A data center server network redundant link monitoring method and device

Abstract

The present invention discloses a data center server network redundant link monitoring method, including: S1, mathematical modeling of the planning and deployment of data center cabinets, switches and servers, and establishing a first mapping relationship between the cabinet U position and the switch port; S2, connecting and mapping the server network port to the corresponding port of the switch; S3, the monitoring platform obtains the first switch side port information and the first server side network port information to generate a second mapping relationship as the initial network baseline information; S4, the monitoring platform obtains the second switch side port information and the second server side network port information, and generates the dynamic network baseline information of the server, compares the dynamic network baseline information with the initial network baseline information, and issues an alarm if they are inconsistent. The present invention also discloses a data center server network redundant link monitoring device. The present invention has the characteristics of high fault identification accuracy, high degree of automation and low business intrusion, and can be widely used in the field of network monitoring.

Description

Technical Field

[0001] The present invention relates to network monitoring and fault management technology, and in particular to a data center server network redundant link monitoring method and device. Background Art

[0002] Server network redundant links refer to the use of multiple redundant network links for data transmission in the server architecture in order to improve the stability and reliability of network connections. When a link fails or is congested, the system can automatically switch to other available links to ensure that data transmission is not interrupted or blocked. At present, the main means of monitoring server redundant links include: heartbeat frame technology, switch data analysis, and regular inspections and maintenance. By real-time monitoring and ensuring the stability and reliability of server network connections, single point failures can be prevented and network performance can be optimized.

[0003] However, compared with the growing demand for network monitoring, the current mainstream technologies for server network redundant link monitoring have their own shortcomings. They can be compared and analyzed from three aspects: fault identification accuracy, degree of automation, and business intrusion:

[0004] 1. Heartbeat frame technology relies on the underlying operating system of the server. When the redundant link is lost, it can correspond to the faulty server and related network cards, but it occupies business bandwidth and is invasive to the business. It has high fault identification accuracy, high degree of automation, and high business invasiveness;

[0005] 2. Switch data analysis can detect abnormalities in the status of a switch port, which can be matched to the MAC address of the downstream server network port, but cannot be associated with the specific server that has lost the redundant link. The fault identification accuracy is low, the degree of automation is high, and the business intrusion is low;

[0006] 3. Regular inspection and maintenance rely heavily on manual troubleshooting, lack of automated processes, and inability to monitor 24 hours a day. Fault identification accuracy is moderate, automation is low, and business intrusion is low. Summary of the invention

[0007] The purpose of the present invention is to overcome the deficiencies of the above-mentioned background technology and to provide a data center server network redundant link monitoring method and device, which have the characteristics of high fault identification accuracy, high degree of automation and low business intrusion.

[0008] The present invention provides a data center server network redundant link monitoring method, comprising the following steps: S1, when planning a data center, mathematical abstraction and mathematical modeling are performed on the planning and deployment of cabinets, switches and servers in the data center, and a first mapping relationship between a cabinet U position and a switch port is established based on the proposed mathematical model; S2, based on the first mapping relationship between the cabinet U position and the switch port, when deploying a server at the lowest U position, the server network port is connected and mapped to the corresponding port of the corresponding switch; S3, using an out-of-band management network of the data center, a monitoring platform respectively obtains first switch side port information and first server side network port information and generates a second mapping relationship between the switch port and the server network port, and stores the second mapping relationship as the initial network baseline information of the server; S4, initializing the initial network baseline information of the server, the monitoring platform provides 24-hour cyclic dynamic monitoring, the monitoring platform uses the out-of-band management network of the data center to obtain second switch side port information and second server side network port information in real time, and generates dynamic network baseline information of the server by using the second switch side port information and the second server side network port information, compares the dynamic network baseline information with the initial network baseline information, and issues an alarm if there is inconsistency.

[0009] In the above technical solution, in step S1, the specific process of the mathematical modeling is as follows: the cabinets of the data center are divided into multiple cabinet groups, each cabinet group contains at least 2 cabinets; each server is equipped with at least two physical network cards, and each physical network card is connected to a corresponding switch; the switches corresponding to each server are deployed in their respective cabinets, minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches.

