A service provider sla evaluation dynamic business model implementation method and system

Abstract

The application belongs to the technical field of business intelligence evaluation, and specifically provides a service provider SLA evaluation dynamic business model implementation method and system. The method mainly comprises: obtaining SLA evaluation source data of a service provider, wherein the SLA evaluation source data comprises operator SLA data, enterprise end-to-end APM data, third-party network quality data and dependent link probe data between service providers; taking the service provider as a node, setting a credibility weight for each node according to a preset rule, determining a directed edge between nodes according to a preset dependent relationship between service providers, and determining a connection weight of the directed edge according to the dependent link probe data. The application realizes dynamic and accurate evaluation under multi-dimensional data fusion, improves the objectivity and credibility of SLA achievement rate evaluation results, and provides reliable decision support for enterprise screening of cloud service providers and optimization of service deployment.

Description

Technical Field

[0001] This application belongs to the field of business intelligence assessment technology, and specifically relates to a method and system for implementing a dynamic business model for service provider SLA assessment. Background Technology

[0002] In the field of assessing the achievement of service level agreements (SLAs) by cloud service providers, existing assessment methods generally use the self-monitoring SLA data provided by operators as the core assessment basis to conduct calculations and output the assessment comparison results of multiple service providers, providing a reference for business units to select service providers.

[0003] However, in the actual cloud service delivery process, the collection and generation of SLA data by operators' self-monitoring are based on their own deployed monitoring nodes and statistical methods. The data presentation results are easily affected by their own operation and maintenance management strategies, resulting in a deviation between the final SLA achievement rate assessment results and the actual service level of the service provider. Such deviations can interfere with the judgment of the business side, making it difficult for enterprises to accurately select cloud service providers that meet their own needs, which is not conducive to the optimization of subsequent service deployment. Summary of the Invention

[0004] This application provides a method and system for implementing a dynamic business model for service provider SLA evaluation. It effectively solves the problem in the prior art where the evaluation results deviate from the actual service level due to reliance on operator self-monitoring SLA data, interfering with the business side's selection decision. It improves the objectivity and accuracy of SLA evaluation and can provide reliable decision support for enterprises to accurately select service providers.

[0005] To achieve the above objectives, this application adopts the following technical solution:

[0006] Firstly, this application provides a method for implementing a dynamic business model for service provider SLA evaluation, comprising: acquiring SLA evaluation source data of service providers, wherein the SLA evaluation source data includes operator SLA data, enterprise end-to-end APM data, third-party network quality data, and dependency link probe data between service providers; taking service providers as nodes, assigning a credibility weight to each node according to preset rules, determining directed edges between nodes according to preset dependency relationships between service providers, and determining the connection weight of directed edges according to dependency link probe data; calculating preliminary deviation data of service providers based on the SLA evaluation source data; when the preliminary deviation data of the first service provider exceeds a preset first threshold, and the enterprise end-to-end APM data or third-party network quality data of at least one second service provider on which the first service provider depends exceeds a preset second threshold within a time window, updating the credibility weight of the first service provider according to the credibility weight of the second service provider, operator SLA data, and connection weight; calculating the SLA achievement rate data of the first service provider based on the credibility weight and SLA evaluation source data; and generating an evaluation result based on the SLA achievement rate data of multiple first service providers.

[0007] Furthermore, the process of acquiring dependency link probe data between service providers includes: acquiring performance metadata collected by probes deployed on communication links between service providers with business dependencies; and generating dependency link probe data containing one-way latency, jitter, and packet loss rate based on the performance metadata.

[0008] Furthermore, a credibility weight is assigned to each node according to preset rules, including: for each node, querying the node's historical credibility weight from the preset evaluation history database as the credibility weight; if the evaluation history database does not contain the node's historical credibility weight, then assigning the node a preset default weight value as the credibility weight.

[0009] Furthermore, the connection weights of directed edges are determined based on the dependent link probe data, including: mapping the one-way latency, jitter, and packet loss rate in the dependent link probe data to individual quality scores within a preset numerical range; and taking the minimum value among the individual quality scores as the connection weights of the directed edges.

