Dynamic Load Balancing and Maintenance Method for Information Systems Based on Cloud Computing

Abstract

The present invention discloses a method for dynamic load balancing and maintenance of an information system based on cloud computing, belonging to the field of network communication technology, including S10: establishing a Web server cluster, presetting the Web servers included in the Web server cluster to at least 100, collecting access data of users accessing a preset Web server, where the access data includes the browsing data of the users; marking the collected users as reference users; S20: constructing a database based on the browsing data and analyzing the browsing characteristics of different users in the database; The present invention utilizes intelligent congestion degree evaluation and user request allocation, calculates the congestion degree of the Web server in real time, and dynamically allocates user requests according to the congestion degree, which can effectively balance the load of each server, thereby improving the performance and response speed of the entire system, realizing the efficient operation and stable maintenance of the system, and providing strong technical support for the information system in the cloud computing environment.

Description

Technical Field

[0001] The present invention relates to the field of network communication technologies, and particularly to a method for dynamic load balancing and maintenance of information systems based on cloud computing. Background Art

[0002] Load balancing is a key network technology in the cloud computing environment, aiming to optimize resource allocation, improve system performance and availability. Through load balancing, network traffic can be distributed among multiple servers or computing nodes to ensure that each node can process requests evenly, thereby avoiding single points of failure and improving the overall performance of the system.

[0003] Load balancing technology mainly relies on load balancing algorithms and load balancers. Among them, the load balancing algorithm determines how traffic is distributed among multiple servers. Common algorithms include Round Robin, Weighted Round Robin, and Least Connections, etc.; while the load balancer is responsible for forwarding requests to the appropriate server according to the algorithm.

[0004] Load balancing of information systems plays an important role in multiple industries. Among them, in the Internet and e-commerce, in a Web server cluster, load balancing is used to distribute requests from users to different servers to ensure that the load of each server is relatively balanced, improving the performance and availability of the entire system. For example, during a promotional event of an e-commerce website, user requests are distributed to multiple Web servers through load balancing to avoid overloading a single server and ensure the stable operation of the website.

[0005] In real life, the user access volume of e-commerce websites is usually extremely large during promotional events, especially during the peak hours of promotional events, when the user access volume will rise sharply. However, the existing technologies lack the ability to predict the access volume and / or access duration at the front end when facing a large user access volume, resulting in difficulties in managing and maintaining multiple Web server nodes as the scale of the access volume expands. Especially when dynamically expanding and reducing resources, it is difficult to dynamically adjust the request distribution strategy according to the current load / congestion situation, resulting in a large gap in the load / congestion of some Web servers. When processing large-scale data and analysis tasks, the data processing speed and efficiency cannot be improved, thereby affecting the content loading speed and user experience. Summary of the Invention

[0006] In view of the above problems existing in the current field of network communication technologies, the present invention is proposed.

[0007] Therefore, one of the objectives of the present invention is to provide a method for dynamic load balancing and maintenance of an information system based on cloud computing. By using intelligent congestion degree evaluation and user request allocation, it can effectively balance the load of each server by calculating the congestion degree of the Web server in real time and dynamically allocating user requests according to the congestion degree, thereby improving the performance and response speed of the entire system, achieving the efficient operation and stable maintenance of the system, and providing strong technical support for the information system in the cloud computing environment.

[0008] To solve the above technical problems, the present invention provides the following technical solutions:

[0009] A method for dynamic load balancing and maintenance of an information system based on cloud computing, comprising the following steps:

[0010] S10: Establish a Web server cluster, preset the Web servers included in the Web server cluster to at least 100, collect the access data of users accessing the preset Web servers, where the access data includes the browsing data of users; mark the collected users as reference users;

[0011] S20: Build a database based on the browsing data and analyze the browsing characteristics of different users in the database; the browsing characteristics include page browsing behavior, commodity browsing characteristics, interaction behavior, shopping cart and purchase behavior, device and network characteristics, and time characteristics;

[0012] S30: Calculate the congestion degree of each of the preset Web servers in the Web server cluster in real time, including calculating in the manner of performance index analysis, where the performance index analysis includes response time, throughput, concurrent user number, and error rate; and divide the preset Web servers into low-congestion Web servers, medium-congestion Web servers, and high-congestion Web servers according to the calculation results;

