Server cluster load balancing method based on PC farm array distribution
By forming an operation correction diagram and intermittent transmission solution, the load balancing of the PC farm server cluster is optimized, and the delay problem caused by insufficient redundancy of the server cluster is solved, and a uniform load with low latency is achieved.
Patent Information
- Application Number
- CN202510757517.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-09
- Publication Date
- 2025-07-04
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In PC farms, when a server cluster fails, the load balancing method cannot effectively handle the insufficient redundancy of the server cluster, resulting in significant latency and affecting the user experience.
By forming operation correction diagrams, intermittent transmission schemes and reserved time, the information transmission path is optimized, and the idle transmission volume of the unfailed server is load-balanced, and the sense of delay is controlled.
In the case of insufficient redundancy of the server cluster, intermittent processing and fine-tuning methods, the information transmission delay after uniform load is reduced and the user experience quality is ensured.
Smart Images
Figure CN120256151A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and specifically relates to a server cluster load balancing method based on the array distribution of a PC farm. Background Art
[0002] A PC farm is a high-density computing cluster constructed by stacking multiple independent PC systems, and its array distribution design aims to achieve high performance, flexible expansion, and efficient management. When applying the PC farm, a server cluster assists in data transmission for it. Usually, the server cluster can fully meet the operation of the PC systems in the PC farm, but the server cluster may have faults. Once a fault occurs, it is necessary to reallocate the load of the server cluster. When performing load balancing, the most important thing is the handling of latency. Since the scale of the fault may have a high probability, the redundancy of the server cluster may be insufficient. Although the use of ordinary methods can achieve uniform load, the latency may be relatively obvious, and the quality of uniform load needs to be improved. Summary of the Invention
[0003] To solve the above technical problems, a server cluster load balancing method based on the array distribution of a PC farm is provided, and the technical solution of the present invention solves the problems raised in the above background art.
[0004] To achieve the above object, the technical solution adopted by the present invention is as follows: A server cluster load balancing method based on the array distribution of a PC farm, comprising: Obtain at least one PC system constituting the PC farm, and obtain a server cluster for the uniform load of the PC system, where the server cluster is composed of at least one server; Analyze the running connection relationship between the PC farm and the server cluster to obtain a running topology diagram; Perform numerical setting on the running topology diagram to obtain a running correction diagram; Obtain a faulty server as a fault server, and respectively use at least one PC system connected to the fault server as a target PC system; Based on historical data, obtain the latency allowance value of the PC system. Based on the latency allowance value, obtain at least one server for the retransmission of information of the target PC system, and respectively use them as target servers. Add up the idle transmission amounts of all the target servers to obtain the total idle transmission amount; Add up the information transmission demand amounts of all the target PC systems to obtain the total information transmission demand amount. When the total information transmission demand amount does not exceed the total idle transmission amount of the target servers, allocate the information transmission demand amount of the target PC system to the target servers to complete the transmission; When the total information transmission demand exceeds the total idle transmission capacity of the target server, the time of information transmission is segmented at intervals of the delay allowance value to obtain at least one sample time interval; Form an intermittent transmission scheme for the target PC system based on the idle transmission capacity of the target server; Analyze the PC system served by the target server to obtain the reserved time, and use the reserved time and the intermittent transmission scheme to control the transmission of the target server in the sample time interval; Take the part of the information transmission demand that the target server has not completed as the characteristic demand, and evenly distribute the characteristic demand to the servers other than the target server for transmission.
[0005] Preferably, the analysis of the operating connection relationship between the PC farm and the server cluster to obtain the operating topology diagram includes the following steps: Take the servers as server nodes, take the PC systems as PC system nodes, connect the line segments between the server nodes and the PC system nodes they serve, and take the graph composed of the server nodes, PC system nodes and line segments as the operating topology diagram.