[0010] Furthermore, the preliminary deviation data for service providers is calculated, including: mapping the SLA assessment source data to a unified dimension to obtain standardized data; and calculating the statistical differences between each service provider's operator SLA data, enterprise end-to-end APM data, and third-party network quality data in the standardized data to obtain preliminary deviation data.

[0011] Further, the credibility weight of the first service provider is updated based on the credibility weight of the second service provider, the operator SLA data, and the connection weight. This includes: for each second service provider, calculating the inferred value of the operator SLA data of the second service provider relative to the first service provider based on the operator SLA data and the corresponding connection weight; taking a weighted average of all inferred values ​​based on the credibility weight of each second service provider to obtain a comprehensive inferred value of the operator SLA data of the first service provider; calculating the consistency data between the operator SLA data of the first service provider and the comprehensive inferred value; calculating the adjustment amount of the credibility weight of the first service provider based on the consistency data, the credibility weight of the first service provider, and the dependency link probe data of each second service provider; and updating the credibility weight of the first service provider based on the adjustment amount.

[0012] Furthermore, based on the consistency data, the credibility weight of the first service provider, and the dependency link probe data of each second service provider, the adjustment amount of the credibility weight of the first service provider is calculated, including: subtracting a preset baseline value from the consistency data to obtain the basic deviation; multiplying the credibility weight of the first service provider by the result of subtracting the credibility weight to obtain the sensitivity; calculating the link reward / penalty coefficient based on the dependency link probe data of the second service providers; and multiplying the basic deviation, sensitivity, link reward / penalty coefficient, and a preset learning rate constant to obtain the adjustment amount of the credibility weight of the first service provider.

[0013] Furthermore, based on the dependency link probe data of the second service provider, the link reward and penalty coefficient is calculated, including: for each second service provider, the geometric mean of all individual quality scores is calculated as the comprehensive score of the second service provider; the geometric mean of the comprehensive scores of all second service providers is taken as the link reward and penalty coefficient.

[0014] Furthermore, based on the credibility weight and SLA assessment source data, the SLA achievement rate data of the first service provider is calculated, including: weighted fusion of operator SLA data, enterprise end-to-end APM data and third-party network quality data based on the credibility weight to obtain the SLA achievement rate data of the first service provider.

[0015] Secondly, this application provides a system for implementing a dynamic business model for service provider SLA evaluation, comprising:

[0016] Data acquisition module: Used to acquire SLA assessment source data from service providers. The SLA assessment source data includes operator SLA data, enterprise end-to-end APM data, third-party network quality data, and dependency link probe data between service providers.

[0017] Network construction module: It is used to set the trust weight of each node with service provider as node according to preset rules, determine the directed edges between nodes according to the preset dependency relationship between service providers, and determine the connection weight of directed edges according to dependency link probe data.

[0018] Game Theory Module: Used to calculate the initial deviation data of the service provider based on the SLA assessment source data; when the initial deviation data of the first service provider exceeds the preset first threshold, and the enterprise end-to-end APM data or third-party network quality data of at least one second service provider on which the first service provider depends exceeds the preset second threshold within the time window, the credibility weight of the first service provider is updated according to the credibility weight of the second service provider, the operator's SLA data, and the connection weight.

[0019] Evaluation module: Used to calculate the SLA achievement rate of the first service provider based on credibility weight and SLA evaluation source data.

[0020] Results generation module: Used to generate evaluation results based on SLA achievement rate data from multiple first-tier service providers.

[0021] Thirdly, this application provides a readable storage medium storing computer program instructions, which are read and executed by a processor to perform the steps of a service provider SLA evaluation dynamic business model implementation method.

[0022] The beneficial effects of this application are:

[0023] This application effectively solves the problem in existing technologies where relying on operator-monitored SLA data leads to discrepancies between assessment results and actual service levels, interfering with business-side selection decisions. It achieves dynamic and accurate assessment of service providers based on multi-dimensional data, improves the objectivity and credibility of SLA achievement rate assessment results, and provides reliable decision support for enterprises to select cloud service providers and optimize service deployment. By collecting multi-source SLA assessment data, setting credibility weights for service providers as nodes and connection weights for dependencies as directed edges, and dynamically updating the weights, the application calculates achievement rates and generates comparison results for multiple service providers.

