Industrial production data business micro-service splitting method, medium and system

Abstract

The invention provides an industrial production data business micro-service splitting method, a medium and a system, and belongs to the technical field of computer systems of specific calculation models. The industrial production data business micro-service splitting method comprises the following steps of collecting business circulation data, service calling data and data interaction data of an industrial production data system, wherein the data are used for subsequent analysis and evaluation; constructing a micro-service coupling degree matrix and a micro-service cohesion degree matrix containing multi-dimensional evaluation indexes; calculating a service contribution value reflecting the business value based on the business flow data; establishing a micro-service evaluation model, and calculating an initial micro-service splitting scheme according to the service contribution value, the micro-service coupling degree matrix and the micro-service cohesion degree matrix; according to the method, the problem that industrial production data cannot be objectively quantified in a traditional micro-service splitting method can be solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer systems for specific computing models. Specifically, it relates to a method, medium, and system for splitting microservices of industrial production data services. Background Art

[0002] Industrial production data systems are an important part of modern manufacturing, undertaking key data collection, processing, and application tasks. With the rapid development of intelligent manufacturing and industrial Internet, such systems are facing increasingly complex business requirements and massive data scales. The traditional monolithic application architecture can no longer meet the requirements of system flexibility, scalability, and maintainability, and there is an urgent need to reconstruct using a microservices architecture.

[0003] The microservices architecture splits complex monolithic applications into multiple loosely coupled and highly cohesive small services. Each service focuses on completing specific business functions and interacts through lightweight communication mechanisms. This architecture has many advantages, such as fast deployment, independent scalability, high fault tolerance, and strong technology heterogeneity. However, how to reasonably split industrial production data systems into microservices is a key issue that needs in-depth study.

[0004] Traditional microservice splitting methods mainly rely on expert experience and heuristic rules, suffering from problems such as strong subjectivity, lack of quantitative standards, and difficulty in evaluating optimization effects. On the one hand, the complexity of business processes and the diversity of data characteristics make it difficult to accurately grasp the dependency relationships and internal connections between services based on manual experience alone. On the other hand, non-functional requirements such as system performance, resource utilization, and reliability also require more scientific evaluation methods to guide the division of microservices. Therefore, there is an urgent need for a data-driven and comprehensive evaluation microservice splitting method to improve the architecture quality of industrial production data systems. Summary of the Invention

[0005] In view of this, the present invention provides a method, medium, and system for splitting microservices of industrial production data services, which can solve the problem that traditional microservice splitting methods cannot objectively quantify industrial production data.

[0006] The present invention is implemented as follows:

[0007] The first aspect of the present invention provides a method for splitting microservices of industrial production data services, which includes collecting business process data, service call data, and data interaction data of the industrial production data system, and the data is used for subsequent analysis and evaluation; constructing a microservice coupling degree matrix and a microservice cohesion degree matrix containing multi-dimensional evaluation indicators; calculating a service contribution value reflecting business value based on the business process data; establishing a microservice evaluation model, and calculating an initial microservice splitting scheme according to the service contribution value, the microservice coupling degree matrix, and the microservice cohesion degree matrix; constructing a microservice optimization evaluation equation set composed of multiple equations for iterative optimization to obtain an optimized microservice splitting scheme; deploying the first batch of microservices and collecting performance data, resource data, and reliability data for evaluation; establishing an operation evaluation model containing multiple evaluation dimensions; using the operation evaluation model to calculate the score of each index of the first batch of microservices and determining whether it meets the preset threshold. If it meets, the deployment of the remaining microservices is completed to form a final microservice architecture system.

[0008] Among them, the business process data is used to obtain business response indicators, system throughput indicators, and business integrity indicators reflecting the business performance of the system. The service call data is used to obtain the computing resource occupancy, storage resource occupancy, and network resource occupancy reflecting the system resource usage. The data interaction data is used to obtain the data update frequency, data synchronization delay, data conflict ratio, and data consistency level reflecting the system data synchronization efficiency.

[0009] Further, the microservice coupling degree matrix evaluates the dependency relationship between microservices through service call frequency, data interaction volume, and interface complexity. The microservice cohesion degree matrix evaluates the tightness within the microservice through business similarity degree, data consistency requirement, and transaction integrity requirement.

[0010] Further, the service contribution value reflects the support degree for the business goal through the business contribution degree, reflects the improvement degree of the system performance through the performance contribution degree, and reflects the importance of data processing through the data contribution degree.

[0011] Further, the microservice evaluation model reflects the overall scale of the system through the total number of microservice modules, reflects the development workload through the microservice code scale, and reflects the business coverage through the microservice function points.

[0012] Further, in the microservice optimization evaluation equation set, the service granularity balance equation, the business value evaluation equation, the resource cost equation, and the data consistency equation are respectively used to evaluate the balance of microservice division, the system value gain, the operation and maintenance cost, and the data synchronization efficiency; the service granularity balance coefficient, the business value coefficient, the resource consumption coefficient, and the data consistency coefficient are calculated using the microservice optimization evaluation equation set as the optimization basis.

[0013] Furthermore, the service granularity balance equation evaluates the balance by the total number of microservice modules, the microservice code scale, and the number of microservice function points; the business value evaluation equation evaluates the system benefits by the service contribution value, business response metrics, system throughput metrics, and business integrity metrics; the resource cost equation evaluates the operation and maintenance costs by calculating the resource occupancy, storage resource occupancy, and network resource occupancy; the data consistency equation evaluates the data efficiency by the data update frequency, data synchronization latency, data conflict ratio, and data consistency level.

[0014] Furthermore, in the operation evaluation model, the performance evaluation function evaluates the system performance through the interface response time, system throughput, number of concurrent users, and service call link data; the resource evaluation function evaluates the resource usage through the CPU utilization rate, memory occupancy rate, disk I / O load, network bandwidth utilization rate, and resource scaling records; the reliability evaluation function evaluates the system reliability through the service availability data, fault recovery time, data consistency detection results, call success rate, and retry policy records.

[0015] The second aspect of the present invention provides a computer-readable storage medium, wherein program instructions are stored in the computer-readable storage medium, and when the program instructions run on a computer, they are used to execute the above-mentioned industrial production data service microservice splitting method.

[0016] The third aspect of the present invention provides an industrial production data service microservice splitting system, which includes the above-mentioned computer-readable storage medium.