[0006] Preferably, the numerical setting of the operating topology diagram to obtain the operating correction diagram includes the following steps: Statistically analyze the actual delay time when the server node transmits information to the PC system node, and take the actual delay time as the length of the line segment between the server node and the PC system node to obtain the operating correction diagram.
[0007] Preferably, the obtaining of the delay allowance value of the PC system based on historical data includes the following steps: Obtain at least one sample user, randomly match the transmission between the server and the PC system, record the sample delay time of the PC system and the feedback result of the sample user using the PC system, and correspond the sample delay time of the PC system and the feedback result of the sample user using the PC system; When the feedback result is no delay, take the feedback result as the target feedback result, and take the maximum value of the sample delay time corresponding to the target feedback result as the delay allowance value.
[0008] Preferably, the obtaining of at least one server for retransmitting information for the target PC system based on the delay allowance value includes the following steps: Use the non-faulty servers to transmit the information of the target PC system, and obtain the target delay time of the non-faulty servers serving the target PC system; Take the non-faulty servers with the target delay time less than the delay allowance value as the target servers.
[0009] Preferably, the intermittent transmission scheme for the target PC system formed by the idle transmission volume of the target server includes the following steps: Summarize at least one target PC system served by the target server to form a target PC system set; Accumulate the information transmission requirements of the target PC systems in the target PC system set to obtain the characteristic transmission requirements; Obtain the ratio of the information transmission requirement of the target PC system to the characteristic transmission requirement as the characteristic ratio; Divide the idle transmission volume of the target server according to the characteristic ratio to obtain at least one idle partial transmission volume, pair the idle partial transmission volume with a proportion equal to the characteristic ratio in the idle transmission volume to the target PC system corresponding to the characteristic ratio, and the idle partial transmission volume transmits information to the paired target PC system; Summarize at least one target server serving the target PC system to form a target server set; Accumulate and superimpose the idle transmission volumes of the target servers in the target server set to obtain the target idle transmission volume; Obtain the ratio of the idle transmission volume of the target server in the target server set to the target idle transmission volume as the target ratio; Divide the information transmission requirement of the target PC system according to the target ratio to obtain at least one partial transmission requirement, and pair the partial transmission requirement with the target server, satisfying that the target ratio corresponding to the partial transmission requirement is equal to the target ratio corresponding to the paired target server; The idle partial transmission volume of the target server transmits to the partial transmission requirement paired to the target server in the target PC system; Use the idle partial transmission volume of the target server to perform intermittent transmission on the partial transmission requirement.
[0010] Preferably, the analysis of the PC system served by the target server to obtain the reserved time includes the following steps: Subtract the actual delay time of the PC system originally served by the target server from the delay allowance value to obtain the reserved time.
[0011] Preferably, the control of the transmission of the target server in the sample time interval using the reserved time and the intermittent transmission scheme includes the following steps: Obtain the target transmission time for the target server to transmit information for data with a size equal to the idle partial transmission volume in the target PC system, and divide the delay allowance value by the target transmission time to obtain the characteristic number; The local transmission requirements are segmented according to the magnitude of the idle local transmission volume, obtaining at least one transmission requirement block. Randomly select transmission requirement blocks whose number is equal to the number of features as necessary transmission requirement blocks, and use the remaining transmission requirement blocks as non-necessary transmission requirement blocks; Within each sample time interval, the idle local transmission volume of the target server sequentially transmits the necessary transmission requirement blocks; Within each sample time interval, the sample time interval is divided into a front interval and a rear interval. The length of the front interval is equal to the reserved time. Within the front interval, the idle transmission volume of the target server is for the non-necessary transmission requirement blocks of the target PC system. In the rear interval, the target server performs information transmission to the original service PC system.
