Distributed service traffic scheduling method, device, equipment, medium and program product

Abstract

The application provides a distributed service traffic scheduling method, device, equipment, medium and program product. The method comprises the following steps: determining a group position feature of a user based on obtained multi-dimensional network index data of the user and a plurality of multi-dimensional network index historical data; determining a group performance feature of the user based on the plurality of multi-dimensional network index historical data; selecting a plurality of target scheduling models based on the group position feature and the group performance feature; obtaining a target service traffic scheduling baseline through the plurality of target scheduling models; and dynamically adjusting the traffic scheduling of distributed services based on the target service traffic scheduling baseline. The distributed service traffic scheduling method provided by the application is based on comprehensive multi-dimensional network index data, and the plurality of target scheduling models are used for comprehensive scheduling to obtain an accurate target service traffic scheduling baseline, so that the traffic scheduling of distributed services is dynamically adjusted, and the accuracy of the distributed service traffic scheduling is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of service scheduling, and in particular to a distributed service traffic scheduling method, device, equipment, medium and program product. BACKGROUND

[0002] Enterprise business system refers to a sub-system of an enterprise that pushes its products to the market and maximizes its profits. In terms of modules, the business system includes marketing planning, sales platform, sales process management, customer service management, customer relationship management, and risk prevention. Among them, users can interact with the enterprise business system through the user terminal according to their needs, and the smoothness of user access directly affects user experience and business efficiency of the enterprise. In the user access process, the enterprise business system manages and allocates the received user requests through traffic scheduling. Traffic scheduling includes load balancing, data synchronization, service degradation, and other strategies and technologies, with the purpose of ensuring that the system can efficiently and stably handle a large number of concurrent requests.

[0003] In the prior art, user access and traffic scheduling are mostly based on static rules for one-time matching and segmentation. According to the geographical location of the user, traffic is allocated to different server clusters, for example, user traffic in the eastern region is directed to the data center in the east, and user traffic in the western region is directed to the data center in the west. This is a static rule because the geographical location of the user is relatively fixed and does not change frequently. However, as more and more enterprise business systems adopt distributed and cloud deployment, it is difficult for static scheduling algorithms to more reasonably schedule business traffic, ultimately affecting the user's access experience. SUMMARY

[0004] The present application provides a distributed service traffic scheduling method, device, equipment, medium and program product to solve the problem of inaccurate distributed service traffic scheduling in the prior art.

[0005] In a first aspect, the present application provides a distributed service traffic scheduling method, comprising:

[0006] Obtaining real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at multiple historical time points;

[0007] Based on the multi-dimensional network index data and the multiple multi-dimensional network index historical data, determining a group location feature of the user; the group location feature is used to reflect the characteristic performance of the network location where the user is located;

[0008] Based on the multiple multi-dimensional network index historical data, determining a group performance feature of the user; the group performance feature is used to reflect the characteristic performance of the user to the network performance;

[0009] select a plurality of target scheduling models from a plurality of types of scheduling models based on the group location feature and the group performance feature; any of the scheduling models is obtained based on model training of multidimensional network index samples and corresponding traffic scheduling baseline labels;

[0010] based on the multidimensional network index data, training the plurality of target scheduling models respectively to obtain a target traffic scheduling baseline;

[0011] based on the target traffic scheduling baseline, performing dynamic adjustment of traffic scheduling for distributed services.

[0012] In one embodiment, after obtaining real-time multidimensional network index data of a user and historical multidimensional network index data of the user at a plurality of historical time points, and before determining a group location feature of the user based on the multidimensional network index data and a plurality of the historical multidimensional network index data, the following steps are performed for the multidimensional network index data:

[0013] determining whether each dimension network index data satisfies a preset weight update trigger rule corresponding to the dimension network index data;

[0014] if the dimension network index data satisfies the preset weight update trigger rule corresponding to the dimension network index data, increasing a preset weight value corresponding to the dimension network index data to obtain a target weight value corresponding to each dimension network index data;

[0015] according to the target weight value corresponding to each dimension network index data, sorting the multidimensional network index data according to weight values to obtain a multidimensional network index sequence.

[0016] In one embodiment, the target scheduling model includes a first target scheduling model and a second target scheduling model; the selecting of the plurality of target scheduling models from the plurality of types of scheduling models based on the group location feature and the group performance feature includes:

[0017] determining a first performance indicator of each type of scheduling model on the group location feature and a second performance indicator of each type of scheduling model on the group performance feature;

[0018] determining a scheduling model corresponding to an optimal performance indicator in the plurality of first performance indicators as the first target scheduling model;

[0019] determining a scheduling model corresponding to an optimal performance indicator in the plurality of second performance indicators as the second target scheduling model.

[0020] In one embodiment, the target traffic scheduling baseline is obtained by training the plurality of target scheduling models based on the multi-dimensional network indicator data, comprising:

[0021] The multi-dimensional network indicator sequence is subjected to feature extraction to obtain multi-dimensional network indicator representation data;

[0022] The multi-dimensional network indicator representation data is input into the first target scheduling model to obtain a first traffic scheduling baseline output by the first target scheduling model;

[0023] The multi-dimensional network indicator representation data is input into the second target scheduling model to obtain a second traffic scheduling baseline output by the second target scheduling model;

[0024] The first traffic scheduling baseline and the second traffic scheduling baseline are subjected to mean value calculation to obtain a target traffic scheduling baseline.

[0025] In one embodiment, the target traffic scheduling baseline is obtained by training the plurality of target scheduling models based on the multi-dimensional network indicator data, comprising:

[0026] The first access behavior traffic data of the user is obtained in real time;

[0027] If the first access behavior traffic data deviates from the target traffic scheduling baseline, the distributed service and user access are dynamically scheduled.

[0028] In one embodiment, after the distributed service is dynamically adjusted based on the target traffic scheduling baseline, comprising:

[0029] The second access behavior traffic data of the user is obtained in real time;

[0030] The second access behavior traffic data and the target traffic scheduling baseline are compared and analyzed to obtain a comparison and analysis result;

[0031] The step of obtaining the second access behavior traffic data of the user in real time is iteratively executed until a plurality of comparison and analysis results are obtained after a continuous period of time is executed;

[0032] Based on the plurality of comparison and analysis results, historical dynamic ring ratio data is generated;

[0033] Based on the historical dynamic ring ratio data, the strategy of distributed service traffic scheduling is optimized.