[0017] Compared with the prior art, the beneficial effects of the industrial production data service microservice splitting method, medium, and system provided by the present invention are as follows:

[0018] 1. Objective quantification: This method introduces mathematical models such as the microservice coupling degree matrix, microservice cohesion degree matrix, and service contribution value, converting subjective experience into objective indicators. The coupling degree matrix not only considers the static call relationship but also introduces dynamic change characteristics, being able to more comprehensively reflect the dependence intensity between services. The cohesion degree matrix can predict the development trend of internal connections and identify potential splitting requirements in advance by introducing the second derivative of business similarity. The service contribution value comprehensively considers the supporting role of the service in business goals, system performance, and data processing, providing a strong basis for optimizing the splitting scheme.

[0019] 2. Multidimensional evaluation: This method constructs an optimization evaluation equation system including four dimensions: service granularity, business value, resource cost, and data consistency. Each dimension uses advanced mathematical tools. For example, the service granularity balance equation introduces the gradient modulus to evaluate the optimization space, and the business value evaluation equation uses triple integrals to consider the cumulative effect, significantly improving the comprehensiveness and accuracy of the evaluation.

[0020] 3. Dynamic optimization: By establishing an operation evaluation model, this method realizes the continuous optimization of the microservice deployment plan. The performance evaluation function introduces the Laplace operator to analyze the local balance of the response time, the resource evaluation function uses divergence to describe the resource distribution characteristics, and the reliability evaluation function uses surface integral to characterize the spatial distribution of availability, making the optimization process more intelligent and adaptive.

[0021] 4. Implementation strategy: This method adopts a progressive deployment plan, preferentially deploying microservices with high service contribution values, and evaluating the deployment effect through quantitative indicators to ensure the smooth transition of the system. At the same time, a service performance compensation parameter is introduced to dynamically adjust the deployment strategy of subsequent services, further improving the operability of the plan.

[0022] Generally speaking, compared with the traditional microservice splitting method, the solution proposed by the present invention is more objective, comprehensive and intelligent, providing an effective solution for the microservice transformation of industrial production data systems. Brief description of the drawings

[0023] Figure 1 It is a flowchart of a microservice splitting method for industrial production data services. Detailed implementation manners

[0024] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0025] As Figure 1 shown, it is a flowchart of a microservice splitting method for industrial production data services provided by the present invention. In this method, the following steps are included:

[0026] S10. Collect the business process data, service call data, and data interaction data of the industrial production data system, obtain the business response index, system throughput index, and business integrity index based on the business process data, obtain the computing resource occupancy, storage resource occupancy, and network resource occupancy based on the service call data, and obtain the data update frequency, data synchronization delay, data conflict ratio, and data consistency level based on the data interaction data;

[0027] S20. Construct a microservice coupling degree matrix, which includes service call frequency, data interaction volume, and interface complexity, and construct a microservice cohesion degree matrix, which includes business similarity degree, data consistency requirement, and transaction integrity requirement;

[0028] S30. Calculate the service contribution value of each service interface based on the business process data. The service contribution value includes the business contribution degree, the performance contribution degree, and the data contribution degree.

[0029] S40. Establish a microservice evaluation model based on the service contribution value, the microservice coupling degree matrix, and the microservice cohesion degree matrix, calculate the initial microservice splitting scheme of the industrial production data system, and obtain the total number of microservice modules, the microservice code scale, and the microservice function points.

[0030] S50. Construct a microservice optimization evaluation equation set. The microservice optimization evaluation equation set includes a service granularity balance equation, a business value evaluation equation, a resource cost equation, and a data consistency equation. Use the microservice optimization evaluation equation set to calculate the service granularity balance coefficient, the business value coefficient, the resource consumption coefficient, and the data consistency coefficient.

[0031] S60. Iteratively optimize the initial microservice splitting scheme based on the service granularity balance coefficient, the business value coefficient, the resource consumption coefficient, and the data consistency coefficient to obtain an optimized microservice splitting scheme.

[0032] S70. Deploy the first batch of microservices with the highest service contribution value in the optimized microservice splitting scheme, and collect the following data:

[0033] Performance data: interface response time, system throughput, concurrent user number, service call link data

[0034] Resource data: CPU utilization rate, memory occupancy rate, disk I / O load, network bandwidth usage rate, resource scaling record

[0035] Reliability data: service availability data, fault recovery time, data consistency detection result, call success rate, retry policy record

[0036] Establish an operation evaluation model. The operation evaluation model includes a performance evaluation function, a resource evaluation function, and a reliability evaluation function.

[0037] S80. Use the operation evaluation model to calculate the performance index score, the resource utilization rate score, and the reliability index score of the first batch of microservices, and determine whether they meet the preset thresholds. If they do not meet, return to step S60 for optimization. If they meet, generate service performance compensation parameters based on the operation evaluation model, and complete the deployment of the remaining microservices in the order of the service contribution value from high to low to form a final microservice architecture system.

[0038] The microservice optimization evaluation equation set includes a service granularity balance equation, a business value evaluation equation, a resource cost equation, and a data consistency equation.

[0039] The service granularity balance equation is used to calculate the balance degree of microservice partitioning. The inputs include the total number of microservice modules, the microservice code scale, and the microservice function points, and the output is the service granularity balance coefficient;

[0040] The business value evaluation equation is used to calculate the system value return. The inputs include the service contribution value, the business response index, the system throughput index, and the business integrity index, and the output is the business value coefficient;

[0041] The resource cost equation is used to evaluate the system operation and maintenance cost. The inputs include the computing resource occupancy, the storage resource occupancy, and the network resource occupancy, and the output is the resource consumption coefficient;

[0042] The data consistency equation is used to calculate the data synchronization efficiency. The inputs include the data update frequency, the data synchronization delay, the data conflict ratio, and the data consistency level, and the output is the data consistency coefficient.

[0043] The performance evaluation function is used to calculate the microservice performance score. The inputs include the interface response time, the system throughput, the number of concurrent users, and the service call link data, and the output is the performance index score;

[0044] The resource evaluation function is used to calculate the resource usage score. The inputs include the CPU utilization rate, the memory occupancy rate, the disk I / O load, the network bandwidth utilization rate, and the resource scaling record, and the output is the resource utilization rate score;

[0045] The reliability evaluation function is used to calculate the system reliability score. The inputs include the service availability data, the fault recovery time, the data consistency detection result, the call success rate, and the retry policy record, and the output is the reliability index score.

[0046] The following describes the specific implementation manners of the above steps in detail:

[0047] The specific implementation manner of step S10 is as follows: First, by collecting the business flow data, service call data, and data interaction data of the industrial production data system, key indicators reflecting the system performance and data characteristics are obtained. Among them, the business flow data is used to calculate the business response index, the system throughput index, and the business integrity index, and these indicators can reflect the business performance level of the system. The service call data is used to calculate the computing resource occupancy, the storage resource occupancy, and the network resource occupancy, and these indicators can reflect the resource usage situation of the system. The data interaction data is used to calculate the data update frequency, the data synchronization delay, the data conflict ratio, and the data consistency level, and these indicators can reflect the data synchronization efficiency of the system. These indicator data provide the basic support for the subsequent microservice evaluation and optimization.