[0010] In the above technical solution, in step S1, the algorithm for minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switch is as follows: The problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switch is expressed in the following form:

[0011] ,in, is the total hardware cost of the cabinet group, Defined as the number of cabinets contained in the cabinet group, Defined as the number of switches required for each cabinet group, Corresponding server The span to the left switch on its upstream link, Corresponding server The span to the switch on the right side of its upstream link, Defined as the sum of spans of all servers, Defined as the cabinet height in units of u,p Defined as the number of switch ports, Defined as the server height in u , Defined as the interval between two servers being put on the shelves, in units of u,c Defined as the upper limit of the number of servers that can be mounted in each cabinet due to factors such as power and space; set , , c when When the total hardware cost is the smallest, the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switch is transformed into the independent variable m Optimization problem , select the appropriate m The switch placement strategy is used to minimize the sum of the spans from all servers to the upstream switches. At the same time, the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches is further simplified. m Continuously recursively decompose into sub-problems, namely: ,in, Indicates that it contains m The deployment method of switches in a cabinet group; in particular, for different m The following switch placement strategy minimizes the sum of the spans from all servers to the upstream switches: When the number of cabinets in each cabinet group is When the number of cabinets is When the number of cabinets is When the number of cabinets is When the number of cabinets is When the number of cabinets in each cabinet group is , divide the cabinets in the cabinet group into two parts, and continue recursively until ,and , calculate the optimal solution: , , , , , ,

[0012]

[0013] , .

[0014] In the above technical solution, in step S2, the specific process of connecting and mapping the server network port to the corresponding port of the corresponding switch when deploying the server at the lowest U position is as follows: two switches of the same model are respectively deployed in two cabinets of the same cabinet group and configured as M-LAG or stacking mode; at least 24 ports of the two switches are respectively connected to each cabinet of the corresponding cabinet group, and the U positions in each cabinet with an internal serial number that is an integer multiple of 3 correspond to the corresponding ports of the two switches; each server is deployed with two network cards to configure link aggregation, and are respectively connected to ports with the same serial number on the two switches to ensure the redundancy of the network link.

[0015] In the above technical scheme, the specific process of step S3 is as follows: the monitoring platform uses the out-of-band management network of the data center to obtain the port status and the MAC address of the network port of the downstream server from the switch side respectively, and at the same time, the monitoring platform uses the out-of-band management network of the data center to obtain the network port status and the network port MAC address from the downstream server side respectively; compare whether the MAC addresses obtained from the switch side and the server side are the same. If different, the monitoring platform issues an alarm and re-establishes the first mapping relationship between the cabinet U position and the switch port; if they are the same, the monitoring platform generates a second mapping relationship between the switch port and the downstream server network port and saves the second mapping relationship as the initial network baseline information of the downstream server.

[0016] In the above technical solution, the specific process of step S4 is as follows: the monitoring platform periodically obtains the port status and the MAC address of the downstream server network port from the switch side through the out-of-band management network of the data center, and at the same time obtains the network port status and the network port MAC address from the server side to generate the dynamic network baseline information of the server; compares the dynamic network baseline information with the initial network baseline information to determine whether the server network link redundancy status is normal. If the dynamic network baseline information is consistent with the initial network baseline information, the server network link redundancy status is normal, and this round of monitoring ends; if the dynamic network baseline information is inconsistent with the initial network baseline information, it means that the server network link redundancy status is abnormal, the monitoring platform issues an alarm, and ends this round of monitoring.

[0017] In the above technical solution, in step S4, the specific process of comparing the dynamic network baseline information with the initial network baseline information is as follows: preferentially determine the server network port status and the switch port status in the dynamic network baseline information; if the server network port status and the switch port status are both UP, compare the consistency of the server network port MAC in the initial network baseline information and the dynamic network baseline information; if the server network port status and the switch port status are both DOWN, trigger a server network link alarm; if the server network port status is DOWN and the switch port status is UP, or the server network port status is UP, but the server network port MAC in the initial network baseline information and the dynamic network baseline information are inconsistent, trigger a server network port alarm; if the initial network baseline information and the server network port MAC in the dynamic network baseline information are consistent, compare the consistency of the switch configuration in the initial network baseline information and the dynamic network baseline information; if the server network port status is UP, the server network port MAC in the initial network baseline information and the dynamic network baseline information are the same, but the switch configuration in the initial network baseline information and the dynamic network baseline information are inconsistent, trigger a switch port alarm.

[0018] In the above technical solution, in step S4, after the monitoring platform issues an alarm, if the fault is eliminated, no alarm will be issued in the next round of dynamic monitoring.