[0024] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description and the accompanying drawings. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0026] Figure 1 A flowchart illustrating the implementation method of the dynamic business model for service provider SLA evaluation in this application is shown.

[0027] Figure 2 This paper illustrates a flowchart of the process for obtaining dependency link probe data according to this application.

[0028] Figure 3 A flowchart illustrating the process of updating the credibility weight of the first service provider in this application is shown;

[0029] Figure 4 A flowchart illustrating the adjustment amount for calculating the credibility weight of the first service provider in this application is shown.

[0030] Figure 5 The diagram shows a module schematic of the system for implementing the dynamic business model of the service provider SLA assessment in this application. Detailed Implementation

[0031] To address the problems raised in the background technology, this application collects multi-source data such as operator SLAs and enterprise APMs, sets credibility weights for service providers as nodes and connection weights for dependencies as directed edges, dynamically updates the weights, calculates the achievement rate, and generates comparison results of multiple service providers. This improves the objectivity and accuracy of SLA evaluation and can provide reliable decision support for enterprises to accurately select service providers.

[0032] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0033] In some embodiments, such as Figure 1 As shown, this application provides a method for implementing a dynamic business model for service provider SLA evaluation, including:

[0034] S1. Obtain SLA assessment source data from service providers. SLA assessment source data includes operator SLA data, enterprise end-to-end APM data, third-party network quality data, and dependency link probe data between service providers.

[0035] Carrier SLA data represents performance metrics that service providers publicly declare through their own monitoring systems. Examples include P99 latency and availability percentages obtained via monitoring APIs.

[0036] Enterprise end-to-end APM data represents performance data reflecting the end-user experience, directly measured by enterprises using the service through their application performance monitoring systems. For example, querying an enterprise's APM system reveals the actual P99 latency for accessing a WS application.

[0037] Third-party network quality data represents network and application performance data provided by professional third-party monitoring platforms independent of service providers and enterprises. For example, network latency of a WS server endpoint obtained by calling a third-party API.

[0038] Dependency link probe data is used to quantify the quality of network links between service providers with dependencies.

[0039] S2. Construct a game-theoretic verification network: Using service providers as nodes, assign credibility weights to each node according to preset rules, determine directed edges between nodes based on preset dependencies between service providers, and determine the connection weights of directed edges based on dependency link probe data.

[0040] The credibility weight represents the degree of trust in the carrier SLA data provided by the service provider.

[0041] Establish directed edges between nodes that have dependencies, with the direction pointing from the dependent node to the dependent node. For example, a company uses a Web Application Service Provider (WS), a Database Service Provider (DB), and a Message Queuing Service Provider (MQ). The WS application service needs to call the services provided by DB and MQ at runtime, that is, WS depends on DB and MQ. Establish directed edges from DB nodes to WS nodes, and directed edges from MQ nodes to WS nodes.

[0042] Connection weights are used to quantify the communication quality of the link.

[0043] S3. Calculate the preliminary deviation data for the service provider based on the SLA assessment source data. The preliminary deviation data represents the difference between each service provider's self-reported data and independent monitoring data.

[0044] When the initial deviation data of the first service provider exceeds the preset first threshold, and the enterprise end-to-end APM data or third-party network quality data of at least one second service provider on which the first service provider depends exceeds the preset second threshold within the time window, the credibility weight of the first service provider is updated according to the credibility weight of the second service provider, the operator SLA data, and the connection weight.

[0045] The first service provider represents the service provider to be evaluated. The first threshold represents the upper limit of the tolerable difference between self-reported data and independently monitored data. When the initial deviation data of the first service provider exceeds the preset first threshold, it means that there is a significant difference between the operator's SLA data claimed by the first service provider and the enterprise's actual end-to-end APM data and third-party network quality data.

[0046] The second threshold represents the criterion for judging the abnormal performance of the upstream dependent service itself. When the enterprise end-to-end APM data or third-party network quality data of at least one second service provider relied upon by the first service provider exceeds the preset second threshold within the time window, it means that the independent monitoring data of at least one second service provider relied upon by the first service provider also performs poorly. Therefore, the abnormal performance of the first service provider may be caused by the second service provider, or it may not be.