[0013] S40: Divide users into browsing users, purchasing users, and searching users based on the congestion degree caused by the access data of the reference users to the preset Web servers; among them, mark the browsing users as users with a high congestion degree; mark the purchasing users as users with a medium congestion degree; mark the searching users as users with a low congestion degree; according to the calculation results of the congestion degree of the preset Web servers, when the reference users access the Web server cluster in the future period, allocate Web servers to them based on the divided types of the reference users;

[0014] S50: updating the congestion level of the preset Web server based on the allocation result, and updating the division result of the preset Web server according to the congestion level; when a user other than the reference user accesses the Web server cluster, the user is allocated to a Web server with a low congestion level according to the updated division result of the preset Web server;

[0015] S60: monitoring and early warning management are performed on the preset Web servers in the Web server cluster, and based on the result of the update division of the preset Web servers, the congestion change of the Web servers with low congestion levels is calculated with a calculation cycle of every ten seconds, and when the load change shows an upward trend within three consecutive calculation cycles, a network congestion warning is issued to users who newly access the Web server cluster.

[0016] As a preferred solution of the present invention, wherein: in said S20, said page browsing behavior includes browsing depth, browsing time and page jump path;

[0017] The product browsing characteristics include the number of product browsing times, product browsing time, and product category browsing;

[0018] The interactive behaviors include click behaviors, search behaviors, and comment and evaluation browsing;

[0019] The shopping cart and purchase behavior include shopping cart addition behavior, shopping cart residence time, and purchase conversion rate;

[0020] The device and network characteristics include access device type and network environment;

[0021] The time characteristics include access time and access frequency.

[0022] As a preferred solution of the present invention, wherein: in S30, the response time includes the first byte time and the page loading time; wherein the first byte time includes the time from the user initiating the request to the receipt of the first byte returned by the server; the page loading time includes the time from the start of page loading to the completion of all resources being fully loaded;

[0023] The throughput includes the number of requests processed by the server per unit time;

[0024] The number of concurrent users includes the number of user requests processed simultaneously by the server;

[0025] The error rate includes the proportion of error responses returned by the server;

[0026] Among them, the response time is preset based on the first byte time and the page loading time, the response time corresponding to the first byte time is ≤300 milliseconds; the response time corresponding to the page loading time is ≤4 seconds; if the time from the user initiating the request to the receipt of the first byte returned by the server and / or the time from the start of page loading to the completion of all resources loading exceeds the preset response time, it is determined that the corresponding preset Web server processes the request slowly and is in a congested operating state; otherwise, no determination is made.

[0027] As a preferred solution of the present invention, in said S40, said allocation method includes allocating said browsing users to a Web server with a low congestion level; allocating said purchasing users to a Web server with a medium congestion level; and allocating said searching users to a Web server with a high congestion level.

[0028] As a preferred solution of the present invention, wherein: based on the time characteristics, the reference users are divided into low-access users, medium-access users and high-access users; the low-access users, medium-access users and high-access users correspond to low access time and frequency, medium access time and frequency, and high access time and frequency respectively; when the high-access users access the Web server cluster in the future period, they are assigned to the low-congestion Web server; the medium-access users are assigned to the medium-congestion Web server; and the low-access users are assigned to the high-congestion Web server.

[0029] As a preferred solution of the present invention, among the reference users, the proportion of the number of high-access users to the number of reference users is calculated, and the records of the high-access users' historical access to the Web server cluster are collected; the 100 to 200 accesses with the longest access time are intercepted from the records, and the time pattern of the access is analyzed, and according to the change of the congestion level of the low-congestion Web server, the number of high-access users who access the Web server cluster in the future time period is calculated with each hour as the statistical period, and the time required for the low-congestion Web server to change into the medium-congestion Web server is calculated according to the number of users, and the result is calculated according to the following formula:

[0030] ;in, Indicates When a high-access user accesses the Web server cluster, the Web server with low congestion level The congestion change data bit / s collected each time;

[0031] In the formula, Indicates that based on the proportion of the number of high-access users to the number of reference users, The sum of the data on congestion changes calculated in the proportion change of the second calculation; It represents the average number of high - access users who access the Web server cluster per hour in the statistics;

[0032] According to the result of the required duration calculated, mark the result as the reference duration; if the remaining duration of the future period is less than half of the reference duration, then issue a network congestion warning to the high - access users who newly access the Web server cluster.