[0019] The above technical solution also includes step S5, locating and handling the fault according to the alarm. When the fault handling involves the change of the initial network baseline information, the monitoring platform refreshes the initial network baseline information of the server according to the real-time data after the fault is recovered.

[0020] The present invention also provides a data center server network redundant link monitoring device, which has a computer program, and the computer program can execute the data center server network redundant link monitoring method.

[0021] The data center server network redundant link monitoring method and device of the present invention have the following beneficial effects:

[0022] 1. Automatically scan the redundant availability of server network links around the clock, obtain the network port data of server network cards and related data of switch ports in real time through the out-of-band management network of the data center, and generate the network baseline of the server. The monitoring platform accurately matches the server with lost redundant links and its faulty network card by comparing the changes in the server network baseline;

[0023] 2. The entire redundant link monitoring process does not invade the server operating system, does not occupy business bandwidth, is unaware of the user, and is fully automated. It has the characteristics of high fault identification accuracy, high degree of automation, and low business intrusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A platform architecture diagram for carrying the data center server network redundant link monitoring method and device of the present invention;

[0025] Figure 2 It is a schematic diagram of the overall process of the data center server network redundant link monitoring method of the present invention;

[0026] Figure 3 It is a flow chart of mathematical modeling in the planning and deployment process of data center cabinets, switches and servers in step S1 of the data center server network redundant link monitoring method of the present invention;

[0027] Figure 4 A server arrangement U bit and switch port mapping relationship diagram of a specific embodiment of step S2 in the data center server network redundant link monitoring method of the present invention;

[0028] Figure 5 This is a specific flow chart of step S3 in the data center server network redundant link monitoring method of the present invention;

[0029] Figure 6 It is a specific flow chart of comparing dynamic network baseline information with initial network baseline information in step S4 of the data center server network redundant link monitoring method of the present invention;

[0030] Figure 7 It is a specific flow chart of an embodiment of triggering a platform warning in step S4 of the data center server network redundant link monitoring method of the present invention;

[0031] Figure 8 It is a structural schematic diagram of a data center server network redundant link monitoring device of the present invention. DETAILED DESCRIPTION

[0032] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments, but the embodiments should not be construed as limiting the present invention.

[0033] In view of the shortcomings of the prior art, the present invention provides a data center server network redundant link monitoring method and device, the present invention needs to be built on a platform, the platform architecture is as follows Figure 1 shown.

[0034] See also Figure 2 The data center server network redundant link monitoring method of the present invention comprises the following steps:

[0035] S1. When planning a data center, mathematical abstraction and mathematical modeling are performed on the planning and deployment process of data center cabinets, switches and servers. Based on the proposed mathematical model, the first mapping relationship between the U position of the data center cabinet (the U position of the data center cabinet is the basic unit used by the data center industry to calibrate the physical space of the cabinet to store equipment, and is also the basic unit for calibrating the physical size of servers, storage, switches and other equipment) and the switch port is further established;

[0036] The steps of mathematical modeling in the planning and deployment of data center cabinets, switches and servers are as follows: Figure 3 As shown:

[0037] S11. The cabinets of the data center can be divided into cabinet groups according to business needs and space distribution. Each cabinet group includes 2 or more cabinets. In a preferred embodiment, every four consecutive cabinets form a cabinet group.

[0038] The overall optimization goal of mathematical modeling is: for each cabinet group, eliminate the single point of failure in the data center, while minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches;

[0039] S12. In order to ensure high redundancy of server network links and completely eliminate the risk of single point failure, each server must be configured with at least two independent physical network links. Specifically, each server should be equipped with at least two physical network cards, which need to be connected to two different switches respectively. In addition, in order to further enhance the reliability of the system, considering the cabinet-level redundancy of the switches, these switches should be deployed in different cabinets.

