secondary chassis. Further, the foregoing cascading system may further have the following features: the load balancing node board includes: a first load balancing node board, and a main load balancing node board, configured to be in a normal working state

The ATCA blade server host frame and the blade node board of the ATCA blade server subrack are loaded and balanced, and are switched to the second load balancing node board when the operation is abnormal;

The second load balancing node board is a standby load balancing node board, and is configured to load balance the ATCA blade server host frame and the blade node board of the ATCA blade server sub-frame after the first load balancing node board is switched.

Further, the foregoing cascading system may further have the following feature: the first negative-cut equalization node board is switched to the second load balancing node board by using a hardware main-to-change signal when the operation is abnormal, and the

The Ethernet interface between the first load balancing node board and the second load balancing node board performs load balancing data synchronization with the second load balancing node board; and the second load balancing node board is switched over the first load balancing node board. After that, the address resolution protocol advertisement is sent out.

Further, the cascading system may further have the following feature: the Ethernet interface between the first load balancing node board and the second load balancing node board is a heartbeat Ethernet interface.

Further, the foregoing cascading system may also have the following features: the switch is a layer 3 switch or a layer 3 or more switch.

Further, the foregoing cascading system may further have the following features: the switch further includes: a first switch and a second switch; the first switch and the second switch work in an active/standby mode; the first switch and the first switch The two switches perform port aggregation connection through a plurality of Ethernet interfaces; the first switch and the second switch use the virtual route redundancy protocol to implement Internet protocol layer redundancy.

Further, the above cascade system may also have the following features.

The first switch is connected to the first load balancing node board one by one, and the second switch is connected to the second load balancing node board one by one; and/or the first switch and the second load balancing node board are connected a connection and the second switch

>5 Connect to the first load balancing node board one by one.

Further, the cascading system may further have the following features: the switching node board of the ATCA blade server host frame and the ATCA blade server sub-frame further includes: a first switching node board and a second switching node board; Switch node board and second switch node board and first exchange Cross-connecting between the machine and the second switch; the first switch node board and the second switch node board work in a dual active mode.

In the ATCA blade server multi-frame cascading system of the present invention, the ATCA blade server multi-chassis is set to the ATCA blade server host frame and the ATCA blade server sub-frame, and the ATCA blade server host frame and the ATCA blade server sub-frame are exchanged. The way in which the LUN board is interconnected with the first and second Layer 3 switches, thereby implementing cascading of multiple ATCA blade server chassis. And the present invention sets the first and second load balancing node boards in the ATCA blade server host frame, the first and second load balancing node boards and the first and second Layer 3 switches are in the active/standby mode, and the ATCA blade server host box The switching node board of the secondary chassis of the ATCA blade server is in dual-master mode. Each blade in the frame and the switching node board of the frame are cross-connected through GE. The dual-NIC is used to bind the same IP to ensure that the ATCA blade server is multi-chassis. The high reliability and security of the connected system.

BRIEF abstract

1 is a schematic structural diagram of a multi-chassis cascading system of an ATCA blade server provided by the present invention; FIG. 2 is a schematic structural diagram of a multi-chassis cascading system of a part of an ATCA blade server provided by the present invention;

3 is a flow chart of a multi-chassis cascading method for an ATCA blade server provided by the present invention. FIG. 4 is a flow chart of a multi-chassis cascading method for an ATCA blade server according to a preferred embodiment of the present invention.