[0048] The specific implementation of step S20 is as follows: Based on the data collected in step S10, two matrices are constructed to characterize the coupling degree and cohesion degree of microservices. The microservice coupling degree matrix reflects the dependency relationships between different microservices, where each matrix element c ij is composed of the following five factors: the service call frequency f ij , the data interaction volume d ij , the interface complexity k ij , the change rate of call frequency , and the change rate of data interaction volume . These five factors are weighted and integrated by assigning different weight coefficients α, β, γ, δ, η, reflecting the multi-dimensional dependency relationships between microservices. The microservice cohesion degree matrix reflects the tightness within a single microservice, where each matrix element h ij is composed of the following five factors: the business similarity s ij , the data consistency requirement t ij , the transaction integrity requirement p ij , the first derivative of business similarity , and the second derivative . These five factors are weighted and integrated by assigning different weight coefficients λ, μ, ν, ξ, ζ, reflecting the multi-dimensional correlation within the microservice.

[0049] The specific implementation of step S30 is as follows: Based on the business process data, calculate the service contribution value V i of each service interface, which reflects the importance of the service in the entire system. The service contribution value V i is composed of the following six factors: the business contribution degree B i , the performance contribution degree P i , the data contribution degree D i , the second-order time derivative of the business contribution degree , the second-order time derivative of the performance contribution degree , and the mixed partial derivative of the business contribution degree, performance contribution degree, and data contribution degree . These six factors are weighted and integrated by assigning different weight coefficients ω 1 , ω 2 , ω 3 , ω 4 , ω 5 , ω 6 , comprehensively reflecting the supporting role of the service in business goals, system performance, and data processing.

[0050] The specific implementation of step S40 is as follows: Based on the microservice coupling degree matrix and cohesion degree matrix constructed in step S20, and the service contribution value calculated in step S30, a microservice evaluation model is established. This model comprehensively reflects the overall scale, development workload, and business coverage of the system through three indicators: the total number of microservice modules M, the microservice code size S, and the microservice function point count F. When calculating the initial microservice splitting plan, it is necessary to balance the equilibrium of these three indicators and finally output the total number of microservice modules, the microservice code size, and the microservice function point count.

[0051] The specific implementation of step S50 is as follows: Construct a microservice optimization evaluation equation set, including the following 4 equations:

[0052] (1) Service granularity balance equation G: This equation is used to evaluate the equilibrium degree of microservice partitioning. The inputs include the total number of microservice modules M, the microservice code size S, and the microservice function point count F. Weight coefficients σ 1 , σ 2 , σ 3 of the total number of modules, code size, and function point count, as well as their associated change rates and gradient norm are considered, comprehensively taking into account the overall scale, development difficulty, and business coverage of the system.

[0053] (2) Business value evaluation equation B: This equation is used to evaluate the value gain of the system. The inputs include the service contribution value V i , business response index R, system throughput index T, and business integrity index C. Weight coefficients φ 1 , φ 2 , φ 3 , φ 4 of these indicators, as well as the mixed partial derivatives of the service contribution value with respect to the response index and throughput index, and the triple integral term ∫∫∫V(r, t, c)drdtdc, comprehensively evaluating the business value of the system.

[0054] (3) Resource cost equation E: This equation is used to evaluate the operation and maintenance cost of the system. The inputs include the computing resource occupancy CPU, storage resource occupancy MEM, and network resource occupancy NET. The logarithmic term ln(1 + ) is introduced in the equation to describe the non-linear characteristics of resource occupancy. At the same time, the resource gradient and the second-order partial derivatives of the CPU occupancy with respect to memory and network are considered to evaluate the overall resource overhead of the system.

[0055] (4) Data Consistency Equation D: This equation is used to evaluate the data synchronization efficiency of the system. The inputs include the data update frequency f, the data synchronization delay l, the data conflict ratio r, and the data consistency level q. The exponential function is adopted in the equation to describe the attenuation characteristics of the frequency and delay, and at the same time, the conflict rate r, the consistency level q, and the mixed partial derivatives of the frequency, delay, and conflict rate are introduced and the loop integral term ∮ C to comprehensively characterize the complex characteristics of data synchronization.

[0056] Through the microservice optimization evaluation equation set composed of the above 4 equations, the service granularity balance coefficient G, the business value coefficient B, the resource consumption coefficient E, and the data consistency coefficient D can be calculated, providing a basis for subsequent optimization iterations.

[0057] The specific implementation of step S60 is: Based on the service granularity balance coefficient G, the business value coefficient B, the resource consumption coefficient E, and the data consistency coefficient D calculated in step S50, the initial microservice splitting scheme obtained in step S40 is iteratively optimized. The specific approach is: First, evaluate the overall balance of the current scheme according to the service granularity balance coefficient G. If G deviates from the expected range, the total number of microservice modules M, the microservice code size S, and the microservice function point number F need to be adjusted to improve the balance of splitting. Second, evaluate the revenue level of the current scheme according to the business value coefficient B. If B is low, the service contribution value V of the key services with relatively high i value needs to be increased to enhance the overall value of the system. Third, evaluate the operation and maintenance cost of the current scheme according to the resource consumption coefficient E. If E is too high, the CPU occupancy of computing resources, the MEM occupancy of storage resources, and the NET occupancy of network resources need to be reduced to optimize the resource usage of the system. Finally, evaluate the data synchronization efficiency of the current scheme according to the data consistency coefficient D. If D is low, the data update frequency f, the data synchronization delay l, and the data conflict ratio r need to be reduced to improve the consistency of data processing. By continuously adjusting these key indicators until each evaluation indicator meets the preset target threshold, the optimized microservice splitting scheme can be obtained.

[0058] The specific implementation of step S70 is: First, deploy the service contribution value V in the optimized microservice splitting scheme in step S60 iThe highest first batch of microservices. During the deployment process, the following key performance data is collected: interface response time t, system throughput th, number of concurrent users u, and number of service call links ch; resource usage data: CPU utilization rate cpu, memory occupancy rate mem, disk I / O load io, network bandwidth usage rate bw, and resource scaling record s; and reliability data: service availability a, fault recovery time rt, data consistency detection result dc, call success rate sr, and retry policy score rp.