[0033] As a preferred solution of the present invention, wherein: if the remaining duration of the future period is not less than half of the reference duration, obtain the 10 access times with the largest proportion of high - access user numbers in the analyzed time pattern, sort the access times according to the number of high - access users, and analyze the intervals between different access times; and predict the sum of the congestion change data bit / s that will increase according to the number of high - access users at different access times; based on the sum of the predicted data, if a certain future access time will cause the remaining duration to be less than half of the reference duration due to it, then issue a network congestion warning to the high - access users who access the Web server cluster at the corresponding access time.

[0034] As a preferred solution of the present invention, wherein: divide the 10 access times into several evaluation indicators, in the time pattern, calculate the weight of each evaluation indicator in the 100 - 200 accesses with the most intercepted access times, and formulate a hierarchical maintenance plan corresponding to different levels according to the calculation results; the hierarchical maintenance plan includes dividing several evaluation indicators into 5 evaluation indicators with high weight ratios and 5 evaluation indicators with low weight ratios; among the 5 evaluation indicators with high weight ratios, allocate the high - access users to the Web servers with low congestion levels; while among the 5 evaluation indicators with low weight ratios, allocate the high - access users to the Web servers with medium congestion levels, and at the same time restrict the low - access users from accessing the Web server cluster.

[0035] A computer device, characterized in that it includes a processor, an input interface, an output interface, and a memory, the processor, input interface, output interface, and memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute a method for dynamic load balancing and maintenance of an information system based on cloud computing.

[0036] A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, the computer program includes program instructions, and the program instructions, when executed by a processor, cause the processor to execute a method for dynamic load balancing and maintenance of an information system based on cloud computing.

[0037] Beneficial effects:

[0038] 1. By calculating the congestion degree of Web servers in real time and dynamically allocating user requests according to the congestion degree, it is possible to effectively balance the loads of each Web server, avoid overloading of a single Web server, and thus improve the performance and response speed of the entire system.

[0039] 2. Based on user types and access characteristics, refined server allocation is carried out, and different types of user requests are allocated to Web servers with corresponding congestion degrees, realizing reasonable allocation and optimized utilization of resources, and improving the utilization rate of Web server resources.

[0040] 3. By allocating high-access users to Web servers with low congestion degrees, it ensures that users can obtain faster response speeds and better browsing experiences. Especially during peak access periods, it can effectively alleviate the problem of long user waiting times.

[0041] 4. Real-time monitoring and early warning management of Web servers are carried out. When an abnormal change in the congestion degree of a Web server is detected, an early warning can be issued in a timely manner and corresponding measures can be taken, enhancing the stability and reliability of the system, and reducing the risk of service interruption caused by Web server overload.

[0042] 5. The system can dynamically adjust the allocation strategy of Web servers and the division of congestion degrees according to the real-time congestion degree and user access conditions, has good self-adaptability and flexibility, and can cope with changing network environments and user requirements.

[0043] 6. Through an automated congestion degree calculation and user allocation mechanism, the workload of manual intervention and maintenance is reduced, the maintenance difficulty and operation cost of the system are lowered, and the maintenance efficiency is improved. Description of the drawings

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. Among them:

[0045] Figure 1 It is a schematic flowchart of the method of the embodiment of the present invention;

[0046] Figure 2 It is a schematic flowchart of the embodiment of the present invention. Specific implementation manners

[0047] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present invention fall within the scope of protection of the present invention.

[0048] Due to the deficiency of the prior art in predicting the access volume and / or access duration at the front end in the face of a huge user access volume, it becomes difficult to manage and maintain multiple Web server nodes as the scale of the access volume expands. Especially when dynamically expanding and reducing resources, it is difficult to dynamically adjust the request allocation strategy according to the current load / congestion situation, resulting in a large gap in the load / congestion of some Web servers. When processing large-scale data and analysis tasks, it is impossible to improve the data processing speed and efficiency, thus affecting the content loading speed and user experience.

[0049] Based on this, the present invention proposes a method for dynamic load balancing and maintenance of an information system based on cloud computing. By using intelligent congestion degree evaluation and user request allocation, it can effectively balance the load of each server by calculating the congestion degree of the Web server in real time and dynamically allocating user requests according to the congestion degree, thereby improving the performance and response speed of the entire system, realizing the efficient operation and stable maintenance of the system, and providing strong technical support for the information system in the cloud computing environment.

[0050] The following further specifically describes this solution through embodiments and with reference to the drawings.