[0040] S13. At the same time, the number of cabinets between the server and its upstream switch is defined as the span from the server to the switch. In order to reduce the cost and related risks of wiring, it is required to minimize the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches:

[0041] Based on the above premise, in order to solve the technical problem of minimizing the data center hardware cost and server span, the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches is expressed as follows:

[0042]

[0043] in, is the total hardware cost of the cabinet group, Defined as the number of cabinets contained in the cabinet group, Defined as the number of switches required for each cabinet group. Corresponding server The span to the left switch on its upstream link, Corresponding server The span to the switch on the right side of its upstream link,

[0044] Defined as the sum of spans of all servers, Defined as the cabinet height in units of u , p Defined as the number of switch ports, Defined as the server height in u , Defined as the interval between two servers being put on the shelves, in units of u , c Defined as the upper limit of the number of servers that can be mounted in each cabinet due to factors such as power and space;

[0045] Combined with the actual production scenario, set , , c It can be found that when When , the total hardware cost can be minimized, then the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switch can be transformed into the independent variable m Optimization problem , it is necessary to select an appropriate m and switch placement strategy to minimize the sum of the spans from all servers to the upstream switches.

[0046] At the same time, the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches can be further simplified. m Continue to recursively decompose it into sub-problems, namely:

[0047]

[0048] in, Indicates that it contains m The deployment method of switches in a cabinet group, especially for different m The following switch placement strategy can minimize the sum of the spans from all servers to the upstream switches: When the number of cabinets is When the number of cabinets is When the number of cabinets is When the number of cabinets is When, the switches are deployed in the second, third, and fifth cabinets;

[0049] When the number of cabinets in each cabinet group , divide the cabinets in the cabinet group into two parts, and continue recursively until ,and , using computer programming methods, calculate the optimal solution:

[0050] ,

[0051] ,

[0052] ,

[0053]

[0054] , .

[0055] S2. When the data center server is deployed on the rack, based on the first mapping relationship between the existing data center cabinet U position and the switch port, the lowest U position of the server on the rack can be mapped to the corresponding port of the corresponding switch. Use a network cable to connect the server network port and the corresponding port of the switch to complete the wiring work. The process is as follows:

[0056] After mathematical modeling of the planning and deployment process of data center cabinets, switches and servers, in the optimal embodiment, when the number of cabinets in the cabinet group is 4, the specific process of setting the mapping relationship between the U position and the switch port is described as follows using a specific embodiment. Figure 4 As shown:

[0057] S21. The cabinets of the data center are divided into a group of four consecutive cabinets to form a plurality of cabinet groups. In each cabinet group, the cabinets are numbered from left to right as cabinet 1, cabinet 2, cabinet 3, and cabinet 4. To ensure high reliability of the link, two 48-port 10G optical switches A and B of the same model are respectively deployed in cabinet 2 and cabinet 3 and configured in M-LAG or stacking mode. Of course, in another embodiment, the cabinets of the data center can be divided into a group of two consecutive cabinets. In each cabinet group, cabinet 1 and cabinet 2 are numbered from left to right, and two 48-port 10 Gigabit optical switches A and B of the same model are respectively deployed in the two cabinets, namely cabinet 1 and cabinet 2; in another embodiment, the cabinets of the data center can be divided into groups of three consecutive cabinets, and in each cabinet group, cabinet 1, cabinet 2 and cabinet 3 are numbered from left to right, and two 48-port 10 Gigabit optical switches A and B of the same model are respectively deployed in the two cabinets, namely cabinet 1, cabinet 2 and cabinet 3;

[0058] S22, the 48 ports of switches A and B are respectively connected to the four cabinets of the cabinet group: among them, ports 1-12 are connected to cabinet 1, ports 13-24 are connected to cabinet 2, ports 25-36 are connected to cabinet 3, and ports 37-48 are connected to cabinet 4. Inside each cabinet, from 36U to 3U of the cabinet, with 3U as the interval, they correspond to different switch ports respectively; of course, in another embodiment, the 24 ports of switches A and B are respectively connected to the two cabinets of the cabinet group: among them, ports 1-12 are connected to cabinet 1, ports 13-24 are connected to cabinet 2, ports 25-36 are connected to cabinet 3, and ports 37-48 are connected to cabinet 4. 24 ports are connected to cabinet 2, ports 25-48 are not connected, and inside each cabinet, from 36U to 3U of the cabinet, with 3U as interval, they correspond to different switch ports respectively; in another embodiment, 36 ports of switches A and B are connected to three cabinets of the cabinet group respectively: among them, ports 1-12 are connected to cabinet 1, ports 13-24 are connected to cabinet 2, ports 25-36 are connected to cabinet 3, ports 37-48 are not connected, and inside each cabinet, from 36U to 3U of the cabinet, with 3U as interval, they correspond to different switch ports respectively;

[0059] S23. Each server deploys two network cards to configure link aggregation, which are connected to the ports with the same serial number on switch A and switch B respectively, thereby ensuring the redundancy of the network link.