[0012] Preferably, the evenly distributing the feature requirements to servers other than the target server for transmission includes the following steps: Take the servers other than the target server as non-target servers, obtain the local part of the feature requirements as the part to be measured, and each part to be measured is respectively less than the idle transmission volume of the non-target server; Take the transmission time of the non-target server for the part to be measured as the calibration time, sort the non-target servers in ascending order according to the calibration time to obtain a non-target server sequence; Generate an i-sequence, where the i-sequence is a sequence composed of the first i non-target servers in the non-target server sequence in sequence, and the value range of i is from 1 to the eigenvalue, and the eigenvalue is the number of non-target servers in the non-target server sequence; When the superimposed value of the idle transmission volumes of the non-target servers in the i-sequence is not less than the feature requirements, then take the i-sequence as the target i-sequence; Use the non-target servers in the target i-sequence with the shortest length to transmit the feature requirements.
[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: By forming an operation correction graph, forming an intermittent transmission scheme, and obtaining the reserved time, when a server in the server cluster fails, different uniform load processing can be performed according to the comparison between the total information transmission requirements and the total idle transmission volume. Thus, even when the redundancy of the server cluster is insufficient, through intermittent processing and fine-tuning, the delay of the information transmission after uniform load can be made as low as possible, so that users will not have an obvious sense of delay when using, thereby ensuring the quality of uniform load. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] Figure 1 It is a schematic flow chart of the server cluster load balancing method based on the PC farm array distribution of the present invention; Figure 2Flow chart for obtaining the latency allowance value of a PC system based on historical data according to the present invention; Figure 3 Flow chart for obtaining at least one server for retransmitting information for a target PC system based on the latency allowance value according to the present invention; Figure 4 Flow chart for forming an intermittent transmission scheme for a target PC system based on the idle transmission amount of a target server according to the present invention; Figure 5 Flow chart for controlling the transmission of a target server in a sample time interval using the reserved time and the intermittent transmission scheme according to the present invention; Figure 6 Flow chart for evenly distributing the characteristic demand quantity to servers other than the target server for transmission according to the present invention. Detailed implementation manners
[0015] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.
[0016] Refer to Figure 1 As shown, a server cluster load balancing method based on the distribution of a PC farm array includes: Obtain at least one PC system constituting the PC farm, and obtain a server cluster for evenly loading the PC system. The server cluster is composed of at least one server; Analyze the running connection relationship between the PC farm and the server cluster to obtain a running topology diagram; Perform numerical setting on the running topology diagram to obtain a running corrected diagram; Obtain a server with a fault as a faulty server, and use at least one PC system connected to the faulty server as a target PC system respectively; Based on historical data, obtain the latency allowance value of the PC system. Based on the latency allowance value, obtain at least one server for retransmitting information for the target PC system as a target server respectively. Add up the idle transmission amounts of all the target servers to obtain the total idle transmission amount; Add up the information transmission demand quantities of all the target PC systems to obtain the total information transmission demand quantity. When the total information transmission demand quantity does not exceed the total idle transmission amount of the target servers, allocate the information transmission demand quantity of the target PC system to the target servers to complete the transmission; When the total information transmission demand quantity exceeds the total idle transmission amount of the target servers, divide the time during information transmission at intervals of the latency allowance value to obtain at least one sample time interval; Form an intermittent transmission scheme for the target PC system based on the idle transmission volume of the target server; Analyze the PC system of the target server's service to obtain the reserved time, and use the reserved time and the intermittent transmission scheme to control the transmission of the target server in the sample time interval; Take the uncompleted part of the information transmission demand of the target server as the characteristic demand, and evenly distribute the characteristic demand to the servers other than the target server for transmission.
[0017] In this solution, the situation when the server cluster fails is mainly dealt with. When it fails, in order to avoid too large an adjustment range, the idle transmission volume of the non-failed server is used to replace the transmission job of the failed server. At the same time, in order to control the delay, a server with an insignificant delay needs to be used for the transmission of the PC system. Because there are many servers in the server cluster and their installation locations are different, the distances from the PC system to the servers are different. When selecting, servers with obvious delays need to be excluded and not considered. However, since the number of available servers is limited, when the failure is relatively large, the redundancy of the available servers is insufficient, and some unconventional methods need to be adopted for load balancing, otherwise, the delay will be obvious.