[0051] Referring to Figures 1 to 2 , which is an embodiment of the present invention. This embodiment provides a method for dynamic load balancing and maintenance of an information system based on cloud computing, including the following steps:

[0052] S10: Establish a Web server cluster, preset the Web servers included in the Web server cluster to at least 100, collect the access data of users accessing the preset Web servers, where the access data includes the browsing data of users; mark the collected users as reference users;

[0053] It should be noted that the access data collected includes at least the access data within one year;

[0054] S20: Build a database based on browsing data and analyze the browsing characteristics of different users in the database; the browsing characteristics include page browsing behavior, product browsing characteristics, interaction behavior, shopping cart and purchase behavior, device and network characteristics, and time characteristics.

[0055] It should be noted in this embodiment that in S20, the page browsing behavior includes browsing depth, browsing duration, and page jump path.

[0056] It should be noted that the browsing depth includes the number of pages accessed by the user; users with a higher depth may be more interested in the website content and will browse multiple product detail pages, category pages, etc.

[0057] The browsing duration refers to the time the user stays on the website. A longer browsing duration indicates that the user has a higher degree of attention to the website content and may be carefully selecting products or reading relevant information.

[0058] The page jump path refers to the order of pages accessed by the user on the website; for example, the user may jump from the home page to the category page, then to the product detail page, and finally enter the shopping cart page. Analyzing the jump path helps to understand the user's shopping intentions and behavior habits.

[0059] The product browsing characteristics include the number of product views, product browsing duration, and product category browsing.

[0060] It should be noted that the number of product views is the frequency of a user's view of a certain product. Products with a high number of views may be potential purchase targets or popular products of the user.

[0061] The product browsing duration is the time the user stays on the product detail page; a longer browsing duration may mean that the user has deeply viewed information such as product details and reviews.

[0062] Product category browsing is the behavior of the user browsing different product categories; for example, the user may first browse the clothing category and then the electronic product category, which helps to understand the user's interest range and shopping needs.

[0063] The interaction behavior includes click behavior, search behavior, and comment and review browsing.

[0064] It should be noted that the click behavior refers to the click operations of the user on the website, including clicking on products, categories, search boxes, navigation bars, etc. The click behavior can reflect the user's focus points and operation habits.

[0065] The search behavior refers to the frequency of the user using the website search function and the search keywords; the search behavior reveals the user's shopping intentions and the demand for specific products.

[0066] Review and evaluation browsing refers to the behavior of users browsing product reviews and evaluations, which indicates that users will refer to other consumers' feedback and experience before making a purchase decision;

[0067] Shopping cart and purchase behavior include shopping cart addition behavior, shopping cart dwelling time and purchase conversion rate;

[0068] Device and network characteristics include access device type and network environment;

[0069] It should be noted that users may have different browsing habits and needs based on the type of device they use, such as mobile phones, tablets, PCs, etc.

[0070] The user's network connection speed and stability, network environment affects page loading speed and user experience, which in turn affects the user's browsing behavior;

[0071] Temporal characteristics include visit time and visit frequency;

[0072] It should be noted that the access time is the time period when the user visits the website; for example, the user may visit the website in the evening on weekdays or during the day on weekends, which helps to understand the user's shopping time preference;

[0073] The access frequency is the number of times a user visits a website within a certain period of time. A high access frequency indicates that the user has a high loyalty and demand for the website.

[0074] This embodiment further distinguishes reference users into low-access users, medium-access users and high-access users based on time characteristics; low-access users, medium-access users and high-access users correspond to low access time and frequency, medium access time and frequency, and high access time and frequency respectively; when high-access users access the Web server cluster in the future period, they are allocated to Web servers with low congestion; medium-access users are allocated to Web servers with medium congestion; and low-access users are allocated to Web servers with high congestion;

[0075] S30: calculating the congestion degree of each preset Web server in the Web server cluster in real time, including calculating in accordance with a performance indicator analysis method, wherein the performance indicator analysis includes response time, throughput, number of concurrent users, and error rate; and dividing the preset Web servers into low congestion degree Web servers, medium congestion degree Web servers, and high congestion degree Web servers according to the calculation results;

[0076] Specifically, in S30, the response time includes the first byte time and the page load time; wherein the first byte time includes the time from when the user initiates the request to when the first byte returned by the server is received; the page load time includes the time from when the page starts loading to when all resources are fully loaded;