[0060] S3. After the data center server is put on the rack and the wiring work is completed, the monitoring platform uses the out-of-band management network of the data center to obtain the first switch side port information and the first server side network port information respectively, and then generates a second mapping relationship between the switch port and the server network port, and stores the second mapping relationship as the initial network baseline information of the server. The specific process is as follows: Figure 5 As shown:

[0061] After the server is deployed according to the rules described in step S2, the monitoring platform uses the out-of-band management network of the data center to obtain the port status and the MAC address of the server network port connected to the port from the switch side, and also obtains the network port status and network port MAC address from the server side, and compares whether the MAC addresses obtained from the switch side and the server side are the same. If they are different, it means that step S1 is not executed correctly, and the monitoring platform issues an alarm and re-executes step S1; if they are the same, the monitoring platform generates and saves the initial network baseline information of the server according to the second mapping relationship between the switch port and the server network port;

[0062] In this step, after the data center server is put on the rack and the wiring work is completed, the format of the initial network baseline information of the server generated and the corresponding field description are shown in Table 1 and Table 2 respectively:

[0063] Table 1 Format of the server's initial network baseline information

[0064]

[0065] Table 2 Corresponding fields of the server's initial network baseline information

[0066]

[0067] S4. After the network baseline information of the data center server is initialized, the monitoring platform performs 24-hour uninterrupted dynamic monitoring. Using the out-of-band management network of the data center, the monitoring platform regularly and actively obtains the port information of the second switch side and the network port information of the second server side, and generates the dynamic network baseline information of the server. The dynamic network baseline information is compared with the initial network baseline information. If there is any inconsistency, an alarm is issued. The specific process is as follows: Figure 6 As shown:

[0068] After the network baseline information of the data center server is initialized, the monitoring platform periodically obtains the port status and the MAC address of the downstream server network port from the switch side through the out-of-band management network of the data center, and simultaneously obtains the network port status and network port MAC address from the server side to generate the dynamic network baseline information of the server;

[0069] Compare the dynamic network baseline information with the initial network baseline information generated in step S3 to determine whether the network link redundancy status is normal. If the dynamic network baseline information is consistent with the initial network baseline information, the server network redundancy status is normal, and this round of monitoring ends; if the dynamic network baseline information is inconsistent with the initial network baseline information, it means that the server network redundancy status is abnormal, the monitoring platform issues an alarm, and this round of monitoring ends;

[0070] In a further embodiment, the comparison principle of the dynamic network baseline information and the initial network baseline information is as follows: each set of information of the server in the dynamic network baseline information and the initial network baseline information is compared one by one, including the UP / DOWN status of the server network port and the switch port; the static information of the server network port, such as the network port number, etc.; the matching relationship between the MAC address of the server network port and the MAC address of the network port under the switch port; and the configuration information of the switch port, such as the port VLAN, etc. If there is a data inconsistency during the comparison process, the corresponding type of platform warning will be triggered according to the specific category of the data;

[0071] In a specific embodiment, if data inconsistency occurs during the comparison process, the specific process of triggering a corresponding type of platform warning will be as follows based on the specific category of the data. Figure 7 :

[0072] Prioritize the server network port status and switch port status in the dynamic network baseline information. If both are UP, compare the consistency of the server network port MAC in the initial network baseline information and the dynamic network baseline information. If both are DOWN, trigger the server network link alarm.

[0073] If the server network port status is DOWN and the switch port status is UP, or the server network port status is UP but the server network port MAC in the initial network baseline information is inconsistent with that in the dynamic network baseline information, the server network port alarm is triggered. If the server network port MAC in the initial network baseline information is consistent with that in the dynamic network baseline information, the switch configuration consistency in the initial network baseline information and the dynamic network baseline information is compared;

[0074] If the server network port status is UP, the server network port MAC in the initial network baseline information is the same as that in the dynamic network baseline information, but the switch configuration in the initial network baseline information is inconsistent with that in the dynamic network baseline information, then the switch port alarm is triggered;

[0075] After the monitoring platform issues an alarm, if the fault is eliminated, no alarm will be issued in the next round of monitoring.

[0076] S5. Locate and handle the fault according to the alarm. When the fault handling involves changes in the initial network baseline information, the monitoring platform refreshes the initial network baseline information of the server according to real-time data after the fault is restored. In particular, when the server has a network card replacement operation, the MAC address of the network port changes after the replacement. After the network card replacement process is completed, the monitoring platform automatically refreshes the initial network baseline information of the server according to real-time data.