[0047] When the data from the first service provider is abnormal, and the upstream second service provider it relies on also performs poorly, making it difficult to directly determine the root cause of the problem, a dynamic game calculation simulating arbitration is initiated by using a multi-dimensional evidence chain consisting of the credibility weight of the second service provider, operator SLA data, and connection weight. This allows for cross-validation and intelligent attribution of the first service provider's claims, and a fair adjustment of credibility.

[0048] For example, if the first threshold is 0.30, the second threshold is 20ms, the time window is 5 minutes, and the initial deviation data of WS is 0.35, which is greater than the first threshold, the self-report of WS is seriously inconsistent with objective observation, and the enterprise end-to-end APM data or third-party network quality data of DB on which WS depends has been 100ms in the past 5 minutes, which exceeds the second threshold, then the data of DB and MQ will be used to re-evaluate and update the credibility of the operator SLA data provided by WS.

[0049] S4. Calculate the SLA achievement rate data for the first service provider based on the credibility weight and SLA assessment source data. The SLA achievement rate data reflects the credibility assessment value of the first service provider's service level after dynamic game verification.

[0050] S5. Generate evaluation results based on SLA achievement rate data from multiple first-tier service providers, such as generating comparison results of SLA achievement rate data from multiple first-tier service providers, to provide a global view and decision support for the business.

[0051] In some embodiments, such as Figure 2 As shown, the dependency link probe data between service providers is obtained, including:

[0052] S1.1. Obtain performance metadata collected by probes deployed on communication links between service providers with business dependencies.

[0053] Specifically, network probes are deployed on the network paths between identified service providers with business dependencies. The probes send probe data packets to the target server at a fixed frequency and record the raw results of each probe.

[0054] S1.2. Based on performance metadata, generate dependent link probe data including one-way latency, jitter, and packet loss rate.

[0055] For example, in the evaluation of companies A, B, and C, network probes were deployed on the communication links of WS-DB and WS-MQ. The probes sent ICMP probe packets to the DB and MQ servers once per second. Within the time window, the probes recorded raw performance metadata: the raw latency of the WS-DB link was 25ms, jitter was 5ms, and packet loss rate was 0.01%; the raw latency of the WS-MQ link was 30ms, jitter was 10ms, and packet loss rate was 0.02%.

[0056] In some embodiments, a credibility weight is set for each node according to a preset rule, including: for each node, querying the historical credibility weight of the node from a preset evaluation history database as the credibility weight; if there is no historical credibility weight for the node in the evaluation history database, then assigning a preset default weight value to the node as the credibility weight, for example, the default weight can be 0.5.

[0057] After updating the credibility weight of the first service provider, the updated credibility weight of the first service provider will be updated and stored in the evaluation history database as the new historical credibility weight of the first service provider.

[0058] For example, when constructing a game-theoretic verification network, after querying and evaluating the historical database, it is found that the DB node has a historical weight of 0.8 and the MQ node has a historical weight of 0.7. However, since WS is a new service provider, the WS node has no historical records. Therefore, the credibility weight of WS is assigned to the default weight of 0.5, which represents neutral trust.

[0059] In some embodiments, determining the connection weights of directed edges based on dependent link probe data includes:

[0060] S2.1. Map the one-way delay, jitter, and packet loss rate in the dependent link probe data to individual quality scores within a preset value range.

[0061] A scoring function can be defined for each indicator, mapping its raw value to a preset numerical range.

[0062] For example, the scoring function for one-way delay can be:

[0063] ;in, A single quality score representing one-way delay. This represents the measured one-way delay. An ideal threshold for representing latency, such as 20ms, could be used. This represents the tolerance threshold for latency, which can be set to 100ms.

[0064] The scoring function for jitter can be: ;in, The individual quality score representing jitter. The jitter value obtained from the measurement can be the standard deviation of the delay or a similar statistic. An optimal threshold for jitter can be set, such as 50ms. This represents the tolerance threshold for jitter, which can be set to 50ms.