Preferred embodiment of the invention

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The basic idea of the present invention is: Introducing a switch and a load balancing node board in an ATCA blade server multi-chassis cascading system, implementing cascading of multiple ATCA blade server chassis and ensuring high ATCA blade server multi-chassis cascading system Reliability and security. The ATCA blade server multi-chassis cascading system 100 provided by the present invention, as shown in FIG. 1, includes a switch 101, an ATCA blade server host frame 102, and an ATCA blade server sub-frame 103. among them,

Switch 101 is interconnected with ATCA blade server host block 102 and ATCA blade server slave frame 103. Preferably, the switch 101 is a Layer 3 switch or a switch of more than three layers. The Layer 3 switch refers to a switch with a part of the router function. The performance is focused on the Layer 2 and Layer 3 switching. By using an Application Specific Integrated Circuit (ASIC), IP packets can be routed once and Forwarded multiple times. The switch of the present invention may also be a switch of three or more layers, that is, a four-layer switch, a seven-layer switch, etc., within the scope of protection of the present invention, as long as the functions of the three-layer switch can be implemented.

In order to improve the reliability of the multi-chassis cascade system 100 and eliminate single point of failure, the switch 101 further includes a first switch 1011 and at least one second switch 1012, the first switch 1011 and the second switch 1012 working in the main In standby mode. The first switch 1011 is a master switch, which communicates with the ATCA blade server host block 102 and the ATCA blade server slave frame 103 during normal operation, and switches to the standby switch, the second switch 1012, when the operation is abnormal. When there are a plurality of second switches 1012, a preference may be set on the plurality of second switches 1012, and the first switch 1011 is switched to the appropriate second switch 1012 according to the level of the preference level when the operation is abnormal.

The first switch 1011 and the second switch 1012 implement VRRP (Virtual Protocol Redundancy Protocol) to implement IP (Internet Protocol) layer redundancy, and Layer 2 switching is dual-active mode. The first switch 1011 and the second switch 1012 perform a trunk (port aggregation) connection through multiple Ethernet interfaces to increase the capacity of the physical link, so as to meet the bandwidth increase requirement and implement interface redundancy. , improve the reliability of the system 100. The Ethernet interface is divided into a ten-mega Ethernet interface, a one-hundred-megabit Ethernet interface, and a GE (Giga Ethernet) interface. Based on the data transmission rate, the first switch 1011 and the second switch 1012 of the present invention preferably perform a trunk port connection through multiple GE interfaces. However, the Trunk port connection through other Ethernet interfaces is also within the scope of the present invention. For convenience, the Ethernet interface referred to below is preferably a GE interface.

ATCA blade server host box 102, completely independent of ATCA blade server subrack 103 The ATCA blade server subrack 103 includes a switch node board 105 and a blade node board 106, but does not include the load balancing node board 104, and the ATCA blade server host box 102 includes a switch node board 105 and a blade node board. 106 and load balancing node board 104. The load balancing node board 104 is interconnected with the switch 101 through multiple GE interfaces, and is used for load balancing processing on the ATCA blade server host block 102 and the blade node board 106 of the ATCA blade server sub-frame 103. The load balancing process includes: screening, scheduling, and health monitoring the ATCA blade server host block 102 and the blade node board 106 of the ATCA blade server subrack 103.

The load balancing node board 104 shields, schedules, and monitors the health of the blade node board 106, thereby fully utilizing the computing power of the blade cluster and facilitating the management and expansion of the blade node board 106.

Specifically, the load balancing node board 104 provides a unified MAC (Media Access Control) address and a Virtual IP (Virtual Internet Protocol) address to the network other than the switch and the switch, where The MAC address and VIP are used by external users to access the multi-chassis cascade system.

In addition, each blade node board of the ATCA blade server host block 102 and the ATCA blade server slave frame 103 has a private IP (Internet Protocol) address, which is visible only to the load balancing node board 104. Therefore, the load balancing node board 104 can shield the ATCA blade server host block 102 and the ATCA blade server sub-frame 103 from the blade node board 106, thereby ensuring high security of the system 100. Further, the present invention can integrate the firewall function on the load balancing node board 104, thereby improving the security of the entire server system 100.