[0077] See also Figure 8 The data center server network redundant link monitoring device of the present invention includes the following contents:

[0078] Mathematical modeling module: When planning a data center, mathematical abstraction and mathematical modeling are performed on the planning and deployment of the cabinets, switches and servers of the data center, and the first mapping relationship between the cabinet U position and the switch port is established based on the proposed mathematical model;

[0079] Mapping module: based on the first mapping relationship between the cabinet U position and the switch port, the server network port is connected and mapped to the corresponding port of the corresponding switch when the server is deployed at the lowest U position;

[0080] Initial network baseline information module: using the out-of-band management network of the data center, the monitoring platform obtains the first switch side port information and the first server side network port information respectively and generates a second mapping relationship between the switch port and the server network port, and stores the second mapping relationship as the initial network baseline information of the server;

[0081] Comparison alarm module: Initialize the initial network baseline information of the server. The monitoring platform provides 24-hour cyclic dynamic monitoring. The monitoring platform uses the out-of-band management network of the data center to obtain the second switch side port information and the second server side network port information in real time, and generates the dynamic network baseline information of the server by using the second switch side port information and the second server side network port information. The dynamic network baseline information is compared with the initial network baseline information. If there is any inconsistency, an alarm is issued.

[0082] Real-time monitoring module: locates and handles faults based on alarms. When fault handling involves changes in the initial network baseline information, the monitoring platform refreshes the initial network baseline information of the server based on real-time data after the fault is restored.

[0083] Technical principles and innovations of the present invention:

[0084] 1. When the data center cabinet group has different numbers of cabinets, the deployment planning method of network switches in the cabinet group;

[0085] 2. A method for establishing a mapping relationship between a data center cabinet U position and a switch port;

[0086] 3. Methods and processes for implementing server redundant link monitoring based on the out-of-band management network of the data center.

[0087] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0088] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

Claims

1. A data center server network redundant link monitoring method, characterized in that: The steps include: S1. When planning a data center, mathematical abstraction and mathematical modeling are performed on the planning and deployment of the cabinets, switches and servers of the data center, and a first mapping relationship between the cabinet U position and the switch port is established based on the proposed mathematical model. The specific process of the mathematical modeling is as follows: The cabinets in the data center are divided into multiple cabinet groups, each cabinet group contains at least 2 cabinets; Each server is equipped with at least two physical network cards, and each physical network card is connected to the corresponding switch; The switches corresponding to each server are deployed in their own cabinets, minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches. The algorithm for minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches is as follows: The problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches is expressed as follows: ， in, is the total hardware cost of the cabinet group, Defined as the number of cabinets contained in the cabinet group, Defined as the number of switches required for each cabinet group, Corresponding server The span to the left switch on its upstream link, Corresponding server The span to the switch on the right side of its upstream link, Defined as the sum of spans of all servers, Defined as the cabinet height in units of u,p Defined as the number of switch ports, Defined as the server height in units of u , Defined as the interval between two servers being put on the shelves, in units of u,c Defined as the upper limit of the number of servers that can be mounted in each cabinet due to power and space factors; set up , , c when When the total hardware cost is the smallest, the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switch is transformed into the independent variable m Optimization problem , select the appropriate m and switch placement strategy to minimize the sum of the spans from all servers to the upstream switches. At the same time, the problem of minimizing the total hardware cost of the cabinet group and the sum of the spans from all servers to the upstream switches is further simplified. m Continuously recursively decompose into sub-problems, namely: in, Indicates that it contains m The deployment method of switches in a cabinet group; in particular, for different m The following switch placement strategy minimizes the sum of the spans from all servers to the upstream switches: When the number of cabinets in each cabinet group is When the number of cabinets is When the number of cabinets is When the number of cabinets is When the number of cabinets is When, the switches are deployed in the second, third, and fifth cabinets; When the number of cabinets in each cabinet group , divide the cabinets in the cabinet group into two parts, and continue recursively until ,and , calculate the optimal solution: ， ， ， ， ； S2. Based on the first mapping relationship between the cabinet U position and the switch port, when the server is deployed at the lowest U position, the server network port is connected and mapped to the corresponding port of the corresponding switch; S3. Using the out-of-band management network of the data center, the monitoring platform obtains the first switch side port information and the first server side network port information respectively and generates a second mapping relationship between the switch port and the server network port, and stores the second mapping relationship as the initial network baseline information of the server; S4. Initialize the initial network baseline information of the server. The monitoring platform provides 24-hour cyclic dynamic monitoring. The monitoring platform uses the out-of-band management network of the data center to obtain the second switch side port information and the second server side network port information in real time, and generates the dynamic network baseline information of the server with the second switch side port information and the second server side network port information. The dynamic network baseline information is compared with the initial network baseline information. If there is any inconsistency, an alarm is issued.