[0034] In one embodiment, the distributed service traffic scheduling method further comprises:

[0035] The service support capability data is obtained;

[0036] perform performance bottleneck analysis based on the service support capability data to obtain performance bottleneck analysis results;

[0037] determine a time point for distributed service flow scheduling based on the performance bottleneck analysis results;

[0038] perform dynamic scheduling of distributed services and user access at the time point for distributed service flow scheduling.

[0039] In a second aspect, the present application provides a distributed service flow scheduling device, comprising:

[0040] an acquisition module configured to acquire real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at a plurality of historical time points;

[0041] a first determination module configured to determine group location features of the user based on the multi-dimensional network index data and the plurality of multi-dimensional network index historical data; the group location features are used to reflect characteristic performance of a network location where the user is located;

[0042] a second determination module configured to determine group performance features of the user based on the plurality of multi-dimensional network index historical data; the group performance features are used to reflect characteristic performance of the user on network performance;

[0043] a scheduling model selection module configured to select a plurality of target scheduling models from a plurality of types of scheduling models based on the group location features and the group performance features; any of the scheduling models is obtained based on model training of multi-dimensional network index samples and corresponding service flow scheduling baseline labels;

[0044] a scheduling model prediction module configured to train the plurality of target scheduling models based on the multi-dimensional network index data to obtain target service flow scheduling baselines;

[0045] a scheduling module configured to perform dynamic adjustment of flow scheduling of distributed services based on the target service flow scheduling baselines.

[0046] In a third aspect, the present application provides a device, which comprises an electronic device, the electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the distributed service flow scheduling method according to any of the above aspects when executing the program.

[0047] In a fourth aspect, the present application also provides a medium, which comprises a non-transitory computer-readable storage medium, and a computer program is stored on the medium, and the computer program is executed by a processor to implement the steps of the distributed traffic scheduling method according to any one of the above aspects.

[0048] In a fifth aspect, the present application also provides a product, which comprises a computer program product, and the computer program product comprises a computer program, and the computer program is stored on a non-transitory computer-readable storage medium, and the computer program is executed by the processor to implement the steps of the distributed traffic scheduling method according to any one of the above aspects.

[0049] The distributed traffic scheduling method, device, equipment, medium and program product provided by the present application can obtain comprehensive multi-dimensional network index data, determine the group location characteristics and group performance characteristics of the current user by combining the multi-dimensional network index historical data, select a plurality of target scheduling models suitable for the current user from two different characteristic dimensions by comprehensively considering the group location characteristics and group performance characteristics of the current user, perform comprehensive scheduling by the plurality of target scheduling models based on the comprehensive multi-dimensional network index data, obtain a precise target traffic scheduling baseline, and then dynamically adjust the distributed traffic based on the precise target traffic scheduling baseline, thereby improving the accuracy of the distributed traffic scheduling, enhancing the early perception ability of the business situation, reducing the pressure and security risks caused by the business burst, and improving the user satisfaction. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0051] Figure 1 is one of the flowcharts of the distributed traffic scheduling method provided by the present application.

[0052] Figure 2 is another flowchart of the distributed traffic scheduling method provided by the present application.

[0053] Figure 3 is a schematic diagram of data sorting provided by the present application.

[0054] Figure 4 is a schematic diagram of scheduling model selection provided by the present application.

[0055] Figure 5is a structural schematic diagram of a distributed service traffic scheduling device provided by the present application.

[0056] Figure 6 is a structural schematic diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0057] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions in the present application will be described below in detail with reference to the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0058] The terms "first", "second", and the like in the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein.

[0059] The embodiments of the present application will be described below with reference to the drawings. Figures 1-6 The provided distributed service traffic scheduling method, device, equipment, medium, and program product of the present application are described.

[0060] It should be noted that the distributed service traffic scheduling method provided by the embodiments of the present application is realized based on a distributed service traffic scheduling device. The distributed service traffic scheduling method provided by the embodiments of the present application collects real-time multi-dimensional network index data of a user, sorts and extracts features from the real-time full-factor data, obtains corresponding representation data, further selects two scheduling models for the user based on the full-factor historical data of the user, comprehensively predicts the full-factor real-time data based on the two scheduling models, obtains a distributed service traffic scheduling baseline of the user, and performs distributed service traffic scheduling according to the distributed service traffic scheduling baseline. The distributed service traffic scheduling method provided by the embodiments of the present application can adapt to the development of distributed clouds, meet the needs of high availability and flexible and elastic deployment of services, realize fine and dynamic scheduling of user access traffic, strengthen the advance perception ability of service situation, reduce the pressure and security risks brought by service bursts, and thus improve user satisfaction.

[0061] The embodiments of the present application take a distributed service traffic scheduling device as an example to describe the distributed service traffic scheduling method.

[0062] In combination with Figure 1 and Figure 2 , Figure 1 is one of flowcharts of a distributed service traffic scheduling method provided by the present application, Figure 2Figure 2 is a flowchart of a second embodiment of the distributed service traffic scheduling method provided by the present application.

[0063] As shown in Figure 1 , the method comprises the following steps:

[0064] Step 101, acquiring real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at multiple historical time points;

[0065] Step 102, determining group location features of the user based on the multi-dimensional network index data and the multiple multi-dimensional network index historical data;

[0066] Step 103, determining group performance features of the user based on the multiple multi-dimensional network index historical data;

[0067] Step 104, selecting multiple target scheduling models from multiple types of scheduling models based on the group location features and the group performance features;

[0068] Step 105, training the multiple target scheduling models respectively based on the multi-dimensional network index data to obtain target service traffic scheduling baselines;

[0069] Step 106, dynamically adjusting distributed service traffic scheduling based on the target service traffic scheduling baselines.

[0070] When the user and the service system constitute a client-server relationship, a network connection between the user terminal and the intelligent scheduling server in the service system is established, and the user can communicate and interact with the intelligent scheduling server through the user terminal.

[0071] Specifically, as shown in Figure 2 , the distributed service traffic scheduling device acquires real-time user access traffic full-factor data of a user.

[0072] It should be noted that, since the data amount of the user access traffic full-factor data is large, a large amount of data will be generated if accessed in parallel, and therefore the user access traffic full-factor data needs to be filtered to determine data that has a greater impact on the traffic scheduling result.