[0065] The packet loss rate scoring function can be: ;in, A single quality score representing packet loss rate. This represents the measured packet loss rate. This represents the tolerance threshold for packet loss rate, which can be set to 0.05.

[0066] S2.2. Take the minimum value among the individual quality scores as the connection weight of the directed edge.

[0067] For example, for a WS-DB link, the measured one-way delay Excellent threshold for latency Delay tolerance threshold Then, the individual quality score of the one-way delay is calculated. Approximately 0.8; jitter is 5ms, the measured jitter value Excellent threshold for jitter jitter tolerance threshold The calculated jitter quality score is then... The measured packet loss rate is 1. ,Right now Tolerance threshold for packet loss rate The calculated packet loss rate is then used to determine the individual quality score. .

[0068] In some embodiments, the preliminary deviation data of the calculation service provider includes:

[0069] S3.1. Map the SLA assessment source data to a unified dimension to obtain standardized data.

[0070] Specifically, a performance scoring function based on its own SLA commitment is defined for each service provider being evaluated.

[0071] For example, the scoring function could be: Where x represents any one of the three raw performance metrics: operator SLA data, enterprise end-to-end APM data, and third-party network quality data. This represents the standardized value of the indicator. The SLA commitment target value represents the service provider's commitment. This represents an unacceptable threshold value.

[0072] S3.2. In the standardized data, calculate the statistical difference between each service provider's operator SLA data, enterprise end-to-end APM data, and third-party network quality data to obtain preliminary deviation data. The larger the preliminary deviation data, the more prominent the contradiction between the service provider's self-reported data and the observed values.

[0073] Preliminary deviation data reference formula: ;in, This represents preliminary deviation data. This represents the standardized value of the operator's SLA data. This represents the standardized value of the enterprise's end-to-end APM data. This represents the standardized value of third-party network quality data.

[0074] For example, WS's operator SLA data Service provider's SLA commitment target value Unacceptable threshold value : The standardized value of the operator's SLA data obtained through calculation Enterprise end-to-end APM data The standardized value of the enterprise end-to-end APM data obtained through calculation. Third-party network quality data The calculated value is the standardized value of the third-party network quality data. The final calculated preliminary deviation data .

[0075] In some embodiments, such as Figure 3 As shown, the credibility weight of the first service provider is updated based on the credibility weight of the second service provider, the operator's SLA data, and the connection weight, including:

[0076] S3.3. For each second service provider, based on the operator SLA data and the corresponding connection weight, calculate the inferred value of the operator SLA data of the second service provider for the first service provider. The inferred value reflects the potential impact of the performance of the second service provider on the first service provider.

[0077] Reference formula: ;in, This represents the inferred value of the i-th second service provider's operator SLA data for the first service provider. This represents the standardized operator SLA data for the i-th second service provider. This represents the connection weight between the i-th second service provider and the first service provider.

[0078] For example, if the standardized operator SLA data of DB and MQ are both 1.0, and the credibility weights of DB and MQ are 0.5 and 1.0 respectively, then the calculated inference values ​​of DB and MQ for the operator SLA data of WS are 0.5 and 1.0 respectively.

[0079] S3.4. Calculate a weighted average of all inferred values ​​based on the credibility weight of each second service provider to obtain the comprehensive inferred value of the operator SLA data of the first service provider.

[0080] Reference formula: ;in, This represents the overall inferred value, where n represents the number of second service providers. This represents the credibility weight of the i-th second service provider.

[0081] For example, if the confidence weight of DB is 0.6 and the confidence weight of MQ is 0.8, then the calculated comprehensive inference value is... .

[0082] S3.5. Calculate the consistency data between the first service provider's operator SLA data and the comprehensive inferred value. The consistency data is used to quantify the degree of agreement between the operator SLA data and the comprehensive inferred value. The closer it is to 1, the higher the degree of agreement. The higher the degree of agreement, the more credible the first service provider's self-reported data is.

[0083] Reference formula: ;in, Represents consistency data. This represents the standardized SLA data of the primary service provider.