[0059] Based on these data, three evaluation functions are constructed:

[0060] (1) Performance evaluation function P: This function comprehensively considers the response time t, throughput th, number of concurrent users u, and number of call links ch, and introduces the partial derivative of throughput with respect to time and the Laplace operator of the response time to comprehensively evaluate the performance of the system.

[0061] (2) Resource evaluation function R: This function comprehensively considers the CPU utilization rate cpu, memory occupancy rate mem, disk I / O load io, and network bandwidth usage rate bw, and introduces the resource scaling index s, the change rate of CPU utilization and the divergence of CPU utilization to evaluate the resource usage of the system.

[0062] (3) Reliability evaluation function L: This function comprehensively considers the service availability a, fault recovery time rt, data consistency detection result dc, call success rate sr, and retry policy score rp, and introduces the product of the change rate of availability and the change rate of success rate and the surface integral of availability to evaluate the reliability performance of the system.

[0063] By running these three evaluation functions, the performance index score P, resource utilization score R, and reliability index score L of the first batch of microservices can be calculated to determine whether the preset performance, resource, and reliability thresholds are met. If not, return to step S60 for further optimization; if so, generate service performance compensation parameters according to the evaluation results, and complete the deployment of the remaining microservices in descending order of the service contribution value V i to form the final microservice architecture system.

[0064] The following details the calculation process or equations involved in the present invention.

[0065] The calculation process, matrices, and equations involved in microservice splitting are as follows:

[0066] 1. Microservice coupling degree matrix:

[0067]

[0068] Single - element calculation formula:

[0069]

[0070] where f ij is the call frequency; d ij is the data interaction volume; k ij is the interface complexity; is the call frequency change rate; is the data interaction volume change rate; α, β, γ, δ, η are weight coefficients and satisfy α + β + γ + δ + η = 1; f max , d max , k max are the maximum values of each index respectively.

[0071] 2. Cohesion matrix:

[0072]

[0073] Single - element calculation formula:

[0074]

[0075] where s ij is the business similarity; t ij is the data consistency requirement; p ij is the transaction integrity requirement; are the first - order and second - order derivatives of the business similarity respectively; λ, μ, ν, ξ, ζ are weight coefficients and satisfy λ + μ + ν + ξ + ζ = 1; s max , t max , p max are the maximum values of each index respectively.

[0076] 3. Service contribution value calculation:

[0077]

[0078] where B i is the business contribution degree; P i is the performance contribution degree; D i is the data contribution degree; is the second - order time derivative of the contribution degree; is the mixed partial derivative; ω 1 , ω 2 , ω 3 , ω 4 , ω 5 , ω 6 are weight coefficients and satisfy

[0079] 4. Microservice Optimization Evaluation Equation System:

[0080] (1) Service Granularity Balance Equation:

[0081]

[0082] In the formula, M is the total number of modules; S is the code scale; F is the function point number; M opt , S opt , F opt is the optimal value; is the associated change rate; is the gradient modulus; σ 1 , σ 2 , σ 3 , σ 4 , σ 5 is the weight coefficient; ∈ 1 is the error term, and the range is [0.1, 0.3].

[0083] (2) Business Value Evaluation Equation:

[0084]

[0085] In the formula, V i is the service contribution value; R is the response index; T is the throughput index; C is the integrity index; is the mixed partial derivative; ∫∫∫V(r, t, c)drdtdc is the triple integral term; φ 1 , φ 2 , φ 3 , φ 4 , φ 5 , φ 6 is the weight coefficient; ∈ 2 is the error term, and the range is [0.2, 0.4].

[0086] (3) Resource Cost Equation:

[0087]

[0088] In the formula, CPU, MEM, NET are the resource occupancy; CPU max , MEM max , NET max is the maximum capacity; is the resource gradient; is the mixed partial derivative; θ 1 , θ 2 , θ 3 , θ 4 , θ 5 is the weight coefficient; ∈3 is the error term, with a range of [0.15, 0.35].

[0089] (4) Data consistency equation:

[0090]

[0091] In the formula, f is the update frequency; l is the synchronization delay; r is the conflict rate; q is the consistency level; is the mixed partial derivative; is the loop integral; f 0 , l 0 , r 0 is the reference value; q max is the highest level; ψ 1 , ψ 2 , ψ 3 , ψ 4 , ψ 5 , ψ 6 is the weight coefficient; ∈ 4 is the error term, with a range of [0.1, 0.3].

[0092] 5. Operating evaluation model:

[0093] (1) Performance evaluation function:

[0094]

[0095] In the formula, t is the response time; th is the throughput; u is the number of concurrent users; ch is the number of call links; is the partial derivative of time with respect to throughput; is the Laplace operator; t 0 , th max , u max , ch max is the reference value; ρ 1 , ρ 2 , ρ 3 , ρ 4 , ρ 5 , ρ 6 is the weight coefficient; ∈ 5 is the error term, with a range of [0.2, 0.4].

[0096] (2) Resource evaluation function:

[0097]

[0098] In the formula, cpu, memm, io, bw are the resource utilization rates; s is the scaling index; is the rate of change; is the divergence; cpumax , mem max , io max , bw max is the threshold; η 1 , η 2 , η 3 , η 4 , η 5 , η 6 , η 7 is the weight coefficient; ∈ 6 is the error term, and the range is [0.1, 0.3].

[0099] (3) Reliability evaluation function:

[0100]

[0101] In the formula, a is the availability; rt is the recovery time; dc is the consistency detection result; sr is the success rate; rp is the retry score; is the product of change rates; is the area integral; a max , rt 0 , dc max , sr max , rp max is the reference value; ξ 1 , ξ 2 , ξ 3 , ξ 4 , ξ 5 , ξ 6 , ξ 7 is the weight coefficient; ∈ 7 is the error term, and the range is [0.15, 0.35].

[0102] 1. Derivation process of microservice coupling degree matrix:

[0103] The basic coupling degree is constructed using a linear weighted model, including three dimensions: call frequency, data interaction volume, and interface complexity. Among them, the call frequency fi j is obtained by statistically analyzing the system logs, specifically including the number of interface calls between services, the number of message queue interactions, etc.; the data interaction volume d ij is statistically obtained by analyzing database access logs, cache access records, etc., and the data volume size of read and write operations needs to be considered; the interface complexity k ij Based on the interface definition, static analysis is performed. Considering factors such as the number of parameters, the complexity of parameter types, and the complexity of return values, the complexity score can be calculated by weighted summation.