The load balancing node board 104 can periodically perform health monitoring on the blade node board 102 of the ATCA blade server host box 102 and the ATCA blade server sub-frame 103. When the faulty blade node board is detected, the faulty blade node board is The box is removed. The load balancing node board 104 also supports a session hold function to ensure consistency of the associated session processing by each blade node board 106. In order to improve the reliability of the multi-chassis cascade system 100 and eliminate single point of failure, the load balancing node board 104 further includes: a first load balancing node board 1041 and a second load balancing node board 1042. among them,

The first load balancing node board 1041 is a load balancing node board for load balancing processing of the ATCA blade server host frame 102 and the blade node board 106 of the ATCA blade server secondary chassis 103 when it is working normally, and is working. When abnormal, it is switched to the second load balancing node board 1042.

The second load balancing node board 1042 is a standby load balancing node board, and is configured to perform the ATCA blade server host block 102 and the blade node board 106 of the ATCA blade server sub-frame 103 after the first load balancing node board 1041 is switched. Load balancing processing.

In the present invention, the first load balancing node board 1041 and the second load balancing node board 1042 have a heartbeat Ethernet interface and a hardware main standby line to implement the first load balancing node board 1041 and the second load balancing node board. 1042 fast switching and data synchronization, as shown in Figure 2. "Heartbeat" is a communication term, here used to determine if there is an available Ethernet interface between the first load balancing node board 1041 and the second load balancing node board 1042. The heartbeat Ethernet interface in the present invention is preferably a heartbeat GE interface. The hardware master and backup lines are used to provide hardware master/slave switchover signals. The first load balancing node board 1041 is switched to the second load balancing node board 1042 by using the hardware active/standby switching signal when the operation is abnormal, and the load balancing data synchronization is performed by the heartbeat GE interface and the second load balancing node board 1042; the second load balancing is performed. After the first load balancing node board 1041 is switched, the node board 1042 sends an ARP (Address Resolution Protocol) advertisement to the outside. After receiving the notification, the switch 101 updates its own exchange table and forwards the system data to the second. The load balancing node board 1042 is then load balanced by the second load balancing node board 1042 for the ATCA blade server host frame and the blade node board of the ATCA blade server secondary chassis.

As shown in FIG. 1 and FIG. 2, the first load balancing node board 1041 is connected to the first switch 1011 one by one, and the second load balancing node board 1042 is connected to the second switch 1012. The physical connection manner is Multiple GE interfaces are connected to the trunk port. The first switch 1011 and the second load balancing node board 1042 are connected to each other, and the second switch 1012 is connected to the first load balancing node board 1041. The physical connection mode is performed through multiple GE interfaces. The trunk port is connected, which is also within the scope of the protection of the present invention.

The switch node board 105 is present in both the ATCA blade server host block 102 and the ATCA blade server slave frame 103. In order to improve the reliability of the multi-chassis cascade system 100 and eliminate single points of failure, The switching node board 105 includes a first switching node board 1051 and a second switching node board 1052, and the first switching node board 1051 and the second switching node board 1052 operate in a dual active mode. In the present invention, the first switching node board 1051 and the second switching node board 1052 support Layer 2 Ethernet switching and STP (Spanning Tree Protocol) to eliminate cross-connection between the switch 101 and the switch 101. The Ethernet loop achieves redundancy at the same time. Preferably, the first switching node board 1051 and the second switching node board 1052 support Layer 2 Ethernet switching and Rapid Spanning Tree Protocol (RSTP) to shorten the convergence time of the bridge network.

Referring to FIG. 1 and FIG. 2, the first switching node board 1051 and the second switching node board 1052 are interconnected with the first switch 1011 and the second switch 1012, and the physical connection manner is to perform a trunk port through multiple GE interfaces. connection.

Similarly, the ATCA blade server host block 102 and the ATCA blade server slave frame 103 each have a plurality of blade node boards 106, and the plurality of blade node boards 106 pass through the GE interface with the first switch node board 1051 and the second switch node board. 1052 is cross-connected. In the present invention, a primary and backup dual network card may be disposed on the plurality of blade node boards 106, and the same IP is bound to the dual network card, and the network port driver uses physical link connection health detection and ARP health detection mode for the primary use. The NIC performs the detection. If the primary NIC is abnormal, the active/standby switchover is performed. After the primary NIC is switched to the standby NIC, the ARP advertisement is sent out.