[0073] Further, the distributed service traffic scheduling device performs individual retrieval on the acquired user access traffic full-factor data based on a retrieval method, filters out data that has no value for distributed service traffic scheduling, repeated data, and missing data contained in the user access traffic full-factor data, to obtain multi-dimensional network index data that has value and is useful for distributed service traffic scheduling. The purpose of this step is to filter data, remove data that has no value, is repeated, or is missing, and improve data quality.

[0074] Among them, the multi-dimensional network index data includes but is not limited to user network location information, user network delay and jitter information, application resource availability, application response rate and success rate, which are important for distributed traffic scheduling. Specifically, in the process of distributed traffic scheduling, user network location information can help the system better understand the geographical area and network topology where the user is located, which helps to balance the load, fault tolerance and optimize user experience, etc. User network delay and jitter information is one of the key factors affecting user experience and application performance. High delay and jitter can cause slow data transmission speed, long application response time, etc., thereby affecting user experience and application stability. Application resource availability is crucial to the stability and reliability of the application. If the resource is not available, the application will not run normally. Application response rate and success rate are directly related to user experience and application performance. Fast response and high success rate mean that users can use the application smoothly, while slow response and low success rate may lead to user loss or dissatisfaction.

[0075] In an embodiment, the multi-dimensional network index data such as user network location information, user network delay and jitter information, application resource availability, application response rate and success rate can be obtained in the following way.

[0076] User network location information: When the user terminal communicates with the intelligent scheduling server, the intelligent scheduling server automatically obtains the network location information of the user terminal. The network location information can be an Internet Protocol (IP) address, Global Positioning System (GPS) coordinates, base station positioning and wireless network Wi-Fi (Wireless Fidelity) positioning, etc. After the intelligent scheduling server obtains the network location information of the user terminal, the intelligent scheduling server sends a confirmation instruction to the user terminal to verify the obtained network location information of the user terminal. After the user terminal receives the network location information instruction sent by the intelligent scheduling server, the user terminal responds to the confirmation instruction and verifies the network location information of the user terminal obtained by the intelligent scheduling server. After the verification is successful, the user terminal sends a location confirmation feedback to the intelligent scheduling server. After the intelligent scheduling server receives the location confirmation feedback sent by the user terminal, the intelligent scheduling server automatically displays the network location information of the user terminal and determines the user network location information.

[0077] User network delay and jitter information: Measure the round-trip time between the user terminal and the server, evaluate the delay through the round-trip time between the user terminal and the server, and detect network jitter by continuously measuring the change of round-trip time.

[0078] Application resource availability: Use application performance management tools to monitor application resource usage, including monitoring central processing units, memory, and storage, to confirm application resource availability.

[0079] Application response rate and success rate: Use real-time monitoring tools to track application response time and business completion rate.

[0080] In addition, while obtaining the user's real-time user access traffic full-factor data, the distributed business traffic scheduling device also obtains the user's user access traffic full-factor historical data at multiple historical time points. Further, the distributed business traffic scheduling device performs data screening on the user access traffic full-factor historical data at each historical time point, filters out data that is of no value to the distributed business traffic scheduling, repeated data, and missing data contained in the user access traffic full-factor historical data, to obtain screened multi-dimensional network index historical data at multiple historical time points. Typically, the dimensions of the multi-dimensional network index historical data correspond to the dimensions of the multi-dimensional network index data.

[0081] Further, the distributed business traffic scheduling device sorts the multi-dimensional network index data based on an adaptive sorting method, places data of more value to the distributed business traffic scheduling in the front, and obtains a sorted multi-dimensional network index sequence.

[0082] Similarly, the distributed business traffic scheduling device sorts the multi-dimensional network index historical data at each historical time point based on an adaptive sorting method, obtains a sorted multi-dimensional network index historical sequence at each historical time point, and arranges the multi-dimensional network index historical sequences at each historical time point in chronological order, i.e., obtains a sequence of multiple multi-dimensional network index historical sequences, which can be referred to as a multi-dimensional network index historical sequence set.

[0083] Further, the distributed business traffic scheduling device determines the group location characteristics of the user based on the multi-dimensional network index sequence and the multi-dimensional network index historical sequence set.

[0084] It should be noted that, since the servers of the distributed business system are distributed in multiple different locations, different users are located in different locations of the servers, so that there is a difference in the scheduling delay when scheduling in a distributed manner, that is, the network location of the user can be used to determine the group location characteristics of the user in a geographical location partition manner. Therefore, first, the current location of the user can be used as the main factor of the group location characteristics of the user, and the current location of the user belongs to a certain area, which may correspond to a specific server cluster or data center, and can reflect the characteristics of the network location of the user. In addition, the historical data of the user can be further used to analyze the behavior mode and performance requirement of the user, and the analyzed behavior mode and performance requirement are used as other factors of the group location characteristics of the user. Therefore, the group location characteristics of the user can be determined by the current location of the user and the behavior mode and performance requirement of the user.

[0085] Therefore, it can be understood that the distributed business traffic scheduling device determines the current location of the user according to the multi-dimensional network index sequence, analyzes the behavior mode and performance requirement of the user according to the multi-dimensional network index historical sequence set, and further determines the group location characteristics of the user according to the current location of the user and the behavior mode and performance requirement of the user.

[0086] Further, the distributed business traffic scheduling device determines the group performance characteristics of the user based on the multi-dimensional network index historical sequence set.

[0087] It should be noted that, since the network data change of the user can change the above-mentioned sorting result, the user is divided into a high-delay-sensitive user, a high-throughput-demand user, and a stable-connection user according to the network data change corresponding to the sorting, and the types can be used to determine the group performance characteristics of the user.

[0088] Therefore, it can be understood that the distributed business traffic scheduling device determines the network data change of the user according to the multi-dimensional network index historical sequence set, further analyzes the network data change of the user, and determines the group performance characteristics of the user.

[0089] Further, the distributed business traffic scheduling device selects a target scheduling model with optimal group location characteristics from a plurality of types of scheduling models based on the group location characteristics, and selects a target scheduling model with optimal group performance characteristics from a plurality of types of scheduling models based on the group performance characteristics, so as to select the most suitable scheduling model for the user according to the historical data of the user.

[0090] The plurality of types of scheduling models include but are not limited to a time sequence prediction scheduling model, a time sequence prediction scheduling model, and a reinforcement learning scheduling model.

[0091] Further, the distributed service traffic scheduling device trains multiple target scheduling models based on the multi-dimensional network index data to obtain a target service traffic scheduling baseline. It should be noted that each type of scheduling model is obtained by model training based on multi-dimensional network index samples and the bass service traffic scheduling baseline label, which is no different from the existing model training process, and will not be described here.