[0077] Throughput includes the number of requests processed by the server per unit time;

[0078] The number of concurrent users includes the number of user requests processed by the server at the same time;

[0079] The error rate includes the proportion of error responses returned by the server;

[0080] The response time is preset based on the first byte time and the page loading time. The response time corresponding to the first byte time is ≤300 milliseconds; the response time corresponding to the page loading time is ≤4 seconds. If the time from the user initiating the request to the receipt of the first byte returned by the server and / or the time from the start of page loading to the completion of all resources loading exceeds the preset response time, it is determined that the corresponding preset Web server processes the request slowly and is in a congested operation state; otherwise, no determination is made.

[0081] S40: Based on the congestion degree caused by the access data of the reference user to the preset Web server, the users are divided into browsing users, purchasing users and searching users; wherein the browsing users are marked as users causing a high congestion degree; the purchasing users are marked as users causing a medium congestion degree; and the searching users are marked as users causing a low congestion degree; according to the calculation result of the congestion degree of the preset Web server, when the reference user accesses the Web server cluster in the future period, a Web server is allocated to the reference user based on the classification type of the reference user;

[0082] It should be emphasized in this embodiment that in S40, the allocation method includes allocating browsing users to Web servers with low congestion levels; allocating purchasing users to Web servers with medium congestion levels; and allocating search users to Web servers with high congestion levels.

[0083] It should be noted in this embodiment that browsing users mainly browse products and information on the website, but have a low willingness to buy. They may be interested in multiple products and spend more time browsing product detail pages, evaluation pages, etc., but rarely add products to shopping carts or make purchases. Browsing users may be in the information collection stage, comparing and researching products, but have not yet made a purchase decision, so they are assigned to Web servers with low congestion levels;

[0084] Purchase-type users have clear purchase intentions. When browsing products, they pay more attention to information such as product prices, specifications, and inventory. They will add products to the shopping cart and finally complete the purchase. The browsing paths of such users are usually relatively direct, quickly jumping from the product category page or search result page to the product details page, and then entering the shopping cart and settlement page. The browsing duration of purchase-type users is relatively short, but the conversion rate is relatively high. Therefore, they are assigned to Web servers with medium congestion levels.

[0085] Search-type users search for specific products or information through the website's search function. They usually enter keywords in the search box and then browse the products on the search result page. The browsing behavior of such users is relatively concentrated, mainly focusing on products related to the search keywords and browsing fewer other products. Search-type users may have clear requirements for brands, functions, or prices and will screen and compare according to the search results. Therefore, they are assigned to Web servers with high congestion levels.

[0086] S50: Update the congestion level of the preset Web server based on the allocation result, and update the division result of the preset Web server according to the congestion level; when users other than the reference users access the Web server cluster, assign the users to the Web servers with low congestion levels according to the updated division result of the preset Web server.

[0087] S60: Monitor and manage early warnings for the preset Web servers in the Web server cluster, and calculate the congestion change of the Web servers with low congestion levels every ten seconds based on the Web servers with low congestion levels according to the updated division result of the preset Web server. When the load change shows an upward trend in 3 consecutive calculation cycles, issue a network congestion warning to the users newly accessing the Web server cluster.

[0088] This embodiment further includes: calculating the proportion of the number of high-access users in the reference users among the total number of reference users, and collecting the records of the historical access of the high-access users to the Web server cluster; intercepting 100 to 200 accesses with the most access times from the records, analyzing their access time patterns, and calculating the number of high-access users accessing the Web server cluster in the future period every hour according to the congestion level change of the Web servers with low congestion levels, and calculating the time required for the Web servers with low congestion levels to change into Web servers with medium congestion levels according to the following formula:

[0089] ; where represents the th high-access user's access to the Web server cluster to the Congestion change data collected for the first time bit / s;

[0090] Wherein, Represents the sum of the th calculated proportion change based on the proportion of the high-access user number to the reference user number, and the th sum of congestion change data; Represents the average number of high-access users accessing the Web server cluster per hour statistically;

[0091] According to the result of the calculated required duration, mark the result as the reference duration; if the remaining duration of the future period is less than half of the reference duration, then issue a network congestion warning to the high-access users newly accessing the Web server cluster;