[0018] Analyze the running connection relationship between the PC farm and the server cluster to obtain the running topology diagram, including the following steps: Take the servers as server nodes, take the PC systems as PC system nodes, connect the line segments between the server nodes and the PC system nodes they serve, and take the graph composed of server nodes, PC system nodes and line segments as the running topology diagram.
[0019] Perform numerical settings on the running topology diagram to obtain the running corrected diagram, including the following steps: Statistically record the actual delay time when the server node transmits information to the PC system node, and take the actual delay time as the length of the line segment between the server node and the PC system node to obtain the running corrected diagram.
[0020] The main function of the running corrected diagram is to be used for server screening, which can more intuitively display the selected servers and the screening process, and it is easy to find selection errors.
[0021] Refer to Figure 2 As shown, based on historical data, obtaining the delay tolerance value of the PC system includes the following steps: Obtain at least one sample user, randomly match the transmission between the server and the PC system, record the sample delay time of the PC system and the feedback results of the sample user using the PC system, and correspond the sample delay time of the PC system and the feedback results of the sample user using the PC system; When the feedback result indicates no delay, the feedback result is used as the target feedback result, and the maximum value of the sample delay time corresponding to the target feedback result is used as the delay allowance value.
[0022] During information transmission, no matter how fast the speed is, there will be delays. However, when the delay is small, users will not feel the delay, but when the delay is large, users can feel the delay. Therefore, in order to control the uniformly loaded delay, a delay allowance value needs to be formed to identify the delay situation and perform subsequent differential processing based on the identified result.
[0023] Refer to Figure 3 As shown, based on the delay allowance value, obtaining at least one server for retransmitting information for the target PC system includes the following steps: Use the non-faulty servers to transmit the information of the target PC system and obtain the target delay time of the non-faulty servers serving the target PC system; Take the non-faulty servers with a target delay time less than the delay allowance value as the target servers.
[0024] During matching, the target PC system will match many target servers, and the target servers will also be matched to multiple target PC systems. Therefore, it is necessary to allocate the idle transmission volume of the target servers. Moreover, since the total amount of the information transmission demand of the target PC system exceeds the total amount of the idle transmission volume of the target servers, direct allocation is not possible, and multiple steps need to be set for gradual allocation.
[0025] Refer to Figure 4 As shown, forming an intermittent transmission scheme for the target PC system for the idle transmission volume of the target servers includes the following steps: Summarize at least one target PC system served by the target servers to form a target PC system set; Accumulate the information transmission demands of the target PC systems in the target PC system set to obtain the characteristic transmission demand; Obtain the ratio of the information transmission demand of the target PC system to the characteristic transmission demand as the characteristic ratio; Divide the idle transmission volume of the target servers according to the characteristic ratio to obtain at least one idle partial transmission volume. Pair the idle partial transmission volume with a proportion equal to the characteristic ratio in the idle transmission volume to the target PC system corresponding to the characteristic ratio, and the idle partial transmission volume performs information transmission to the paired target PC system; Summarize at least one target server serving the target PC system to form a target server set; Accumulate and superimpose the idle transmission volumes of the target servers in the target server set to obtain the target idle transmission volume; Obtain the ratio of the idle transmission volume of the target server in the target server set to the target idle transmission volume as the target ratio; Divide the information transmission demand of the target PC system according to the target ratio to obtain at least one partial transmission demand, and pair the partial transmission demand with the target server, so that the target ratio corresponding to the partial transmission demand is equal to the target ratio corresponding to the paired target server; The idle partial transmission volume of the target server transmits the partial transmission demand paired to the target server in the target PC system; Use the idle partial transmission volume of the target server to intermittently transmit the partial transmission demand.