[0092] Further, the distributed service traffic scheduling device dynamically adjusts the traffic scheduling of the distributed service based on the target service traffic scheduling baseline.

[0093] The distributed service traffic scheduling method provided by the application obtains comprehensive multi-dimensional network index data, determines the group location characteristics and group performance characteristics of the current user by combining multi-dimensional network index historical data, selects multiple target scheduling models suitable for the current user from two different characteristic dimensions by comprehensively considering the group location characteristics and group performance characteristics of the current user, obtains a precise target service traffic scheduling baseline by comprehensively scheduling through multiple target scheduling models based on comprehensive multi-dimensional network index data, and dynamically adjusts the traffic scheduling of the distributed service based on the precise target service traffic scheduling baseline, thereby improving the accuracy of distributed service traffic scheduling, enhancing the early perception ability of service situation, reducing the pressure and security risks brought by service bursts, and improving user satisfaction.

[0094] Further, after obtaining the real-time multi-dimensional network index data of the user and the multi-dimensional network index historical data of the user at multiple historical time points, and before determining the group location characteristics of the user based on the multi-dimensional network index data and the multiple multi-dimensional network index historical data, the following steps are performed for the multi-dimensional network index data:

[0095] determine whether each dimension network index data satisfies the preset weight update trigger rule corresponding to the dimension network index data;

[0096] If the dimension network index data satisfies the preset weight update trigger rule corresponding to the dimension network index data, the preset weight value corresponding to the dimension network index data is increased to obtain a target weight value corresponding to each dimension network index data;

[0097] According to the target weight value corresponding to each dimension network index data, the multi-dimensional network index data is sorted according to the weight value size to obtain a multi-dimensional network index sequence.

[0098] The following describes Figure 2 the refinement of step S2.

[0099] It should be noted that the above description is combined with the following description Figure 3 , Figure 3 is a schematic diagram of data sorting provided by the present application. Before the sorting process, a preset weight is defined for each dimension network index data, that is, the user's network location information, the user's network delay and jitter information, the application resource availability, the application response rate and the success rate each have a preset weight, which can be defined as the same base weight, such as 0.5. At the same time, a preset weight update trigger rule is defined for each dimension network index data. Since the types of each dimension network index data are different, the preset weight update trigger rules defined for each dimension network index data are different. After the preset weight update trigger rule corresponding to any dimension network index data is triggered, the corresponding preset weight is increased, and then the dimension network index data is sorted according to the adjusted weight value from large to small. If the preset weight update trigger rule corresponding to any dimension network index data is not triggered, the corresponding preset weight is not updated.

[0100] Specifically, the distributed business traffic scheduling device determines whether each dimension network index data meets the preset weight update trigger rule corresponding to the dimension network index data.

[0101] Further, if the dimension network index data meets the preset weight update trigger rule corresponding to the dimension network index data, the distributed business traffic scheduling device increases the preset weight value corresponding to the dimension network index data to obtain the target weight value corresponding to the dimension network index data.

[0102] If the dimension network index data does not meet the preset weight update trigger rule corresponding to the dimension network index data, the distributed business traffic scheduling device does not update the preset weight value corresponding to the dimension target index data, and determines it as the target weight value corresponding to the dimension network index data.

[0103] Through the above method, the target weight value corresponding to each dimension network index data can be obtained.

[0104] Further, the distributed business traffic scheduling device sorts the multi-dimensional network index data according to the target weight value corresponding to each dimension network index data according to the weight value size to obtain a multi-dimensional network index sequence.

[0105] It should be noted that in the sorting process, the larger the target weight value, the earlier the corresponding network index data is arranged, and the network index data with the same target weight value can be randomly arranged.

[0106] In an embodiment, the preset weights of the network location information of the user, the user network delay and jitter information, the application resource availability, the application response rate and success rate are all 0.5, and it is assumed that the preset weight update trigger rule corresponding to the user network delay and jitter information is that the weight of the network delay and jitter information is increased when the network delay exceeds a certain threshold or the jitter information exceeds a certain threshold. At this time, if at least one of the network delay or the jitter information of the user exceeds the corresponding threshold, the basic weight corresponding to the user network delay and jitter information is increased by 0.1. At this time, the target weight value of the user network delay and jitter information is 0.6, and the target weight values of the network location information of the user, the application resource availability, the application response rate and success rate are all 0.5. Therefore, in the sorting, the user network delay and jitter information is ranked first, and the network location information of the user, the application resource availability, the application response rate and success rate are randomly arranged.

[0107] Similarly, the multi-dimensional network index historical data of the user at multiple historical time points is also adaptively sorted in the above-mentioned sorting manner to obtain a multi-dimensional network index historical sequence set. It can be understood that the sorting results of the network location information of the user, the user network delay and jitter information, the application resource availability, the application response rate and success rate of the same user at different times or different users may be different.

[0108] In an embodiment, the multi-dimensional network index historical sequence set can be represented as follows:

[0109] Historical time 1: [network location information of the user, user network delay and jitter information, application resource availability, application response rate and success rate];

[0110] Historical time 2: [application resource availability, user network delay and jitter information, network location information of the user, application response rate and success rate];

[0111] Historical time 3: [user network delay and jitter information, application response rate and success rate, network location information of the user, application resource availability];

[0112] ……;

[0113] Historical time n: [application response rate and success rate, user network delay and jitter information, network location information of the user, application resource availability].

[0114] The embodiment of the present application can represent the value degree of multi-dimensional data on distributed business traffic scheduling through multi-dimensional network index data, facilitate subsequent accurate prediction through a scheduling model, in addition, the multi-dimensional network index historical data of multiple historical time points can determine the network change of the user at different times, and then the group location feature and the group performance feature of the user are determined combined with the multi-dimensional network index data, the scheduling model most suitable for the user is selected through the group location feature and the group performance feature of the user to perform accurate prediction, and the prediction accuracy is improved.

[0115] Further, based on step 104, the target scheduling model is selected from the plurality of types of scheduling models based on the group location feature and the group performance feature, comprising:

[0116] determining the first performance index of each type of scheduling model on the group location feature and the second performance index of each type of scheduling model on the group performance feature;

[0117] determining the scheduling model corresponding to the optimal performance index in the plurality of first performance indexes as the first target scheduling model;

[0118] determining the scheduling model corresponding to the optimal performance index in the plurality of second performance indexes as the second target scheduling model.