2. The data center server network redundant link monitoring method according to claim 1, characterized in that: In step S2, the specific process of connecting and mapping the server network port to the corresponding port of the corresponding switch when deploying the server at the lowest U position is as follows: Two switches of the same model are deployed in two cabinets of the same cabinet group and configured in M-LAG or stacking mode; At least 24 ports of the two switches are connected to the cabinets of the corresponding cabinet group, and the U bits in each cabinet whose serial numbers are integer multiples of 3 correspond to the corresponding ports of the two switches; Each server deploys two network cards to configure link aggregation and connects to the ports with the same serial number on the two switches to ensure the redundancy of the network link.

3. The data center server network redundant link monitoring method according to claim 2, characterized in that: The specific process of step S3 is as follows: The monitoring platform uses the out-of-band management network of the data center to obtain the port status and the MAC address of the port of the server connected to the port from the switch side. At the same time, the monitoring platform uses the out-of-band management network of the data center to obtain the port status and the MAC address of the port from the server side. Compare whether the MAC addresses obtained on the switch side and the server side are the same. If they are different, the monitoring platform will issue an alarm and re-establish the first mapping relationship between the cabinet U position and the switch port; if they are the same, the monitoring platform will generate a second mapping relationship between the switch port and the downstream server network port and save the second mapping relationship as the initial network baseline information of the downstream server.

4. The data center server network redundant link monitoring method according to claim 3, characterized in that: The specific process of step S4 is as follows: The monitoring platform periodically obtains the port status and the MAC address of the downstream server network port from the switch side through the out-of-band management network of the data center. At the same time, the monitoring platform obtains the network port status and network port MAC address from the server side to generate dynamic network baseline information for the server; Compare the dynamic network baseline information with the initial network baseline information to determine whether the server network link redundancy status is normal. If the dynamic network baseline information is consistent with the initial network baseline information, the server network link redundancy status is normal and this round of monitoring ends. If the dynamic network baseline information is inconsistent with the initial network baseline information, it means that the server network link redundancy status is abnormal, the monitoring platform issues an alarm, and ends this round of monitoring.

5. The data center server network redundant link monitoring method according to claim 4, characterized in that: In step S4, the specific process of comparing the dynamic network baseline information with the initial network baseline information is as follows: Prioritize the server network port status and switch port status in the dynamic network baseline information. If both are UP, compare the consistency of the server network port MAC in the initial network baseline information and the dynamic network baseline information. If both are DOWN, trigger the server network link alarm. If the server network port status is DOWN and the switch port status is UP, or the server network port status is UP but the server network port MAC in the initial network baseline information is inconsistent with that in the dynamic network baseline information, the server network port alarm is triggered. If the server network port MAC in the initial network baseline information is consistent with that in the dynamic network baseline information, the switch configuration consistency in the initial network baseline information and the dynamic network baseline information is compared; If the server network port status is UP, the server network port MAC in the initial network baseline information is the same as that in the dynamic network baseline information, but the switch configuration in the initial network baseline information is inconsistent with that in the dynamic network baseline information, a switch port alarm is triggered.

6. The data center server network redundant link monitoring method according to claim 5, characterized in that: In step S4, after the monitoring platform issues an alarm, if the fault is eliminated, no alarm will be issued in the next round of dynamic monitoring.

7. The data center server network redundant link monitoring method according to claim 6, characterized in that: The method also includes step S5, locating and handling the fault according to the alarm. When the fault handling involves a change in the initial network baseline information, the monitoring platform refreshes the initial network baseline information of the server according to real-time data after the fault is recovered.

8. A data center server network redundant link monitoring device, comprising a computer program, characterized in that: The computer program can execute the data center server network redundant link monitoring method as described in any one of claims 1 to 7.

Images