[0104] The dynamic characteristic analysis introduces the time dimension and reflects the system evolution characteristics by calculating the change rate of indicators. The change rate calculation uses the sliding time window method, and the window size is determined according to the business characteristics, usually taking 5 - 10 minutes. Within each time window, the change values of the call frequency and data interaction volume are calculated, and the exponential smoothing method is used to process the original data to reduce the impact of random fluctuations. The change rate is calculated using the difference method, that is

[0105] Normalization ensures the comparability of indicators with different dimensions and adopts the maximum value normalization method. The maximum value f of each indicator max ,d max ,k max is obtained through historical data statistics and is updated regularly to adapt to system changes. The weight coefficients α, β, γ, δ, η are determined by the analytic hierarchy process, and the business characteristics and optimization objectives need to be considered. The final weight values are determined by combining expert scoring and data verification.

[0106] 2. Derivation process of the cohesion matrix:

[0107] The basic similarity quantification adopts a multi-dimensional evaluation method. The business similarity s ij uses text analysis technology. First, the text such as the service business description document and interface description is preprocessed, including steps such as word segmentation and stop word removal. Then, the Word2Vec model is used to convert the text into vector representation. Finally, the cosine similarity between vectors is calculated as the business similarity index. The data consistency requirement t ij is based on data model analysis. Considering factors such as entity relationship strength and transaction boundaries, the connection strength between nodes is calculated by constructing a data dependency graph. The transaction integrity requirement p ij is determined by analyzing the business process and transaction definition, and characteristics such as the scope and isolation level of the transaction need to be considered.

[0108] The dynamic characteristic modeling introduces the first and second derivatives to capture the system evolution characteristics. The first derivative reflects the change speed of similarity and is calculated by the difference between adjacent time windows; the second derivative represents the change rate of the change speed and is used to predict the development trend. The selection of the time window needs to balance the requirements of real-time and stability. For a typical microservice system, an hourly time window can be selected.

[0109] Parameter standardization and weight determination adopt a data-driven method. The maximum values s of each indicator are determined through historical data analysis max ,t max ,p max, the weight coefficients λ, μ, ν, ξ, ζ are initially uniformly distributed and then optimized through machine learning methods. The gradient descent algorithm can be used to minimize the prediction error.

[0110] 3. Derivation process of service contribution value calculation:

[0111] The basic contribution degree evaluation adopts a multi-dimensional quantization method. The business contribution degree B i is calculated by analyzing factors such as the number of business processes covered by the service, the importance of business processes, and the execution frequency of business processes, and is quantified by weighted summation; the performance contribution degree P i is evaluated based on the performance optimization effect of the service, including indicators such as the improvement degree of response time, the increase ratio of throughput, and the optimization of resource utilization; the data contribution degree D i is determined according to the data value processed by the service, considering factors such as the amount of data, the data update frequency, and the importance of data.

[0112] The analysis of dynamic characteristics and interaction effects introduces higher-order derivative terms. The second-order time derivative is obtained through time series analysis, reflects the acceleration characteristics of the indicator, and is calculated by the second-order difference method. The mixed partial derivative evaluates the coupling relationship between indicators by constructing a cross-impact matrix, and the matrix elements are obtained through regression analysis of historical data.

[0113] The optimization of weight coefficients adopts machine learning methods. By constructing a training data set, including historical service evolution data and performance data, algorithms such as support vector regression (SVR) are used to train the model to optimize the weight coefficients ω 1 , ω 2 , ω 3 , ω 4 , ω 5 , ω 6 , and the loss function of the model needs to consider the balance between prediction accuracy and model complexity.

[0114] 4. Derivation process of microservice optimization evaluation equations:

[0115] The service granularity balance equation is constructed based on the optimization theory. First, the optimal number of modules M is determined through domain-driven design opt , considering factors such as business boundaries and team structures; the optimal code scale S is determined through code complexity analysis opt , using indicators such as cyclomatic complexity and dependency; the optimal function point number F is determined based on the principle of functional cohesion opt . The square term is designed to amplify the deviation and promote the optimization process to converge to the optimal value. The associated change rate is obtained through regression analysis and reflects the dependency relationship between indicators. The gradient norm For evaluating the distribution characteristics of the optimization space.

[0116] The business value evaluation equation adopts the cumulative effect analysis method. The basic value is accumulated obtained by calculating the service contribution value; the performance indicators R, T, and C are obtained by statistically analyzing the system monitoring data. The mixed partial derivative reflects the interactive influence between response time and throughput, and is obtained by analyzing the performance test data. The triple integral ∫∫∫V(r, t, c)drdtdc considers the distribution characteristics of value in the multi-dimensional space, and the integration domain is determined according to the system operation range.

[0117] Parameter optimization and error handling adopt the data-driven method. The weight coefficient is obtained by training the machine learning model, and the error term ∈ 1 , ∈ 2 is modeled by the normal distribution, and the parameters are determined by maximum likelihood estimation. The cross-validation method is used for model verification to ensure the reliability and stability of the evaluation results.

[0118] The microservice splitting of traditional industrial production data services mainly relies on expert experience and heuristic rules, and its specific implementation process is as follows: First, the domain expert divides the preliminary boundary according to business knowledge, mainly considering factors such as the relevance of business processes, the frequency of data access, and the organizational structure of the team; Second, the system architect conducts evaluation and adjustment based on the technical dimension, including aspects such as the complexity of the interface, the scale of the service, and the convenience of deployment; Finally, through the review and practical verification of the R & D team, the splitting plan is continuously iteratively optimized. This method has problems such as strong subjectivity, lack of quantitative standards, and difficulty in evaluating the optimization effect.

[0119] The microservice splitting method proposed by the present invention has the following significant advantages:

[0120] 1. Objective quantification: By introducing mathematical models such as the coupling degree matrix, cohesion degree matrix, and service contribution value, subjective experience is transformed into objective indicators. The coupling degree matrix not only considers the static call relationship but also introduces the dynamic change rate and can accurately reflect the actual dependence strength between services. The cohesion degree matrix can predict the development trend of business similarity by introducing the second-order derivative and can identify potential splitting requirements in advance.

[0121] 2. Multi-dimensional evaluation: A complete optimization evaluation equation set is constructed, including four dimensions: service granularity, business value, resource cost, and data consistency. Each dimension uses high-order mathematical tools. For example, the service granularity balance equation introduces the gradient to evaluate the optimization space, and the business value evaluation equation uses the triple integral ∫∫∫V(r, t, c)drdtdc to consider the cumulative effect, significantly improving the comprehensiveness and accuracy of the evaluation.

[0122] 3. Dynamic Optimization: Continuous optimization is achieved by running the evaluation model. The performance evaluation function introduces the Laplace operator to analyze the local balance of the response time, and the resource evaluation function uses divergence to describe the characteristics of resource distribution, and the reliability evaluation function adopts surface integral to characterize the spatial distribution of availability, making the optimization process more intelligent and adaptive.