FIG. 3 is a flowchart of a multi-chassis cascading method for an ATCA blade server according to the present invention. The method is used to construct an ATCA blade server multi-chassis cascading system 100 as shown in FIG. 1, wherein each method specifically includes the following:

Step S301, the ATCA blade server multi-chassis frame is respectively set to an ATCA blade server host frame 102 and an ATCA blade server sub-frame 103.

The method further includes: inserting a load balancing node board 104 on the ATCA blade server host block 102 to load balance the blade node board 106 of the ATCA blade server host block 102 and the ATCA blade server chassis 103.

In step S302, the switch node board 105 of the ATCA blade server host block 102 and the ATCA blade server slave frame 103 are interconnected with the switch 101. In this step, the switch node board 105 and the switch 101 are interconnected, and the physical connection mode is connected through the trunk port of the GE interface.

The ATCA blade server multi-chassis cascading method provided by the preferred embodiment of the present invention is shown in FIG. 4, which is used to construct the ATCA blade server multi-chassis cascading system 100 as shown in FIG. 1 and FIG. 2, wherein each The ATCA blade server chassis includes a switch node board 105 and an N-block blade node board 106. The method specifically includes the following:

Step S401, an ATCA blade server host box 102 is set in the ATCA blade server multi-machine box, and the other chassis is set to the ATCA blade server sub-frame 103.

Step S402, inserting a first load balancing node board 1041 and a second load balancing node board 1042 on the ATCA blade server host block 102.

Step S403, a GE heartbeat interface and a hardware active/standby switchover line are set between the first load balancing node board 1041 and the second load balancing node board 1042.

Step S404, the switching node board 105 and the N-block blade node board 106 are interconnected on the ATCA blade server host box 102 and the ATCA blade server sub-frame 103. The physical connection mode is to connect the trunk port through the GE interface.

Step S405, a primary and backup dual network card is set on the switching node board 105 of the ATCA blade server host box 102 and the ATCA blade server secondary box 103, and the dual network card is bound to the same IP.

The network port driver uses the physical link connection health detection and ARP health detection mode to detect the primary network card. If the primary network adapter is abnormal, the active/standby switchover is performed. After the primary network adapter switches to the standby network adapter, the ARP advertisement is sent out.

Step S406, the first switch node board 1051 and the second switch node board 1052 are respectively set up, and the first switch node board 1051 and the second switch node board 105 are connected to the switch node board 105 of the ATCA blade server chassis 102 and the ATCA blade server slave frame 103, respectively. The working mode of the switching node board 1052 is set to the dual active mode. Both the first switching node board 1051 and the second switching node board 1052 support Layer 2 Ethernet switching and RSTP protocols. Step S407, the first switch 1011 and the second switch 1012 are set up and cross-connected with the first switching node board 1051 and the second switching node board 1052. The physical connection mode is The GE interface is connected to the trunk port.

Step S408, the first switch 1011 and the second switch 1012 are set to work in the primary standby mode, and the first switch 1011 is set as the primary switch, and the second switch 1012 is set as the standby switch.

5 The VRRP protocol is used between the first switch 1011 and the second switch 1012 to implement IP layer redundancy, and the Layer 2 switching works in dual active mode.

Step S409, the first load balancing node board 1041 and the second load balancing node board 1042 are set to operate in the primary standby mode, and the first load balancing node board 1041 is set as the primary load balancing node board, and the second load balancing node is configured. The board 1042 is set as a standby load balancing node board.