[0119] The following description Figure 2 refines the content of step S4.

[0120] It should be noted that the above Figure 4 , Figure 4 is a schematic diagram of the scheduling model selection provided by the present application. In the pre-training process of the scheduling model, the performance index of each type of scheduling model on each group location feature is recorded, so that a scheduling model with the optimal performance index can be selected according to the group location feature of the user. Similarly, in the pre-training process of the scheduling model, the performance index of each type of scheduling model on each group performance feature is recorded, so that a scheduling model with the optimal performance index can be selected according to the group performance feature of the user. The performance index includes but is not limited to accuracy, precision, recall rate and F1 score.

[0121] Specifically, the distributed business traffic scheduling device determines the first performance index of each type of scheduling model on the group location feature of the user and the second performance index of each type of scheduling model on the group performance feature.

[0122] ​Further, the distributed service traffic scheduling device determines an optimal performance indicator meeting the group location characteristics of the user from the plurality of first performance indicators, and determines an optimal performance indicator meeting the group performance characteristics of the user from the second performance indicators.

[0123] Further, the distributed service traffic scheduling device determines a scheduling model corresponding to the optimal performance indicator meeting the group location characteristics of the user as the first target scheduling model.

[0124] Further, the distributed service traffic scheduling device determines a scheduling model corresponding to the optimal performance indicator meeting the group performance characteristics of the user as the second target scheduling model.

[0125] In an embodiment, it is assumed that user A often uses services in city B, and city B corresponds to data center X. By analyzing the historical data of user A, it is found that user A is very sensitive to network delay in data center X. Therefore, it is necessary to select a scheduling model optimized for low delay in data center X, and determine it as the first target scheduling model. At the same time, if the network usage mode of user A is different on weekdays and weekends, it is also possible to adjust the scheduling strategy according to these modes to ensure the best service experience at different time periods.

[0126] In another embodiment, by analyzing the historical data of user A, it is found that the network used by user A usually has high throughput, indicating that the customer belongs to a high-throughput demand user. Therefore, it is necessary to select a scheduling model optimized for high throughput, and determine it as the second target scheduling model.

[0127] The embodiment of the application selects a plurality of target scheduling models from a plurality of model types based on the group location characteristics and the group performance characteristics determined based on the historical data of the user, can select the target scheduling model most suitable for the user according to different dimensional characteristics, and then performs comprehensive scheduling through the plurality of target scheduling models to obtain a target service traffic scheduling baseline, which can improve the accuracy of prediction of the target scheduling model.

[0128] Further, based on step 105, the target service traffic scheduling baseline is obtained by training the plurality of target scheduling models based on the multi-dimensional network indicator data, including:

[0129] performing feature extraction on the multi-dimensional network indicator sequence to obtain multi-dimensional network indicator representation data;

[0130] inputting the multi-dimensional network indicator representation data into the first target scheduling model to obtain a first service traffic scheduling baseline output by the first target scheduling model;

[0131] Input the multi-dimensional network index representation data into the second target scheduling model to obtain a second service traffic scheduling baseline output by the second target scheduling model;

[0132] Perform mean value calculation based on the first service traffic scheduling baseline and the second service traffic scheduling baseline to obtain a target service traffic scheduling baseline.

[0133] The following describes Figure 2 the refinement of step S5 in the method.

[0134] Specifically, the distributed service traffic scheduling device performs feature extraction on the multi-dimensional network index sequence to obtain multi-dimensional network index representation data.

[0135] Further, the distributed service traffic scheduling device inputs the multi-dimensional network index representation data into a first target scheduling model to obtain a first service traffic scheduling baseline output by the first target scheduling model.

[0136] Further, the distributed service traffic scheduling device inputs the multi-dimensional network index representation data into a second target scheduling model to obtain a second service traffic scheduling baseline output by the second target scheduling model.

[0137] Further, the distributed service traffic scheduling device performs mean value calculation based on the first service traffic scheduling baseline and the second service traffic scheduling baseline to obtain a target service traffic scheduling baseline, which can be considered as a baseline for user access and computing power scheduling, and is used for measuring the difference between actual user access behavior traffic and expectation.

[0138] The embodiment of the application performs comprehensive prediction on the multi-dimensional network index sequence by using the target scheduling model suitable for the user selected based on different dimensional features, and performs mean value calculation on the prediction results output by each target scheduling model to obtain an accurate target service traffic scheduling baseline, thereby improving the prediction accuracy of the scheduling model, and achieving accurate distributed service traffic scheduling based on the accurate target service traffic scheduling baseline.

[0139] Further, based on step 106, the dynamic adjustment of distributed service traffic scheduling based on the target service traffic scheduling baseline comprises:

[0140] Obtain real-time first access behavior traffic data of the user;

[0141] If the first access behavior traffic data deviates from the target service traffic scheduling baseline, the distributed service and user access are dynamically scheduled.

[0142] The following describes Figure 2 the refinement of step S6 in the method.

[0143] Specifically, the distributed business traffic scheduling device acquires real-time first access behavior traffic data of a user, wherein the first access behavior traffic data records access traffic generated by the user when accessing a system, including access frequency, data transmission volume, etc.

[0144] Further, the distributed business traffic scheduling device compares and analyzes the first access behavior traffic data and a target business traffic scheduling baseline.

[0145] If the first access behavior traffic data deviates from the target business traffic scheduling baseline, the distributed business traffic scheduling device dynamically schedules distributed businesses and user access.

[0146] If the first access behavior traffic data does not deviate from the target business traffic scheduling baseline, the distributed business traffic scheduling device continues to use an original scheduling strategy of the distributed businesses and the user access, and records the first access behavior traffic data of the user as user access behavior traffic historical data of the user, which can be used for subsequent optimization of distributed business traffic scheduling.

[0147] When the first access behavior traffic data of the user deviates from the target business traffic scheduling baseline, the embodiment of the application timely dynamically schedules the distributed businesses and the user access, can adapt to development of a distributed cloud, meets demands of high availability, flexible and elastic deployment of a business, realizes fine and dynamic scheduling of user access traffic, simultaneously strengthens an early perception ability of a business situation, reduces pressure and security risks brought by a business burst, and thus improves user satisfaction.

