Networking method of data center network and data center network

By adopting the CLOS network structure and deploying odd-numbered switches and even-numbered servers in the data center network, combined with numbering rules and connection strategies, a non-blocking three-dimensional ring network is formed, which solves the scalability and efficiency problems of traditional data center networks and achieves efficient data transmission and bandwidth utilization.

CN120692166APending Publication Date: 2025-09-23HAINAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410315674.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-03-20
Publication Date
2025-09-23

AI Technical Summary

Technical Problem

Traditional data center network topology construction methods lack flexibility and scalability, resulting in low data transmission efficiency and insufficient bandwidth, making it difficult to meet the needs of high performance and high cost-effectiveness.

Method used

A data center network networking method based on the CLOS network structure is adopted. By deploying odd-numbered switches and even-numbered servers, combined with numbering rules and connection strategies, a non-blocking three-dimensional ring network structure is formed, making full use of switch ports to achieve non-blocking information transmission.

Benefits of technology

It improves data transmission efficiency and network bandwidth, realizes the scalability and fault tolerance of data center networks, maximizes switch utilization, and forms an efficient networking structure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120692166A_ABST
    Figure CN120692166A_ABST
Patent Text Reader

Abstract

The invention provides a networking method of a data center network and the data center network, which are used for constructing a high-performance data center network. In logic, the proposed architecture is constructed by inserting a dual port server in each pair of adjacent switches in a switch fabric, where the switches are connected in the form of a ring CLOS fabric. The data center network architecture is mainly composed of two devices including port switches and dual-port servers, the servers and the switches are respectively deployed in columns (odd numbers), the number of the n-port switches in each switch column is n, the number of the dual-port servers in each server column is n, the number of the n-port switches in each server column is n, and the number of the n-port servers in each server column is n. And the server columns and the switch columns are alternately arranged in a circular topology form. According to the annular data center network, networking of the data center network which is large in network bandwidth, balanced in load, good in fault tolerance and capable of achieving non-blocking transmission is achieved with low cost.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of network technology, and particularly relates to a data center network networking method and a data center network. Background Art

[0002] With the rapid development of information technology such as computers and communications, and the surge in applications such as electronic media, cloud computing, and big data, the demand for data exchange is increasing, placing higher demands on data center networks. Due to the rapid growth in the number of users and service content, current data center networks need to continuously expand to meet service demands. Traditional data center network topology construction methods and the operational mechanisms of the network layer control plane are rigid and can no longer meet the growing demand for high performance and cost-effectiveness in the new environment.

[0003] Current data center networks usually adopt the classic fat-tree structure, which lacks design flexibility and high efficiency in expansion. The present invention is based on the CLOS network structure. The basic form of the data center network is formed by folding an ordinary CLOS network and connecting it into a ring, that is, connecting the switches in the input and output stages of the CLOS network. Therefore, a dual-port server can be inserted between any two switches at different levels. This structure has two effective expansion methods: quantity expansion and level expansion, so it has good scalability. The connection between the switches is based on the CLOS network structure, which can realize non-blocking switching function. A large number of the above-mentioned switches are connected to the server to form a data center network with good scalability and non-blocking. By adopting reasonable identification, each server can be numbered so that it has a unique location coordinate. Under this premise, the present invention provides an efficient routing method. The networking method of the present invention has good scalability and strong fault tolerance, and can efficiently realize data center network networking. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems of low data transmission efficiency and insufficient bandwidth in the above-mentioned prior art, and to provide a data center network networking method and a data center network based on dual-port servers and multi-port switches, so as to improve data transmission efficiency and network bandwidth.

[0005] The present invention is achieved through the following technical solutions: The embodiment of the present invention provides a method for networking a data center network, wherein the data center network includes m rows of dual-port servers and List port switch, and is an odd number, and the deployment is carried out by placing switches in odd columns and servers in even columns, where the number of switches in each column is , the number of servers in each column is .therefore Total number of switch ports , the total number of dual-port servers . and The number directly reflects the size of the data center network.

[0006] The certain rules are rules for interconnecting two adjacent columns of switches and servers, including a connection strategy. The connection strategy is based on numbering the servers and switches, wherein the numbering method of the switches and servers is the same, but the numbers of the two are independent of each other.

[0007] Each switch is identified by its row and column coordinates, using The logo is located OK, Column switches. The value ranges are . Similarly, use The logo is located OK, Column switches. The value ranges are .

[0008] After numbering the switches and servers in the deployment, the present invention provides the following connection strategy: List Port switch and In a network consisting of a row of dual-port servers, the servers are connected to switches in the left column of switches. Each server in the column , will be connected to the right port of the left switch column. and switches There is such a relationship: , .