[0026] During allocation, due to the different idle conditions of the target servers, the different information transmission demands of the target PC system, and the complex relationship between the target servers and the target PC system, which is not a one-to-one relationship, but a relationship where one target server corresponds to multiple target PC systems and multiple target servers correspond to one target PC system. Therefore, it is necessary to determine the correspondence between the idle partial transmission volume of the target server and the partial transmission demand, so that in the follow-up, transmission can be carried out according to the correspondence, and further fine-tuning can be carried out according to the transmission results. Because the total information transmission demand exceeds the total idle transmission volume of the target server, the idle partial transmission volume cannot complete the transmission demand of the partial transmission demand. Therefore, the untransmitted part needs to be transmitted in other ways, that is, using the subsequent reserved time for transmission.
[0027] Analyze the PC systems served by the target server to obtain the reserved time, including the following steps: Subtract the actual delay time of the PC system originally served by the target server from the delay allowance value to obtain the reserved time.
[0028] The reserved time is the part where the actual delay time is smaller than the delay allowance value. As long as the delay does not reach the delay allowance value, it will not be felt. Therefore, for the reserved time, it can be used to transmit the untransmitted data, as long as the sum of the added data and the actual delay time of the PC system originally served by the target server does not exceed the delay allowance value. Because the delay of the PC system originally served by the target server will not exceed the delay allowance value, it will not affect the original transmission. And because the untransmitted data is transmitted immediately, according to the initial selection of the target server, it satisfies that the delay will not exceed the delay allowance value. Therefore, for this part of the transmission, there will also be no obvious delay. However, in this case of transmission supplement, there may still be some information that is not transmitted. For these information, it can only be completed by means of delayed transmission, which will be further explained later.
[0029] Refer toFigure 5 As shown, controlling the transmission of the target server in the sample time interval using the reserved time and intermittent transmission scheme includes the following steps: Obtain the target transmission time for the target server to transmit information for data in the target PC system with a size equal to the idle local transmission volume. Divide the delay allowance value by the target transmission time to obtain the number of features; Divide the local transmission demand according to the size of the idle local transmission volume to obtain at least one transmission demand block. Randomly select the number of transmission demand blocks equal to the number of features as the necessary transmission demand blocks, and use the remaining transmission demand blocks as non-necessary transmission demand blocks; Within each sample time interval, the target server's idle local transmission volume sequentially transmits the necessary transmission demand blocks; Within each sample time interval, divide the sample time interval into a front interval and a back interval. The length of the front interval is equal to the reserved time. In the front interval, the target server's idle transmission volume is for the non-necessary transmission demand blocks of the target PC system. In the back interval, the target server transmits information to the original service PC system.
[0030] Refer to Figure 6 As shown, evenly distributing the feature demand to servers other than the target server for transmission includes the following steps: Regard the services other than the target server as non-target servers. Obtain the local part of the feature demand as the part to be measured, and each part to be measured is respectively smaller than the idle transmission volume of the non-target server; Take the transmission time of the non-target server for the part to be measured as the calibration time, and sort the non-target servers in ascending order according to the calibration time to obtain the non-target server sequence; Generate an i-sequence, where the i-sequence is a sequence composed of the first i non-target servers in the non-target server sequence in sequence, and the value range of i is from 1 to the feature value, and the feature value is the number of non-target servers in the non-target server sequence; When the superimposed value of the idle transmission volumes of the non-target servers in the i-sequence is not less than the feature demand, then regard the i-sequence as the target i-sequence; Use the non-target servers in the target i-sequence with the shortest length to transmit the feature demand.
[0031] All the remaining information transmission tasks need to be completed here. Therefore, the condition for setting the target i-sequence is that it can complete the transmission of the feature demand. However, since the delay of the non-target server for the transmission of the feature demand is perceivable, because it may be far from the PC system, therefore, in order to control its delay, try to use the non-target server with a small delay to complete the transmission. Thus, use the non-target servers in the target i-sequence with the shortest length to transmit the feature demand.
[0032] Furthermore, the present solution also proposes a storage medium, on which a computer-readable program is stored. When the computer-readable program is called, it executes the above-mentioned server cluster load balancing method based on the PC farm array distribution.