[0148] Further, after dynamically adjusting the distributed businesses based on the target business traffic scheduling baseline, the following steps are included:

[0149] Acquiring real-time second access behavior traffic data of the user;

[0150] Comparing and analyzing the second access behavior traffic data and the target business traffic scheduling baseline to obtain a comparison and analysis result;

[0151] Iteratively performing the step of acquiring the real-time second access behavior traffic data of the user until a plurality of comparison and analysis results are obtained after a continuous time period is executed;

[0152] Generating historical dynamic year-on-year data based on the plurality of comparison and analysis results;

[0153] Optimizing a strategy of distributed business traffic scheduling based on the historical dynamic year-on-year data.

[0154] Specifically, the distributed business traffic scheduling device acquires real-time second access behavior traffic data of a user.

[0155] Further, the distributed business traffic scheduling device performs comparative analysis based on the second access behavior traffic data and the target business traffic scheduling baseline, obtaining a comparative analysis result, which reflects the deviation of the user access behavior traffic from the baseline and the trend of the user access behavior traffic.

[0156] Further, the distributed business traffic scheduling device iteratively performs the step of obtaining real-time second access behavior traffic data of the user until a plurality of comparative analysis results are obtained after a continuous time period, which can be set according to actual conditions.

[0157] Further, the distributed business traffic scheduling device generates historical dynamic year-on-year data based on the plurality of comparative analysis results.

[0158] Further, the distributed business traffic scheduling device optimizes the strategy of distributed business traffic scheduling based on the historical dynamic year-on-year data.

[0159] In an embodiment, the historical dynamic year-on-year data can be used to optimize the scheduling strategy. The historical dynamic year-on-year data records the changes of user access behavior traffic over time, which can help analyze user behavior patterns and traffic trends, thereby optimizing the scheduling strategy.

[0160] In another embodiment, the historical dynamic year-on-year data can be used to train and adjust the machine learning model. As input, the historical dynamic year-on-year data can be used to train and adjust the machine learning model to improve the prediction accuracy and adaptability of the model.

[0161] In another embodiment, the historical dynamic year-on-year data can be used to automatically adjust the scheduling algorithm. On the basis of the machine learning model, an automatic adjustment scheduling algorithm is added, which uses the historical dynamic year-on-year data to adjust the scheduling of business and access in real time to adapt to the changing business needs of the enterprise system.

[0162] In another embodiment, the historical dynamic year-on-year data can be used to predict and cope with traffic peaks. By analyzing the historical dynamic year-on-year data, traffic peak periods can be predicted, and resource allocation and scheduling can be performed in advance to avoid system overload.

[0163] The embodiment of the present application generates historical dynamic year-on-year data by continuously obtaining comparative analysis results between user access behavior traffic data and target business traffic scheduling baseline, and further optimizes the measurement of distributed business traffic scheduling based on historical dynamic year-on-year data, so that the system can continuously learn and adapt to changes in user behavior, and realize more intelligent and automated distributed business traffic scheduling.

[0164] The following description Figure 2 refines the content of step S6.

[0165] Furthermore, the distributed service traffic scheduling method also includes:

[0166] Obtain business support capability data;

[0167] Based on the aforementioned business support capability data, a performance bottleneck analysis is performed to obtain the performance bottleneck analysis results.

[0168] Based on the performance bottleneck analysis results, the timing of distributed service traffic scheduling is determined.

[0169] At the time point of the distributed service traffic scheduling, distributed services and user access are dynamically scheduled.

[0170] Specifically, the distributed business traffic scheduling device acquires business support capability data.

[0171] The business support capability data includes, but is not limited to, server load, response time, and resource utilization (CPU, memory, storage, network), which reflect the system's performance and carrying capacity.

[0172] Furthermore, the distributed business traffic scheduling device performs performance bottleneck analysis based on business support capability data to obtain performance bottleneck analysis results, which can reflect the current status and performance bottlenecks of the business system.

[0173] Furthermore, based on the performance bottleneck analysis results, the distributed service traffic scheduling device determines the timing of distributed service traffic scheduling so that distributed services and user access can be dynamically scheduled in advance before the service support capacity data becomes overloaded.

[0174] Furthermore, the distributed service traffic scheduling device dynamically schedules distributed services and user access at the time points of distributed service traffic scheduling.

[0175] It should be noted that after dynamically scheduling distributed services and user access based on business support capability data, historical dynamic month-on-month data will be gradually formed and accumulated. Based on the historical dynamic month-on-month data, an automatic adjustment scheduling algorithm will be added to the scheduling model to adapt to the ever-changing business needs of the business system.

[0176] This invention, through acquiring business support capability data and performing performance bottleneck analysis, can identify performance bottlenecks in the system. Furthermore, based on the performance bottleneck analysis results, it can determine appropriate time points for distributed business traffic scheduling, thereby optimizing the utilization of system resources, reducing the impact of performance bottlenecks, improving the overall system response speed and stability, and enhancing the user experience.

[0177] The distributed service traffic scheduling device provided by the application is described below, and the distributed service traffic scheduling device described below can be correspondingly referred to the distributed service traffic scheduling method described above.

[0178] Referring to Figure 5 , Figure 5 FIG. 1 is a structural schematic diagram of the distributed service traffic scheduling device provided by the application.

[0179] The distributed service traffic scheduling device comprises:

[0180] The acquisition module 510 is configured to acquire real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at a plurality of historical time points.

[0181] The first determination module 520 is configured to determine a group position feature of the user based on the multi-dimensional network index data and the plurality of multi-dimensional network index historical data; the group position feature is used to reflect a characteristic performance of a network position where the user is located.

[0182] The second determination module 530 is configured to determine a group performance feature of the user based on the plurality of multi-dimensional network index historical data; the group performance feature is used to reflect a characteristic performance of the user on network performance.

[0183] The scheduling model selection module 540 is configured to select a plurality of target scheduling models from a plurality of types of scheduling models based on the group position feature and the group performance feature; any of the scheduling models is obtained based on model training of multi-dimensional network index samples and corresponding service traffic scheduling baseline labels.

[0184] The scheduling model prediction module 550 is configured to train the plurality of target scheduling models respectively based on the multi-dimensional network index data to obtain a target service traffic scheduling baseline.

[0185] The scheduling module 560 is configured to dynamically adjust distributed service traffic scheduling based on the target service traffic scheduling baseline.