[0084] S3.6. Based on the consistency data, the credibility weight of the first service provider, and the dependency link probe data of each second service provider, calculate the adjustment amount of the credibility weight of the first service provider.

[0085] S3.7. Update the credibility weight of the first service provider based on the adjustment amount.

[0086] For example, if WS, as the primary service provider, has an initial deviation of 0.35, which is greater than the first threshold of 0.3, and the enterprise APM data of the database it relies on is 100ms, which is greater than the second threshold of 20ms, then the credibility weight of WS will be updated, and the inference value of the database for WS will be calculated. MQ's inferred value for WS Comprehensive inferred value Consistency data .

[0087] In some embodiments, such as Figure 4 As shown, based on consistency data, the credibility weight of the first service provider, and the dependency link probe data of each second service provider, the adjustment amount of the credibility weight of the first service provider is calculated, including:

[0088] S3.6.1. Subtract a preset benchmark value from the consistency data to obtain the basic deviation, referring to the formula: ;in, Represents the basic deviation. This represents the preset baseline value, such as 0.5. A positive value indicates that the primary service provider has high credibility. A negative value indicates that the reliability of the first service is low.

[0089] Multiply the credibility weight of the first service provider by the result of subtracting the credibility weight, and you get the sensitivity score, using the following formula: ;in, Represents sensitivity, The credibility weight represents the primary service provider.

[0090] The adjustment range is largest when the credibility of the first service provider is in the middle, and the adjustment automatically slows down when the credibility of the first service provider is extremely high or extremely low, to prevent excessive oscillation of weight or saturation, thereby ensuring that the change in credibility is gradual and reasonable.

[0091] The link reward / penalty coefficient is calculated based on the dependency link probe data of the second service provider.

[0092] The link reward / penalty coefficient is positively correlated with the quality of all dependent link probe data. The better the quality of all dependent link probe data, the clearer and more reliable the operator SLA data of the second service provider should be, and a higher confidence level should be given, i.e., the closer the link reward / penalty coefficient is to 1. The worse the quality of all dependent link probe data, the more likely the operator SLA data of the second service provider may be distorted or have reduced reference value, and a lower confidence level should be given, i.e., the closer the link reward / penalty coefficient is to 0.

[0093] S3.6.2. Multiply a base deviation, sensitivity, link reward / penalty coefficient and a preset learning rate constant to obtain the adjustment amount of the first service provider's credibility weight.

[0094] Reference formula: in, The amount of adjustment representing the weighting of the first service provider's credibility. The representative link reward / penalty coefficient. This represents the learning rate constant, used to control the overall step size of weight adjustments to adapt to different business scenarios and performance requirements.

[0095] A grid search combined with historical data backtesting can be used to determine this; specifically, a set of candidate values ​​should be preset within a reasonable numerical range. The value is determined by collecting complete SLA evaluation source data and manually verified weight adjustment records over a specific time period as a validation set. Using this validation set data, each candidate value is then evaluated. The value is calculated using the service provider's SLA evaluation dynamic business model implementation method, and each... The adjustment amount obtained at the given value is selected based on the degree of agreement between it and the adjustment amount manually confirmed in the validation set, with the highest degree of agreement being chosen. As a learning rate constant.

[0096] For example, if =0.3, the calculated basic deviation B=0.337, sensitivity S=0.25, link reward / penalty coefficient Then the adjustment amount of the credibility weight of the first service provider is calculated. .

[0097] In some embodiments, the link reward / penalty coefficient is calculated based on the dependency link probe data of the second service provider, including:

[0098] For each second service provider, the geometric mean of all individual quality scores is calculated as the comprehensive score of the second service provider, which reflects the quality of the second service provider's dependency link probe data.

[0099] Reference formula: ;in, This represents the overall score of the second service provider.

[0100] The geometric mean of the overall scores of all second-tier service providers can be used as the link reward / penalty coefficient, as shown in the following formula: , where n represents the number of second service providers.

[0101] For example, the calculated DB individual score MQ's individual score The calculated link reward / penalty coefficient K .