[0123] 4. Implementation Strategy: Adopt a progressive deployment plan, prioritize the deployment of microservices with high service contribution values, and evaluate the deployment effect through quantitative indicators such as performance metric scores, resource utilization scores, and reliability metric scores to ensure a smooth transition of the system. At the same time, introduce service performance compensation parameters to dynamically adjust the deployment strategy of subsequent services.

[0124] Generally speaking, by mathematizing and modeling the microservice splitting process, the present invention effectively solves the problems of strong subjectivity and difficult quantification of optimization effects in traditional methods, provides a scientific methodological support for the microservice transformation of industrial production data business systems, and has important theoretical value and practical significance. This method is not only applicable to industrial production data business systems, but also can be extended to the microservice architecture design in other fields, and has broad application prospects.

[0125] The second aspect of the present invention provides a computer-readable storage medium, wherein program instructions are stored in the computer-readable storage medium, and when the program instructions run in a computer, they are used to execute the above-mentioned method for splitting microservices of industrial production data services.

[0126] The third aspect of the present invention provides a system for splitting microservices of industrial production data services, which includes the above-mentioned computer-readable storage medium.

[0127] The core idea of the present invention is to collect key data of the industrial production data system, construct a multi-dimensional evaluation model, scientifically evaluate and dynamically optimize the microservice splitting scheme to improve the architecture quality of the system. The specific technical principle is as follows:

[0128] First of all, the present invention collects system data from three dimensions: business process flow, service call, and data interaction to comprehensively reflect the business characteristics, resource usage, and data characteristics of the system. Among them, business process flow data is used to calculate business response metrics, system throughput metrics, and business integrity metrics that reflect system performance; service call data is used to calculate computing resource occupancy, storage resource occupancy, and network resource occupancy that reflect resource usage; data interaction data is used to calculate data update frequency, data synchronization delay, data conflict ratio, and data consistency level that reflect data synchronization efficiency. These metric data provide a basic basis for subsequent microservice evaluation and optimization.

[0129] Secondly, the present invention constructs a microservice coupling degree matrix and a microservice cohesion degree matrix to quantitatively describe the dependency relationships and internal connections among microservices. Among them, the microservice coupling degree matrix includes factors such as service call frequency, data interaction volume, interface complexity, etc., and introduces a dynamic change rate, which can more accurately reflect the actual dependency strength between services. The microservice cohesion degree matrix includes factors such as business similarity, data consistency requirements, and transaction integrity requirements, and introduces the second derivative of business similarity, which can predict the development trend of internal connections. These two matrices lay a mathematical foundation for the evaluation of microservice splitting schemes.

[0130] Thirdly, the present invention proposes a calculation model for service contribution value, comprehensively considering the support degree of the service to business goals, system performance, and data processing, and quantifying the importance of each service. This model not only includes static contribution degree indicators, but also introduces dynamic second derivatives and mixed partial derivatives, which can more comprehensively reflect the actual value of the service. The service contribution value provides the core parameters for the subsequent microservice evaluation model.

[0131] Finally, the present invention constructs an optimization evaluation equation set including four equations: service granularity balance, business value evaluation, resource cost, and data consistency. Among them, the service granularity balance equation evaluates the overall balance of the splitting scheme through the trade-off of the total number of microservice modules, code scale, and function points; the business value evaluation equation comprehensively considers service contribution value, business response, system throughput, and business integrity to evaluate the revenue level of the scheme; the resource cost equation evaluates the operation and maintenance overhead of the system from the perspectives of computing, storage, and network resources; the data consistency equation evaluates the efficiency of data processing from dimensions such as data update frequency, synchronization delay, conflict ratio, and consistency level. By dynamically optimizing these four equations, the microservice splitting scheme can be continuously improved, and the overall architecture quality of the system can be enhanced.

[0132] Compared with traditional methods, the technical solution of the present invention has the following key features:

[0133] 1. Make full use of system operation data, adopt the method of mathematical modeling, and quantitatively evaluate microservice splitting schemes from multiple dimensions, overcoming the limitations of relying on subjective experience.

[0134] 2. Construct an optimization evaluation equation set including higher-order partial derivatives and integral terms, which can comprehensively reflect the complex dependency relationships and internal connections among microservices, and improve the accuracy of evaluation.

[0135] 3. Through the strategy of combining continuous optimization and progressive deployment, realize the dynamic adjustment and smooth transition of the microservice architecture, and ensure the stable operation of the system.

[0136] A specific embodiment 1 of the present invention is provided below. The specific implementation methods of each step in this embodiment 1 are described in detail as follows:

[0137] The specific implementation method of step S10 is as follows: First, by collecting the business process data, service call data, and data interaction data of the industrial production data system, key indicators reflecting the system performance and data characteristics are obtained. Among them, the business process data is used to calculate the business response index R, the system throughput index T, and the business integrity index C. The business response index R reflects the processing speed of the system for business requirements, the system throughput index T reflects the overall production capacity of the system, and the business integrity index C reflects the coverage of the system for business processes. These indicators can be used to evaluate the business performance level of the system. The service call data is used to calculate the CPU occupancy of computing resources, the MEM occupancy of storage resources, and the NET occupancy of network resources. These indicators can reflect the resource usage of the system. The data interaction data is used to calculate the data update frequency f, the data synchronization delay l, the data conflict ratio r, and the data consistency level q. These indicators can reflect the data synchronization efficiency of the system. These indicator data provide a basic support for subsequent microservice evaluation and optimization.

[0138] The specific implementation method of step S20 is: Based on the data collected in step S10, two matrices are constructed to characterize the coupling degree and cohesion degree of microservices.

[0139] The microservice coupling degree matrix C reflects the dependency relationship between different microservices, where each matrix element c ij consists of the following 5 factors:

[0140]

[0141] Among them, f ij is the call frequency between microservice i and microservice j, d ij is the data interaction volume between microservice i and microservice j, k ij is the complexity of the interface between microservice i and microservice j, is the change rate of the call frequency, is the change rate of the data interaction volume. α, β, γ, δ, η are the weight coefficients of each factor, satisfying α + β + γ + δ + η = 1. f max , d max , k max are the maximum values of each indicator respectively. This coupling degree matrix not only considers the static call relationship but also introduces dynamic change characteristics, and can more comprehensively characterize the dependency strength between microservices.