[0186] The distributed service traffic scheduling device provided by the application obtains comprehensive multi-dimensional network index data, determines the group location characteristics and group performance characteristics of the current user by combining the multi-dimensional network index historical data, selects a plurality of target scheduling models suitable for the current user from two different characteristic dimensions by comprehensively considering the group location characteristics and group performance characteristics of the current user, performs comprehensive scheduling through the plurality of target scheduling models based on the comprehensive multi-dimensional network index data, obtains a precise target service traffic scheduling baseline, and then dynamically adjusts the traffic scheduling of the distributed service based on the precise target service traffic scheduling baseline, improves the accuracy of the distributed service traffic scheduling, strengthens the early perception ability of the service situation, reduces the pressure and security risks caused by service bursts, and thus improves the user satisfaction.

[0187] Further, the distributed service traffic scheduling device is further used for:

[0188] determining whether each dimensional network index data meets the preset weight update trigger rule corresponding to the dimensional network index data;

[0189] if the dimensional network index data meets the preset weight update trigger rule corresponding to the dimensional network index data, increasing the preset weight value corresponding to the dimensional network index data to obtain a target weight value corresponding to each dimensional network index data;

[0190] according to the target weight value corresponding to each dimensional network index data, sorting the multi-dimensional network index data according to the weight value to obtain a multi-dimensional network index sequence.

[0191] Further, the scheduling model selection module 540 is further used for:

[0192] determining a first performance index of each type of scheduling model on the group location characteristics and a second performance index of each type of scheduling model on the group performance characteristics;

[0193] determining the scheduling model corresponding to the optimal performance index in the plurality of first performance indexes as a first target scheduling model;

[0194] determining the scheduling model corresponding to the optimal performance index in the plurality of second performance indexes as a second target scheduling model.

[0195] Further, the scheduling model prediction module 550 is further used for:

[0196] performing feature extraction on the multi-dimensional network index sequence to obtain multi-dimensional network index representation data;

[0197] input the multi-dimensional network index characterization data into the first target scheduling model to obtain a first service traffic scheduling baseline output by the first target scheduling model;

[0198] input the multi-dimensional network index characterization data into the second target scheduling model to obtain a second service traffic scheduling baseline output by the second target scheduling model;

[0199] perform mean value calculation based on the first service traffic scheduling baseline and the second service traffic scheduling baseline to obtain a target service traffic scheduling baseline.

[0200] Further, the scheduling module 560 is further configured to:

[0201] obtain real-time first access behavior traffic data of the user;

[0202] if the first access behavior traffic data deviates from the target service traffic scheduling baseline, perform dynamic scheduling on distributed service and user access.

[0203] Further, the distributed service traffic scheduling apparatus is further configured to:

[0204] obtain real-time second access behavior traffic data of the user;

[0205] perform comparative analysis based on the second access behavior traffic data and the target service traffic scheduling baseline to obtain a comparative analysis result;

[0206] iteratively perform the step of obtaining real-time second access behavior traffic data of the user until a plurality of comparative analysis results are obtained after a continuous time period is executed;

[0207] generate historical dynamic year-on-year data based on the plurality of comparative analysis results;

[0208] optimize the strategy of distributed service traffic scheduling based on the historical dynamic year-on-year data.

[0209] Further, the distributed service traffic scheduling apparatus is further configured to:

[0210] obtain service support capability data;

[0211] perform performance bottleneck analysis based on the service support capability data to obtain a performance bottleneck analysis result;

[0212] determine a time point of distributed service traffic scheduling based on the performance bottleneck analysis result;

[0213] perform dynamic scheduling on distributed service and user access at the time point of distributed service traffic scheduling.

[0214] It should be noted that the distributed service traffic scheduling device provided by the present application can execute the distributed service traffic scheduling method described in any of the above embodiments during actual operation, and the present embodiment will not be described here.

[0215] Figure 6 is a structural schematic diagram of an electronic device provided by the present application, as Figure 6 shown, the electronic device can include a processor 610, a communications interface 620, a memory 630, and a communications bus 640, wherein the processor 610, the communications interface 620, and the memory 630 complete mutual communication through the communications bus 640. The processor 610 can invoke the logical instructions in the memory 630 to execute a distributed service traffic scheduling method, which includes: obtaining real-time multi-dimensional network indicator data of a user and multi-dimensional network indicator historical data of the user at multiple historical time points; determining group location characteristics of the user based on the multi-dimensional network indicator data and the multiple multi-dimensional network indicator historical data; the group location characteristics are used to reflect the characteristic performance of the network location where the user is located; determining group performance characteristics of the user based on the multiple multi-dimensional network indicator historical data; the group performance characteristics are used to reflect the characteristic performance of the network performance of the user; selecting multiple target scheduling models from multiple types of scheduling models based on the group location characteristics and the group performance characteristics; any of the scheduling models is obtained based on model training of multi-dimensional network indicator samples and corresponding service traffic scheduling baseline labels; training the multiple target scheduling models respectively based on the multi-dimensional network indicator data to obtain a target service traffic scheduling baseline; and dynamically adjusting the distributed service traffic based on the target service traffic scheduling baseline.

[0216] In addition, the logical instructions in the memory 630 described above can be implemented in the form of a software functional unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium, includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0217] In another aspect, the present application also provides a computer program product, which comprises a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the distributed traffic scheduling method provided by any of the above embodiments, the method comprising: obtaining real-time multi-dimensional network indicator data of a user and multi-dimensional network indicator historical data of the user at a plurality of historical time points; determining a group location feature of the user based on the multi-dimensional network indicator data and the plurality of multi-dimensional network indicator historical data, the group location feature being used to reflect a characteristic performance of a network location where the user is located; determining a group performance feature of the user based on the plurality of multi-dimensional network indicator historical data, the group performance feature being used to reflect a characteristic performance of the user on network performance; selecting a plurality of target scheduling models from a plurality of types of scheduling models based on the group location feature and the group performance feature, any of the scheduling models being obtained based on model training of multi-dimensional network indicator samples and corresponding traffic scheduling baseline labels; training the plurality of target scheduling models based on the multi-dimensional network indicator data to obtain target traffic scheduling baselines; and dynamically adjusting traffic scheduling of distributed services based on the target traffic scheduling baselines.