[0092] Based on the above, in this embodiment, if the remaining duration of the future period is not less than half of the reference duration, obtain the 10 access times with the largest proportion of high-access user numbers in the analyzed time pattern, sort the access times according to the number of high-access users, and analyze the intervals between different access times; and predict the sum of the congestion change data bit / s that will increase based on the number of high-access users at different access times; based on the predicted sum of data, if a certain future access time will cause the remaining duration to be less than half of the reference duration due to it, then issue a network congestion warning to the high-access users accessing the Web server cluster at the corresponding access time;

[0093] Further in this embodiment, divide the 10 access times into several evaluation indicators, calculate the weight of each evaluation indicator in the 100 - 200 accesses with the most intercepted access times in the time pattern, and formulate a hierarchical maintenance plan corresponding to different levels according to the calculation results; the hierarchical maintenance plan includes dividing several evaluation indicators into 5 evaluation indicators with high weight ratios and 5 evaluation indicators with low weight ratios; among the 5 evaluation indicators with high weight ratios, allocate high-access users to Web servers with low congestion levels; while among the 5 evaluation indicators with low weight ratios, allocate high-access users to Web servers with medium congestion levels, and at the same time restrict low-access users from accessing the Web server cluster.

[0094] A computer device, characterized in that it includes a processor, an input interface, an output interface, and a memory, and the processor, input interface, output interface, and memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method for dynamic load balancing and maintenance of an information system based on cloud computing.

[0095] A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, the computer program includes program instructions, and the program instructions cause a processor to execute a method for dynamic load balancing and maintenance of an information system based on cloud computing when executed by the processor.

[0096] In summary, the present invention utilizes intelligent congestion degree evaluation and user request allocation. By calculating the congestion degree of the Web server in real time and dynamically allocating user requests according to the congestion degree, it can effectively balance the loads of each server, thereby improving the performance and response speed of the entire system, realizing the efficient operation and stable maintenance of the system, and providing strong technical support for the information system in the cloud computing environment.

[0097] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for dynamic load balancing and maintenance of information systems based on cloud computing, characterized in that: The following steps are involved: S10: Establishing a Web server cluster, presetting the number of Web servers included in the Web server cluster to at least 100, collecting access data of users accessing the preset Web servers, wherein the access data includes browsing data of the users; and marking the collected users as reference users; S20: Building a database based on the browsing data, and analyzing browsing characteristics of different users in the database; the browsing characteristics include page browsing behavior, product browsing characteristics, interactive behavior, shopping cart and purchase behavior, device and network characteristics, and time characteristics; S30: calculating the congestion degree of each of the preset Web servers in the Web server cluster in real time, including calculating in a manner of performance indicator analysis, wherein the performance indicator analysis includes response time, throughput, number of concurrent users, and error rate; and dividing the preset Web servers into low congestion degree Web servers, medium congestion degree Web servers, and high congestion degree Web servers according to the calculation results; S40: Based on the congestion degree caused by the access data of the reference user to the preset Web server, the users are divided into browsing users, purchasing users and searching users; wherein the browsing users are marked as users causing a high congestion degree; the purchasing users are marked as users causing a medium congestion degree; and the searching users are marked as users causing a low congestion degree; according to the calculation result of the congestion degree of the preset Web server, when the reference user visits the Web server cluster in the future period, a Web server is allocated to the reference user based on the classification type of the reference user; The allocation method includes allocating the browsing users to a Web server with a low congestion level; allocating the purchasing users to a Web server with a medium congestion level; and allocating the searching users to a Web server with a high congestion level. Or based on the time feature, the reference users are divided into low-access users, medium-access users and high-access users; the low-access users, medium-access users and high-access users correspond to low access time and frequency, medium access time and frequency, and high access time and frequency respectively; when the high-access users access the Web server cluster in the future period, they are allocated to the low-congestion Web server; the medium-access users are allocated to the medium-congestion Web server; and the low-access users are allocated to the high-congestion Web server; S50: updating the congestion level of the preset Web server based on the allocation result, and updating the division result of the preset Web server according to the congestion level; when a user other than the reference user accesses the Web server cluster, the user is allocated to a Web server with a low congestion level according to the updated division result of the preset Web server; S60: monitoring and early warning management are performed on the preset Web servers in the Web server cluster, and based on the result of the update division of the preset Web servers, the congestion change of the Web servers with low congestion levels is calculated with a calculation cycle of every ten seconds, and when the load change shows an upward trend within three consecutive calculation cycles, a network congestion warning is issued to users who newly access the Web server cluster.