10 The load balancing node board 104 provides a unified MAC address and a VIP address. The primary load balancing node board monitors the connection status between the node board and the switch 101 through a protocol and a physical link detection mechanism. If there is an abnormality, the hardware master The standby signal is switched and the data is synchronized through the heartbeat GE interface. After the switchover is complete, the new active load balancing node board sends ARP advertisements.

In step S410, the first load balancing node board 1041 and the first switch 1011 are connected, and the L5 is connected to the second load balancing node board 1042 and the second switch 1012.

The physical connection mode of the first load balancing node board 1041 and the first switch 1011 and the physical connection mode of the second load balancing node board 1042 and the second switch 1012 are connected through a GE interface.

When the ATCA blade server of the ATCA blade server in FIG. 1 and FIG. 2 is constructed according to the cascading method shown in FIG. 4, the data processing flow of the cascading system 100 is as follows:

When the first load balancing node board 1041 and the first switch 1011 are working normally, the external user accesses the ATCA blade server multi-chassis cascade system 100 through the VIP address provided by the load balancing node board, and the data stream is exchanged to the first switch 1011. The first load balancing node board 1041 in the host frame. The first load balancing node board 1041 detects the content of the packet 15 according to different load balancing functions configured, and according to the load status and health status of each blade node board 106 of the current ATCA blade server host frame 102 and the ATCA blade server secondary chassis 103. And the services that can be provided, according to a certain scheduling policy, scheduling the data packets at the fourth layer or the application layer, and uniformly forwarding the external connection requests (or forwarding according to the corresponding scheduling algorithm) to the blade node boards 106, thereby Da For the purpose of load balancing, the response from the server (that is, the blade node board in the ATCA box) is forwarded to the first switch 1011 in the same manner (for forwarding by the load balancing node board, using NAT mode), and the first load balancing. The node board 1041 also needs to support session hold functionality to ensure consistency of the blade node board 106 for associated session processing. The first load balancing node board 1041 also periodically performs health check on each blade node board 106 to remove the faulty blade from the blade group.

When the first load balancing node board 1041 works abnormally, the first load balancing node board 1041 is switched to the second load balancing node board 1042, and then the external user accesses the ATCA blade server multi-chassis level through the VIP address provided by the load balancing node board. In the system 100, the data flows through the first switch 1011, through the Layer 3 switching (ie, IP switching) to the second switch 1012, and the second switch 1012 performs Layer 2 switching to the second load balancing node board 1042 in the host frame. The specific data processing process is similar to the data processing process of the first load balancing node board 1041 and the first switch 1011 in the normal working state, which is limited by the space, and is not expanded here. For details, refer to the first load balancing node board 1041 and the first A data processing process in which a switch 1011 is operating normally.

When the first switch 1011 is abnormal, the first switch 1011 is switched to the second switch.

1012, the external user accesses the ATCA blade server multi-chassis cascading system 100 through the VIP address, and the data flow passes through the first switch 1011 (when the first switch is abnormal, the switch's Layer 2 switching is still normal), and is switched through the second layer to The second switch 1012, 1012 performs the three-layer switching to the first switch 1011, and the first switch 1011 exchanges data to the first load balancing node board 1041 in the host frame through the Layer 2 switching. The specific data processing process is the same as the data processing process of the first load balancing node board 1041 and the first switch 1011 in the normal working state, and is limited by the space, and is not expanded here. For details, refer to the first load balancing node board 1041 and The data processing process of the first switch 1011 when it is working normally.

There are a variety of other embodiments of the present invention, and various modifications and changes can be made thereto in accordance with the present invention without departing from the spirit and scope of the invention. Changes and modifications are intended to be included within the scope of the appended claims.