[0102] In some embodiments, the SLA achievement rate data of the first service provider is calculated based on the credibility weight and SLA assessment source data, including:

[0103] Based on the credibility weight of the first service provider, the operator's SLA data, enterprise end-to-end APM data, and third-party network quality data are weighted and integrated to obtain the first service provider's SLA achievement rate data. The SLA achievement rate data quantifies the degree of conformity between the service provider's actual service level and its promised SLA.

[0104] Reference formula: ;in, This represents the SLA achievement rate data for the first service provider after normalization. This represents the standardized value of the operator's SLA data. This represents the standardized value of the enterprise's end-to-end APM data. This represents the standardized value of third-party network quality data. represent The weight, i.e., the updated credibility weight, , Represent , The weights can be fixed empirical values, such as 0.8 and 0.6 respectively.

[0105] For example, if Web Application Service Provider 1, Web Application Service Provider 2, and Web Application Service Provider 3 are each designated as the first service provider, the SLA achievement rate data can be calculated. , , The scores are 0.828, 0.723, and 0.628 respectively. The ranking clearly shows that Web Application Service Provider 2 performed the best and Web Application Service Provider 3 performed the worst, which helps businesses quickly identify the priority of different service providers.

[0106] In some embodiments, such as Figure 5 As shown, this application provides a system for implementing a dynamic business model for service provider SLA evaluation, including:

[0107] Data acquisition module: Used to acquire SLA assessment source data from service providers. The SLA assessment source data includes operator SLA data, enterprise end-to-end APM data, third-party network quality data, and dependency link probe data between service providers.

[0108] Network construction module: It is used to set the trust weight of each node with service provider as node according to preset rules, determine the directed edges between nodes according to the preset dependency relationship between service providers, and determine the connection weight of directed edges according to dependency link probe data.

[0109] Game Theory Module: Used to calculate the initial deviation data of the service provider based on the SLA assessment source data; when the initial deviation data of the first service provider exceeds the preset first threshold, and the enterprise end-to-end APM data or third-party network quality data of at least one second service provider on which the first service provider depends exceeds the preset second threshold within the time window, the credibility weight of the first service provider is updated according to the credibility weight of the second service provider, the operator's SLA data, and the connection weight.

[0110] Evaluation module: Used to calculate the SLA achievement rate of the first service provider based on credibility weight and SLA evaluation source data.

[0111] Results generation module: Used to generate evaluation results based on SLA achievement rate data from multiple first-tier service providers.

[0112] This system for implementing a dynamic business model for service provider SLA evaluation has all the advantages of the method for implementing a dynamic business model for service provider SLA evaluation, and can automate the steps of implementing the method.

[0113] In some embodiments, this application provides a readable storage medium storing computer program instructions, which are read and executed by a processor to perform the steps of a service provider SLA evaluation dynamic business model implementation method.

[0114] It should be noted that, in this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0115] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory.

[0116] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for implementing a dynamic business model for service provider SLA evaluation, characterized in that, include: Obtain SLA assessment source data from service providers, including operator SLA data, enterprise end-to-end APM data, third-party network quality data, and dependency link probe data between service providers; Using service providers as nodes, a trust weight is assigned to each node according to preset rules, directed edges between nodes are determined according to preset dependencies between service providers, and the connection weight of directed edges is determined according to dependency link probe data. The service provider's preliminary deviation data is calculated based on the SLA assessment source data; when the first service provider's preliminary deviation data exceeds a preset first threshold, and the enterprise end-to-end APM data or third-party network quality data of at least one second service provider on which the first service provider depends exceeds a preset second threshold within a time window, the first service provider's credibility weight is updated based on the second service provider's credibility weight, the operator's SLA data, and the connection weight. Based on the aforementioned credibility weights and SLA assessment source data, calculate the SLA achievement rate data of the first service provider; The evaluation results are generated based on SLA achievement rate data from multiple first-tier service providers.

2. The method according to claim 1, characterized in that, Obtain dependency link probe data between service providers, including: Acquire performance metadata collected by probes deployed on communication links between service providers with business dependencies; Based on the performance metadata, dependent link probe data containing one-way latency, jitter, and packet loss rate is generated.