2. The method for dynamic load balancing and maintenance of information systems based on cloud computing according to claim 1, characterized in that: In S20, the page browsing behavior includes browsing depth, browsing time and page jump path; The product browsing characteristics include the number of product browsing times, product browsing time, and product category browsing; The interactive behaviors include click behaviors, search behaviors, and comment and evaluation browsing; The shopping cart and purchase behavior include shopping cart addition behavior, shopping cart residence time, and purchase conversion rate; The device and network characteristics include access device type and network environment; The time characteristics include access time and access frequency.

3. The method for dynamic load balancing and maintenance of information systems based on cloud computing according to claim 1, characterized in that: In S30, the response time includes the first byte time and the page loading time; wherein the first byte time includes the time from the user initiating the request to the receipt of the first byte returned by the server; the page loading time includes the time from the start of page loading to the completion of all resources loading; The throughput includes the number of requests processed by the server per unit time; The number of concurrent users includes the number of user requests processed simultaneously by the server; The error rate includes the proportion of error responses returned by the server; Among them, the response time is preset based on the first byte time and the page loading time, the response time corresponding to the first byte time is ≤300 milliseconds; the response time corresponding to the page loading time is ≤4 seconds; if the time from the user initiating the request to the receipt of the first byte returned by the server and / or the time from the start of page loading to the completion of all resources loading exceeds the preset response time, it is determined that the corresponding preset Web server processes the request slowly and is in a congested operating state; otherwise, no determination is made.

4. The method for dynamic load balancing and maintenance of information systems based on cloud computing according to claim 1, characterized in that: Among the reference users, the proportion of the number of high-access users to the number of reference users is calculated, and the records of the high-access users' historical access to the Web server cluster are collected; the 100 to 200 visits with the longest access time are intercepted from the records, and the time pattern of the access is analyzed, and according to the change of the congestion level of the low-congestion Web server, the number of high-access users who access the Web server cluster in the future time period is calculated with each hour as the statistical period, and the time required for the low-congestion Web server to change into the medium-congestion Web server is calculated according to the number of users, and the result is calculated according to the following formula: ;in, Indicates When a high-access user accesses the Web server cluster, the low-congestion Web server The congestion change data bit / s collected each time; In the formula, Indicates that based on the proportion of the number of high-access users to the number of reference users, The calculated weight change is calculated in the The sum of the congestion change data; Indicates the average number of high-access users accessing the Web server cluster per hour; According to the result of the calculated required duration, the result is marked as a reference duration; if the remaining duration of the future period is less than half of the reference duration, a network congestion warning is issued to high-access users who newly access the Web server cluster.

5. The method for dynamic load balancing and maintenance of information systems based on cloud computing according to claim 4, characterized in that: If the remaining duration of the future time period is not less than half of the reference duration, the 10 access times with the largest number of high-access users are obtained from the analyzed time pattern, and the access times are sorted according to the number of high-access users, and the intervals between different access times are analyzed; and the sum of the increased congestion change data bit / s is predicted based on the number of high-access users at different access times; based on the sum of the predicted data, if a certain access time in the future will cause the remaining duration to be less than half of the reference duration, a network congestion warning will be issued to high-access users accessing the Web server cluster when the corresponding access time arrives.

6. The method for dynamic load balancing and maintenance of information systems based on cloud computing according to claim 5, characterized in that: The 10 access times are divided into a number of evaluation indicators, and in the time pattern, the weight of each evaluation indicator is calculated according to the intercepted 100 to 200 visits with the most access times, and a hierarchical maintenance plan corresponding to different levels is formulated according to the calculation results; the hierarchical maintenance plan includes dividing the several evaluation indicators into 5 evaluation indicators with high weight ratios and 5 evaluation indicators with low weight ratios; among the 5 evaluation indicators with high weight ratios, the high-access users are allocated to the Web server with low congestion level; Among the five evaluation indicators with low weights, the high-access users are allocated to the Web servers with medium congestion levels, while the low-access users are restricted from accessing the Web server cluster.

7. A computer device, characterized in that: The method comprises a processor, an input interface, an output interface and a memory, wherein the processor, the input interface, the output interface and the memory are interconnected, wherein the memory is used to store a computer program, the computer program comprises program instructions, and the processor is configured to call the program instructions to execute the method according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is enabled to perform the method according to any one of claims 1 to 6.

Images