[0142] The microservice cohesion degree matrix H reflects the tightness within a single microservice, where each matrix element h ij consists of the following 5 factors:

[0143]

[0144] where s ij is the business similarity between microservice i and microservice j, and t ij is the data consistency requirement between microservice i and microservice j, and p ij is the transaction integrity requirement between microservice i and microservice j. is the first derivative of the business similarity, is the second derivative of the business similarity. λ, μ, ν, ξ, ζ are the weight coefficients of each factor, satisfying λ + μ + ν + ξ + ζ = 1. s max , t max , p max are the maximum values of each index respectively. By introducing the dynamic change characteristics of the business similarity, this cohesion matrix can better predict the internal association trend of microservices.

[0145] The specific implementation of step S30 is: calculating the service contribution value V of each service interface based on the business flow data i , which reflects the importance of this service in the whole system. The service contribution value V i consists of the following 6 factors:

[0146]

[0147] where B i is the business contribution degree of service i, P i is the performance contribution degree of service i, and D i is the data contribution degree of service i. are the second-order time derivatives of the business contribution degree and the performance contribution degree respectively, is the mixed partial derivative of the business contribution degree with respect to the performance contribution degree and the data contribution degree. ω 1 , ω 2 , ω 3 , ω 4 , ω 5 , ω 6 are the weight coefficients of each factor, satisfying The business contribution degree B i reflects the support degree of service i for the system business goal, the performance contribution degree P i reflects the improvement degree of service i for the system performance, and the data contribution degree D i reflects the importance of service i for data processing. Combining these factors can comprehensively evaluate the overall contribution value of each service.

[0148] The specific implementation of step S40 is as follows: Based on the microservice coupling degree matrix C and cohesion degree matrix H constructed in step S20, and the service contribution value V calculated in step S30, i , a microservice evaluation model is established. This model comprehensively reflects the overall scale, development workload, and business coverage of the system through the following three indicators:

[0149] Total number of microservice modules M: Reflects the overall scale of the system.

[0150] Microservice code size S: Reflects the development workload of the system.

[0151] Number of microservice function points F: Reflects the business coverage of the system.

[0152] When calculating the initial microservice splitting scheme, it is necessary to balance the equilibrium of these three indicators. Therefore, the service granularity balance equation G is introduced:

[0153]

[0154] where M opt , S opt , F opt are the optimal values of M, S, and F respectively. Reflects the associated change rate between these three indicators, reflects their gradient modulus. σ 1 , σ 2 , σ 3 , σ 4 , σ 5 is the weight coefficient of each item, ∈ 1 is the error term, and its value range is [0.1, 0.3]. By optimizing G, the initial microservice splitting scheme can be obtained, and the total number of microservice modules M, microservice code size S, and number of microservice function points F can be output.

[0155] The specific implementation of step S50 is as follows: Construct a microservice optimization evaluation equation set, including the following four equations:

[0156] (1) Business value evaluation equation B:

[0157]

[0158] This equation is used to evaluate the value benefit of the system. The inputs include the service contribution value V i , business response index R, system throughput index T, and business integrity index C. φ 1 , φ 2 , φ 3 , φ 4 , φ 5 , φ 6is the weight coefficient for each item. is the mixed partial derivative of the service contribution value with respect to the response index and throughput index, and ∫∫∫V(r, t, c)drdtdc is the triple integral term of the service contribution value, ∈ 2 is the error term, and its value range is [0.2, 0.4].

[0159] (2) Resource cost equation E:

[0160]

[0161] This equation is used to evaluate the operation and maintenance cost of the system. The inputs include the CPU occupancy of computing resources, the MEM occupancy of storage resources, and the NET occupancy of network resources. θ 1 , θ 2 , θ 3 , θ 4 , θ 5 are the weight coefficients for each item. reflects the gradient of resource occupancy. reflects the mixed partial derivative of CPU occupancy with respect to memory and network, ∈ 3 is the error term, and its value range is [0.15, 0.35].

[0162] (3) Data consistency equation D:

[0163]

[0164] This equation is used to evaluate the data synchronization efficiency of the system. The inputs include the data update frequency f, the data synchronization delay l, the data conflict ratio r, and the data consistency level q. ψ 1 , ψ 2 , ψ 3 , ψ 4 , ψ 5 , ψ 6 are the weight coefficients for each item. reflects the mixed partial derivative of frequency, delay, and conflict rate. is the loop integral of frequency, ∈ 4 is the error term, and its value range is [0.1, 0.3].

[0165] (4) Service granularity balance equation G: This equation has been given in step S40.

[0166] Through the microservice optimization evaluation equation set composed of the above 4 equations, the business value coefficient B, the resource consumption coefficient E, the data consistency coefficient D, and the service granularity balance coefficient G can be calculated, providing a basis for subsequent optimization iterations.

[0167] The specific implementation of step S60 is as follows: Based on the four evaluation coefficients calculated in step S50, the initial microservice splitting scheme obtained in step S40 is iteratively optimized.

[0168] First, evaluate the overall balance of the current scheme according to the service granularity balance coefficient G. If G deviates from the expected range, it is necessary to adjust the total number of microservice modules M, the microservice code size S, and the microservice function point number F to improve the balance of splitting. Specifically, if G is too large, it means the overall system scale is too large and M needs to be reduced; if G is too small, it means the overall system scale is too small and M needs to be increased. At the same time, S and F also need to be appropriately adjusted to maintain the relative balance of the three indicators.

[0169] Secondly, evaluate the revenue level of the current scheme according to the business value coefficient B. If B is low, it means the overall business value of the system is not high, and it is necessary to increase the service contribution value V i For key services with relatively high 1 , φ 2 , φ 3 , φ 4 and other weight coefficients to highlight the contribution of key services.

[0170] Furthermore, evaluate the operation and maintenance cost of the current scheme according to the resource consumption coefficient E. If E is too high, it means the resource overhead of the system is too large, and it is necessary to reduce the CPU occupancy of computing resources, the MEM occupancy of storage resources, and the NET occupancy of network resources to optimize the resource usage of the system. The consumption of different resources can be controlled by adjusting θ 1 , θ 2 , θ 3 and other weight coefficients.

[0171] Finally, evaluate the data synchronization efficiency of the current scheme according to the data consistency coefficient D. If D is low, it means the data synchronization efficiency of the system is poor, and it is necessary to reduce the data update frequency f, the data synchronization delay l, and the data conflict ratio r to improve the consistency of data processing. The data synchronization indicators can be optimized by adjusting ψ 1 , ψ 2 , ψ 3 , ψ 4 and other weight coefficients.

[0172] By continuously adjusting these key indicators until each evaluation indicator meets the preset target threshold, the optimized microservice splitting scheme can be obtained.