[0033] It can be understood that the storage medium can be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid state disk (SSD).
[0034] In summary, the advantages of the present invention are as follows: By forming a running correction diagram, forming an intermittent transmission scheme, and obtaining a reserved time, when a server in the server cluster fails, different uniform load processing can be performed according to the comparison between the total information transmission demand and the total idle transmission volume. Thus, even when the redundancy of the server cluster is insufficient, the delay of information transmission after uniform load can still be made as low as possible through intermittent processing and fine-tuning, so that users will not have an obvious sense of delay when using it, thereby ensuring the quality of uniform load.
[0035] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.
Claims
1. A server cluster load balancing method based on the array distribution of a PC farm, characterized in that, Including: Obtain at least one PC system that constitutes a PC farm, and obtain a server cluster for the uniform load of the PC system. The server cluster is composed of at least one server; Analyze the running connection relationship between the PC farm and the server cluster to obtain a running topology diagram; Perform numerical setting on the running topology diagram to obtain a running corrected diagram; Obtain the server with a fault as a faulty server, and use at least one PC system connected to the faulty server as a target PC system respectively; Based on historical data, obtain the delay allowance value of the PC system. Based on the delay allowance value, obtain at least one server for retransmitting the information of the target PC system as a target server respectively. Add up the idle transmission amounts of all target servers to obtain the total idle transmission amount; Add up the information transmission demand amounts of all target PC systems to obtain the total information transmission demand amount. When the total information transmission demand amount does not exceed the total idle transmission amount of the target servers, allocate the information transmission demand amount of the target PC system to the target servers to complete the transmission. Among them, both the total information transmission demand amount and the total idle transmission amount are statistical results within the delay allowance value; When the total information transmission demand amount exceeds the total idle transmission amount of the target servers, divide the time during information transmission at intervals of the delay allowance value to obtain at least one sample time interval; Form an intermittent transmission scheme for the target PC system based on the idle transmission amount of the target server; Analyze the PC systems served by the target server to obtain the reserved time, and use the reserved time and the intermittent transmission scheme to control the transmission of the target server in the sample time interval; Take the part of the information transmission demand amount that the target server has not completed as the characteristic demand amount, and evenly distribute the characteristic demand amount to the servers other than the target server for transmission.
2. A server cluster load balancing method based on the array distribution of a PC farm according to claim 1, characterized in that, The analysis of the running connection relationship between the PC farm and the server cluster to obtain the running topology diagram includes the following steps: Take the server as a server node, take the PC systems as PC system nodes, connect the line segments between the server node and the PC system nodes it serves, and take the graph composed of the server node, the PC system node and the line segments as the running topology diagram.
3. A server cluster load balancing method based on the array distribution of a PC farm according to claim 2, characterized in that, The numerical setting of the running topology diagram to obtain the running corrected diagram includes the following steps: Statistically record the actual delay time when the server node transmits information to the PC system node, and take the actual delay time as the length of the line segment between the server node and the PC system node to obtain the running corrected diagram.
4. A server cluster load balancing method based on the array distribution of a PC farm according to claim 3, characterized in that, The obtaining of the delay allowance value of the PC system based on historical data includes the following steps: Obtain at least one sample user, randomly match and transmit between the server and the PC system, record the sample delay time of the PC system and the feedback result of the sample user using the PC system, and correspond the sample delay time of the PC system and the feedback result of the sample user using the PC system; When the feedback result is no delay, take the feedback result as the target feedback result, and take the maximum value of the sample delay time corresponding to the target feedback result as the delay allowance value.
5. A server cluster load balancing method based on the array distribution of a PC farm according to claim 4, characterized in that, The steps for obtaining at least one server for retransmitting information for the target PC system based on the delay allowance value are as follows: Transmit the information of the target PC system using the non-faulty servers, and obtain the target delay time for the non-faulty servers to serve the target PC system; Regard the non-faulty servers with a target delay time less than the delay allowance value as the target servers.