Industrial applicability In the ATCA blade server multi-frame cascading system of the present invention, the ATCA blade server multi-chassis is set to the ATCA blade server host frame and the ATCA blade server sub-frame, and the ATCA blade server host frame and the ATCA blade server sub-frame are exchanged. The cascading board is interconnected with the first and second Layer 3 switches, thereby implementing the ATCA blade server multi-chassis cascading. And the present invention sets the first and second load balancing node boards in the ATCA blade server host frame, the first and second load balancing node boards and the first and second layer 3 switches are in the active/standby mode, and the ATCA blade server host box The switching node board of the ATCA blade server subrack is in dual-active mode. Each blade in the frame and the local switching node board are cross-connected through GE. The dual-NIC is used to bind the same IP, thus implementing the ATCA blade server multi-machine. The flattening of the frame cascading system ensures the high reliability and security of the ATCA blade server multi-chassis cascading system.

Claims

Claim

An ATCA blade server multi-frame cascading system, each ATCA blade server chassis includes a plurality of switch node boards and blade node boards, wherein the cascading system includes the ATCA blade server chassis and switches;

A switch node board of each of the ATCA blade chassis is interconnected with the switch to implement an ATCA blade server multi-chassis cascade.

2. The cascading system according to claim 1, wherein the ATCA blade server multi-chassis is respectively set to an ATCA blade server host frame, an ATCA blade server sub-frame, and the ATCA blade server host frame is further configured. Includes:

Load balancing node board, interconnected with the switch, used to host the ATCA blade server and

The blade node board of the ATCA blade server subrack is subjected to load balancing processing, which includes shielding, scheduling, and health monitoring of the ATCA blade server host frame and the blade node board of the ATCA blade server subrack.

The cascading system according to claim 2, wherein the load balancing node board provides a unified interface access control address and a virtual internet protocol address for external users to access the multi-chassis cascade system. .

4. The cascading system according to claim 2, wherein each of the ATCA blade server host frame and the ATCA blade server sub-frame has a private internet protocol address, and the internet protocol address is only The load balancing node board is visible to implement a load balancing node board to shield the ATCA blade server host frame and the blade node board of the ATCA blade server secondary frame.

The cascading system according to claim 2, wherein the load balancing node board comprises:

The first load balancing node board is used as a load balancing node board for load balancing of the blade node boards of the ATCA blade server host frame and the ATCA blade server secondary chassis when it is working normally, and is switched when the operation is abnormal. To the second load balancing node board;

The second load balancing node board is a standby load balancing node board, and is used for the ATCA blade server host frame and the ATCA blade server secondary chassis after the first load balancing node board is switched. The blade node board performs load balancing processing.

The cascading system according to claim 5, wherein the first load balancing node board is switched to the second load balancing node board by the hardware active/standby switching signal when the operation is abnormal, and the first load balancing is performed at the same time. Ethernet interface between the node board and the second load balancing node board

The second load balancing node board performs load balancing data synchronization; after the first load balancing node board is switched, the second load balancing node board sends an address resolution protocol advertisement to the outside.

The cascading system according to claim 6, wherein the Ethernet interface between the first load balancing node board and the second load balancing node board is a heartbeat Ethernet interface.

The cascading system according to any one of claims 1 to 7, wherein the exchange is a three-layer switch or a three-layer switch.

The cascading system according to claim 8, wherein the switch further comprises: a first switch and a second switch; the first switch and the second switch work in an active/standby mode; A switch and a second switch perform port aggregation connection through a plurality of Ethernet interfaces; the first switch and the second switch implement a Internet protocol by using a virtual routing redundancy protocol

15 layers of redundancy.

10. The cascade system of claim 9 wherein:

The first switch is connected to the first load balancing node board one by one, and the second switch is connected to the second load balancing node board one by one; and/or the first switch and the second load balancing node board are one by one Connected, and the second switch >0 is connected to the first load balancing node board one by one.

The cascading system according to claim 9 or 10, wherein the switching node board of the ATCA blade server host frame and the ATCA blade server sub-frame further comprises: a first switching node board and a second switching node a first switching node board and a second switching node board are cross-connected with the first switch and the second switch; the first switching node board and the second switching

>5 node boards work in dual active mode.