[0009] exist The servers in the column will be connected to For servers Each server in the column , will be connected to the right switch For the connected server and switches There is such a relationship: . The servers in the column will be connected to switches in the column, where .

[0010] exist The servers in the column will be connected to For servers Each server in the column , will connect directly to the switch On the left port of the server, a ring structure is formed. and switches There is such a relationship:

[0011] After the switches and servers are connected according to the interconnection method, a data center network can be formed, in which the lines are parallel and do not cross, thereby realizing a three-dimensional ring data center network.

[0012] The present invention provides a data center network networking method based on the CLOS interconnection structure. and The number of switches directly reflects the size of the data center network. The servers fully utilize every port on the switch, maximizing the switch's input and output ports and achieving maximum switch utilization. It can be seen that the two adjacent rows of switches utilize the fully interconnected approach of the CLOS switching network, a rearrangeable, non-blocking connection method that ensures non-blocking information transmission. It can also be seen that the entire network structure is a CLOS modular, recursive interconnection. Thus, the data center network networking method of the present invention forms a large, three-dimensional, ring-shaped CLOS data center network. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the specific implementation of the present invention, the following will briefly introduce the drawings required for describing the specific implementation. Obviously, the drawings described below are only one implementation of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 A vertical view of the data center network structure of the present invention;

[0015] Figure 2 The structure of each basic unit in the present invention;

[0016] Figure 3 An example of a plan view of a network connection of the present invention is provided;

[0017] Figure 4It is a modular recursive interconnection schematic diagram of the network structure of the present invention. Implementation Method

[0018] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the technical solutions in the specific implementation methods of the present invention are clearly and completely described below to further illustrate the present invention. Obviously, the specific implementation methods described are only part of the implementation methods of the present invention, rather than all styles.

[0019] The present invention is further described in detail below with reference to the accompanying drawings:

[0020] like Figure 1 The figure shows a data center network networking method and a vertical view of a data center network structure of the present invention, wherein the data center network includes Column dual port server and List Port switch ( is an odd number and ).

[0021] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. As an example, the present invention is not limited to this example.

[0022] First, as Figure 2 As shown, the basic unit of the entire network is shown, namely Servers connected to one By connecting multiple basic units on the right port of the port switch, a complete network structure can be formed.

[0023] The complete network structure is as follows Figure 3 As shown, there are five columns of switches and servers, and the number of switch ports is 4 (i.e. , ) network plane view.

[0024] Figure 3 Each switch in is identified by its row and column coordinates, for example, the first column of switches The first switch in the column (Figure The coordinates of the switch with column number 0 are .

[0025] like Figure 3 As shown, for the convenience of description, we divide the ports of the switch into two parts as shown in the plan view. The left and right parts each contain half of the number of ports of the switch. The ports of each part are numbered from top to bottom. The switch, the connected server (coordinates are The server connected to it is numbered 0. Similarly, the other server connected to it is numbered 1.

[0026] The servers are connected to the switches in the left switch column. Each server in the column , will be connected to the right port of the left switch column. and switches There is such a relationship: , .exist Figure 2 The coordinates are The two servers are connected at coordinates on the switch.

[0027] The connection rules of the switch on the right side of the server are divided into three parts.

[0028] exist The servers in the column will be connected to For servers Each server in the column , will be connected to the right switch For the connected server and switches There is such a relationship: , .exist Figure 2 In the figure, the coordinates are The server connection is at coordinates on the switch.

[0029] exist The servers in the column will be connected to switches in the column, where .like Figure 4 As shown, in fact, the model is a recursive interconnection of modules.

[0030] exist The servers in the column will be connected to For servers Each server in the column , will connect directly to the switch On the left port of the server, a ring structure is formed. and switches There is such a relationship: exist Figure 2 The coordinates are The server connection is at coordinates on the switch.

Claims

1. A ring-type data center network, characterized in that: The data center network includes Column dual port server and List port switch, and The number of switches in each column is odd. , the number of servers in each column is Therefore, the total number of n-port switches is , the total number of dual-port servers .

2. According to claim 1, we consider a switch and all servers connected to its right port as a basic unit. The connection between the servers and the switch follows the following principle: the server is first connected to the right port of the switch to its left, forming a basic unit. Each unit is then connected to each other. For example, suppose there are three units, namely unit A, unit B, and unit C. Now, we want to connect the server in unit A to the left port of the switch in unit B. Similarly, we connect unit B to unit C, and unit C to unit A, forming a ring-shaped CLOS structure. The structure described herein is not limited to this example.

3. The data center network according to claim 2, wherein: The switches and servers are connected in a fully interconnected manner, which is a non-blocking connection method, forming a modular recursive interconnection structure. The data center network networking method has a series of advantages such as large bandwidth, load balancing, and good fault tolerance.