6. A server cluster load balancing method based on the array distribution of a PC farm according to claim 5, characterized in that, The steps for forming an intermittent transmission scheme for the target PC system based on the idle transmission volume of the target servers are as follows: Summarize at least one target PC system served by the target servers to form a target PC system set; Accumulate the information transmission requirements of the target PC systems in the target PC system set to obtain the characteristic transmission requirement; Obtain the ratio of the information transmission requirement of the target PC system to the characteristic transmission requirement as the characteristic ratio; Divide the idle transmission volume of the target servers according to the characteristic ratio to obtain at least one idle partial transmission volume. Pair the idle partial transmission volume with a proportion equal to the characteristic ratio in the idle transmission volume to the target PC system corresponding to the characteristic ratio, and the idle partial transmission volume transmits information to the paired target PC system; Summarize at least one target server serving the target PC system to form a target server set; Accumulate and superimpose the idle transmission volumes of the target servers in the target server set to obtain the target idle transmission volume; Obtain the ratio of the idle transmission volume of the target server in the target server set to the target idle transmission volume as the target ratio; Divide the information transmission requirement of the target PC system according to the target ratio to obtain at least one partial transmission requirement. Pair the partial transmission requirement with the target server, such that the target ratio corresponding to the partial transmission requirement is equal to the target ratio corresponding to the paired target server; The idle partial transmission volume of the target server transmits to the partial transmission requirement paired to the target server in the target PC system; Use the idle partial transmission volume of the target server to perform intermittent transmission of the partial transmission requirement.
7. A server cluster load balancing method based on the array distribution of a PC farm according to claim 6, characterized in that The steps for analyzing the PC system served by the target server to obtain the reserved time are as follows: Subtract the actual delay time of the PC system originally served by the target server from the delay allowance value to obtain the reserved time.
8. A server cluster load balancing method based on the array distribution of a PC farm according to claim 7, characterized in that, The steps for controlling the transmission of the target server in the sample time interval using the reserved time and the intermittent transmission scheme are as follows: Obtain the target transmission time for the target server to transmit information for data with a size equal to the idle partial transmission volume in the target PC system. Divide the delay allowance value by the target transmission time to obtain the characteristic number; Divide the partial transmission requirement according to the size of the idle partial transmission volume to obtain at least one transmission requirement block. Randomly select the number of transmission requirement blocks equal to the characteristic number as the necessary transmission requirement blocks, and regard the remaining transmission requirement blocks as non-necessary transmission requirement blocks; In each sample time interval, the idle partial transmission volume of the target server sequentially transmits the necessary transmission requirement blocks; Within each sample time interval, the sample time interval is divided into a front interval and a back interval. The length of the front interval is equal to the reserved time. In the front interval, the idle transmission volume of the target server blocks the non-essential transmission demand volume of the target PC system. In the back interval, the target server transmits information to the original service PC system.
9. A server cluster load balancing method based on the array distribution of a PC farm according to claim 8, characterized in that The step of evenly distributing the feature demand volume to servers other than the target server for transmission includes the following steps: Take the services other than the target server as non-target servers, and obtain a part of the feature demand volume as the part to be measured. Each part to be measured is respectively smaller than the idle transmission volume of the non-target server; Take the transmission time of the non-target server for the part to be measured as the calibration time, and sort the non-target servers in ascending order according to the calibration time to obtain a non-target server sequence; Generate an i-sequence, where the i-sequence is a sequence composed of the first i non-target servers in the non-target server sequence in sequence. The value range of i is from 1 to the eigenvalue, and the eigenvalue is the number of non-target servers in the non-target server sequence; When the superimposed value of the idle transmission volumes of the non-target servers in the i-sequence is not less than the feature demand volume, then take the i-sequence as the target i-sequence; Use the non-target servers in the target i-sequence with the shortest length to transmit the feature demand volume.