3. The method according to claim 1, characterized in that, Each node is assigned a credibility weight according to preset rules, including: For each node, the historical credibility weight of the node is retrieved from the preset evaluation history database and used as the credibility weight. If the historical credibility weight of the node is not found in the evaluation history database, a preset default weight value is assigned to the node as the credibility weight.

4. The method according to claim 1, characterized in that, The connection weights of directed edges are determined based on the dependency link probe data, including: The one-way latency, jitter, and packet loss rate in the dependent link probe data are respectively mapped to individual quality scores within a preset value range; The minimum value among the individual quality scores is taken as the connection weight of the directed edge.

5. The method according to claim 1, characterized in that, Calculate the initial deviation data for the service provider, including: The SLA assessment source data is mapped to a unified dimension to obtain standardized data; In the standardized data, the statistical differences between each service provider's operator SLA data, enterprise end-to-end APM data, and third-party network quality data are calculated to obtain preliminary deviation data.

6. The method according to claim 2, characterized in that, The credibility weight of the first service provider is updated based on the credibility weight of the second service provider, operator SLA data, and connection weight, including: For each second service provider, the inferred value of the second service provider's operator SLA data for the first service provider is calculated based on the operator SLA data and the corresponding connection weight. The weighted average of all inferred values ​​is calculated based on the credibility weight of each second service provider to obtain the comprehensive inferred value of the operator SLA data of the first service provider. Calculate the consistency data between the first service provider's operator SLA data and the comprehensive inferred value; Based on the consistency data, the credibility weight of the first service provider, and the dependency link probe data of each second service provider, calculate the adjustment amount of the credibility weight of the first service provider. The credibility weight of the first service provider is updated based on the adjustment amount.

7. The method according to claim 6, characterized in that, Based on the consistency data, the trust weight of the first service provider, and the dependency link probe data of each second service provider, the adjustment amount of the trust weight of the first service provider is calculated, including: Subtract a preset benchmark value from the consistency data to obtain the basic deviation; multiply the credibility weight of the first service provider by the result of subtracting the credibility weight to obtain the sensitivity; calculate the link reward / penalty coefficient based on the dependency link probe data of the second service provider. The adjustment amount of the first service provider's credibility weight is obtained by multiplying the basic deviation, sensitivity, link reward / penalty coefficient and a preset learning rate constant.

8. The method according to claim 7, characterized in that, Based on the dependency link probe data from the second service provider, the link reward / penalty coefficient is calculated, including: The one-way latency, jitter, and packet loss rate in the dependent link probe data are respectively mapped to individual quality scores within a preset value range; For each second service provider, the geometric mean of all the individual quality scores is calculated as the overall score of the second service provider; The geometric mean of the overall scores of all second service providers is taken as the link reward / penalty coefficient.

9. The method according to claim 1, characterized in that, Based on the aforementioned credibility weights and SLA assessment source data, the SLA achievement rate data of the first service provider is calculated, including: Based on the aforementioned credibility weights, the operator's SLA data, enterprise end-to-end APM data, and third-party network quality data are weighted and fused to obtain the SLA achievement rate data of the first service provider.

10. A system for implementing a dynamic business model for service provider SLA evaluation, characterized in that, include: Data acquisition module: Acquires SLA assessment source data from service providers, including operator SLA data, enterprise end-to-end APM data, third-party network quality data, and dependency link probe data between service providers; Network construction module: Taking service providers as nodes, it sets a trust weight for each node according to preset rules, determines the directed edges between nodes according to the preset dependency relationships between service providers, and determines the connection weight of the directed edges according to the dependency link probe data. Game Theory Module: Calculates the initial deviation data of the service provider based on the SLA assessment source data; when the initial deviation data of the first service provider exceeds the preset first threshold, and the enterprise end-to-end APM data or third-party network quality data of at least one second service provider on which the first service provider depends exceeds the preset second threshold within the time window, the credibility weight of the first service provider is updated based on the credibility weight of the second service provider, the operator's SLA data, and the connection weight. Evaluation module: Calculates the SLA achievement rate data of the first service provider based on the credibility weight and SLA evaluation source data; Results generation module: Generates evaluation results based on SLA achievement rate data from multiple first-tier service providers.

Images