[0218] In yet another aspect, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the distributed traffic scheduling method provided by any of the above embodiments, the method comprising: obtaining real-time multi-dimensional network indicator data of a user and multi-dimensional network indicator historical data of the user at a plurality of historical time points; determining a group location feature of the user based on the multi-dimensional network indicator data and the plurality of multi-dimensional network indicator historical data, the group location feature being used to reflect a characteristic performance of a network location where the user is located; determining a group performance feature of the user based on the plurality of multi-dimensional network indicator historical data, the group performance feature being used to reflect a characteristic performance of the user on network performance; selecting a plurality of target scheduling models from a plurality of types of scheduling models based on the group location feature and the group performance feature, any of the scheduling models being obtained based on model training of multi-dimensional network indicator samples and corresponding traffic scheduling baseline labels; training the plurality of target scheduling models based on the multi-dimensional network indicator data to obtain target traffic scheduling baselines; and dynamically adjusting traffic scheduling of distributed services based on the target traffic scheduling baselines.

[0219] The apparatus embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units. Part or all of the modules can be selected according to actual needs to achieve the purposes of the embodiments. Those skilled in the art can understand and implement without creative labor.

[0220] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and necessary universal hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0221] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for distributed traffic scheduling, the method comprising: The method comprises the following steps: acquiring real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at multiple historical time points; determining group position characteristics of the user based on the multi-dimensional network index data and the multi-dimensional network index historical data; the group position characteristics are used to reflect characteristic performance of a network position where the user is located; determining group performance characteristics of the user based on the multi-dimensional network index historical data; the group performance characteristics are used to reflect characteristic performance of the user on network performance; selecting multiple target scheduling models from multiple types of scheduling models based on the group position characteristics and the group performance characteristics; any of the scheduling models is obtained based on model training of multi-dimensional network index samples and corresponding traffic scheduling baseline labels; training the multiple target scheduling models based on the multi-dimensional network index data respectively to obtain target traffic scheduling baselines; dynamically adjusting traffic scheduling of distributed services based on the target traffic scheduling baselines.

2. The method of claim 1, wherein, After acquiring real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at multiple historical time points, and before determining group position characteristics of the user based on the multi-dimensional network index data and the multi-dimensional network index historical data, the following steps are performed for multi-dimensional network index data: determining whether each dimension network index data satisfies a preset weight update trigger rule corresponding to the dimension network index data; if the dimension network index data satisfies the preset weight update trigger rule corresponding to the dimension network index data, increasing a preset weight value corresponding to the dimension network index data to obtain a target weight value corresponding to each dimension network index data; sorting the multi-dimensional network index data according to weight values in size according to the target weight value corresponding to each dimension network index data to obtain a multi-dimensional network index sequence.

3. The method of claim 2, wherein, The target scheduling model comprises a first target scheduling model and a second target scheduling model; the selecting multiple target scheduling models from multiple types of scheduling models based on the group position characteristics and the group performance characteristics comprises: determining a first performance index of each type of scheduling model on the group position characteristics and a second performance index of each type of scheduling model on the group performance characteristics; determining a scheduling model corresponding to an optimal performance index in the multiple first performance indexes as the first target scheduling model; determining a scheduling model corresponding to an optimal performance index in the multiple second performance indexes as the second target scheduling model.

4. The method of claim 3, wherein, The training the multiple target scheduling models based on the multi-dimensional network index data respectively to obtain target traffic scheduling baselines comprises: performing feature extraction on the multi-dimensional network index sequence to obtain multi-dimensional network index representation data; inputting the multi-dimensional network index representation data into the first target scheduling model to obtain a first traffic scheduling baseline output by the first target scheduling model; inputting the multi-dimensional network index characteristic data into the second target scheduling model to obtain a second service traffic scheduling baseline output by the second target scheduling model; performing mean value calculation based on the first service traffic scheduling baseline and the second service traffic scheduling baseline to obtain a target service traffic scheduling baseline.

5. The method of claim 1, wherein, The dynamic adjustment of the distributed service traffic scheduling based on the target service traffic scheduling baseline comprises: obtaining real-time first access behavior traffic data of the user; if the first access behavior traffic data deviates from the target service traffic scheduling baseline, dynamically scheduling the distributed service and user access.

6. The method of claim 5, wherein, After the dynamic adjustment of the distributed service traffic scheduling based on the target service traffic scheduling baseline, comprising: obtaining real-time second access behavior traffic data of the user; comparing and analyzing the second access behavior traffic data and the target service traffic scheduling baseline to obtain a comparison and analysis result; iteratively performing the step of obtaining real-time second access behavior traffic data of the user until a plurality of comparison and analysis results are obtained after a continuous period of time; based on the plurality of comparison and analysis results, generating historical dynamic ring ratio data; based on the historical dynamic ring ratio data, optimizing the strategy of distributed service traffic scheduling.

7. The distributed traffic scheduling method according to any of claims 1-6, characterized in that, The distributed service traffic scheduling method further comprises: obtaining service support capability data; performing performance bottleneck analysis based on the service support capability data to obtain a performance bottleneck analysis result; based on the performance bottleneck analysis result, determining a time point of distributed service traffic scheduling; at the time point of distributed service traffic scheduling, dynamically scheduling the distributed service and user access.

8. A distributed traffic flow scheduling apparatus, characterized by, comprising: an acquisition module for acquiring real-time multi-dimensional network index data of a user and multi-dimensional network index historical data of the user at a plurality of historical time points; a first determination module for determining group location characteristics of the user based on the multi-dimensional network index data and a plurality of the multi-dimensional network index historical data; the group location characteristics are used to reflect the characteristic performance of the network location where the user is located; a second determination module for determining group performance characteristics of the user based on a plurality of the multi-dimensional network index historical data; the group performance characteristics are used to reflect the characteristic performance of the user on network performance; a scheduling model selection module for selecting a plurality of target scheduling models from a plurality of types of scheduling models based on the group location characteristics and the group performance characteristics; any of the scheduling models is obtained by model training based on multi-dimensional network index samples and corresponding service traffic scheduling baseline labels; a scheduling model prediction module for training the plurality of target scheduling models based on the multi-dimensional network index data to obtain a target service traffic scheduling baseline; a scheduling module for dynamically adjusting distributed service traffic scheduling based on the target service traffic scheduling baseline.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, The processor executes the computer program to implement the steps of the distributed service traffic scheduling method according to any one of claims 1 to 7. The processor executes the computer program to implement the steps of the distributed service traffic scheduling method according to any one of claims 1 to 7.

10. A computer-readable storage medium, the computer-readable storage medium comprising a non-transitory computer-readable storage medium, having stored thereon a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the distributed traffic flow scheduling method according to any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the distributed traffic flow scheduling method according to any one of claims 1 to 7.

Images