[0173] The specific implementation of step S70 is as follows: First, deploy the service contribution value V in the optimized microservice splitting scheme in step S60 iThe top first batch of microservices. During the deployment process, the following key performance data is collected: Interface response time t: reflects the processing speed of the system; System throughput th: reflects the overall production capacity of the system; Concurrent user number u: reflects the bearing capacity of the system; Number of service call links ch: reflects the complexity of the system.

[0174] Resource usage data: CPU utilization rate cpu: reflects the usage of computing resources; Memory occupancy rate mem: reflects the usage of storage resources; Disk I / O load io: reflects the usage of disk resources; Network bandwidth usage rate bw: reflects the usage of network resources; Resource scaling record s: reflects the situation of resource elastic scaling;

[0175] Reliability data: Service availability a: reflects the stability of the system; Fault recovery time rt: reflects the fault tolerance of the system; Data consistency detection result dc: reflects the consistency of data processing; Call success rate sr: reflects the robustness of the system; Retry policy score rp: reflects the self-healing ability of the system.

[0176] Based on these data, three evaluation functions are constructed:

[0177] Performance evaluation function P:

[0178]

[0179] This function comprehensively considers the response time t, throughput th, concurrent user number u, and number of call links ch, and introduces the partial derivative of throughput with respect to time and the Laplace operator of response time to comprehensively evaluate the performance of the system. ρ 1 ρ 2 ρ 3 ρ 4 ρ 5 ρ 6 are the weight coefficients of each item, ∈ 5 is the error term, and its value range is [0.2, 0.4].

[0180] Resource evaluation function R:

[0181]

[0182] This function comprehensively considers the CPU utilization rate cpu, memory occupancy rate memm, disk I / O load io, and network bandwidth usage rate bw, and introduces the resource scaling index s, the change rate of CPU utilization rate and the divergence of CPU utilization rate to evaluate the resource usage of the system. η 1 η 2 η3 , η 4 , η 5 , η 6 , η 7 are the weight coefficients of each item, ∈ 6 is the error term, and its value range is [0.1, 0.3].

[0183] Reliability evaluation function L:

[0184]

[0185] This function comprehensively considers service availability a, fault recovery time rt, data consistency detection result dc, call success rate sr, and retry policy score rp, and introduces the product of the availability change rate and the success rate change rate as well as the area integral of availability to evaluate the reliability performance of the system. ξ 1 , ξ 2 , ξ 3 , ξ 4 , ξ 5 , ξ 6 , ξ 7 are the weight coefficients of each item, ∈ 7 is the error term, and its value range is [0.15, 0.35].

[0186] The following provides a specific Embodiment 2 of the present invention. The specific implementation methods of each step in this Embodiment 2 are described in detail as follows:

[0187] A manufacturing enterprise has an industrial production data system to support its intelligent manufacturing business. This system is responsible for key functions such as real-time data collection, quality monitoring, and production plan management during the production process. With the continuous expansion of the business scale and the continuous improvement of the intelligent level, the system architecture faces the need for upgrading and transformation. After full evaluation, the enterprise decides to reconstruct the original monolithic application architecture into a microservices architecture to improve the flexibility, scalability, and maintainability of the system.

[0188] According to the microservices splitting method for industrial production data services proposed by the present invention, the enterprise established a microservices splitting team composed of business experts, system architects, and R & D leaders, and began the specific implementation work.

[0189] First, the team conducted an in-depth analysis of the business processes, service calls, and data interactions of the existing system and collected relevant operation data. By analyzing the business process data, the team calculated the business response index R, system throughput index T, and business integrity index C of the system. Taking the business response index R as an example, this index reflects the processing speed of the system for production tasks. According to historical data statistics, the average value of this index is 85 ms, the maximum value is 120 ms, and the minimum value is 55 ms. Similarly, the average value of the system throughput index T is 3000 units / minute, the maximum value is 4200 units / minute, and the minimum value is 2100 units / minute; the average value of the business integrity index C is 92%, the maximum value is 97%, and the minimum value is 85%.

[0190] By analyzing the service call data, the team calculated the CPU usage of the computing resources, MEM usage of the storage resources, and NET usage of the network resources of the system. Taking the CPU usage of the computing resources as an example, according to the monitoring data, the average CPU utilization rate of the production management service is 65%, the highest can reach 85%, and the lowest is 45%; the average CPU utilization rate of the quality control service is 72%, the highest is 92%, and the lowest is 52%; the average CPU utilization rate of the equipment management service is 78%, the highest is 95%, and the lowest is 60%. Similarly, the MEM usage of the storage resources and the NET usage of the network resources also show differences among different services.

[0191] In addition, by analyzing the data interaction data, the team calculated the data update frequency f, data synchronization delay l, data conflict ratio r, and data consistency level q of the system. Taking the data update frequency f as an example, the data of the production management service is updated on average once every 10 seconds, the highest can reach once every 5 seconds, and the lowest is once every 20 seconds; the data of the quality control service is updated on average once every 15 seconds, the highest can reach once every 8 seconds, and the lowest is once every 25 seconds; the data of the equipment management service is updated on average once every 12 seconds, the highest can reach once every 7 seconds, and the lowest is once every 18 seconds. Similarly, the data synchronization delay l, data conflict ratio r, and data consistency level q also show differences among different services.

[0192] Based on the above data, the team constructed the microservice coupling matrix C and the microservice cohesion matrix H.

[0193] The microservice coupling matrix C is shown in the following table:

[0194] Table 1 Microservice coupling matrix C

[0195] Production Management Quality Control Equipment Management Inventory Management Order Management Production Management - 0.72 0.63 0.48 0.39 Quality Control 0.72 - 0.57 0.41 0.33 Equipment Management 0.63 0.57 - 0.35 0.28 Inventory Management 0.48 0.41 0.35 - 0.21 Order Management 0.39 0.33 0.28 0.21 -

[0196] It can be seen from this matrix that the coupling degree between the production management service and the quality control service is the highest, reaching 0.72; while the coupling degree of the order management service with other services is relatively low. This is mainly because production management and quality control are closely related and need to exchange data frequently, while order management is relatively independent and only needs to interact with other services occasionally.

[0197] The microservice cohesion matrix H is shown in the following table:

[0198] Table 2 Microservice Cohesion Matrix H

[0199] Production Management Quality Control Equipment Management Inventory Management Order Management Production Management - 0.85 0.78 0.61 0.52 Quality Control 0.85 - 0.72 0.55 0.47 Equipment Management 0.78 0.72 - 0.49 0.41 Inventory Management 0.61 0.55 0.49 - 0.35 Order Management 0.52 0.47 0.41 0.35 -

[0200] It can be seen from this matrix that the cohesion degree of the production management service is the highest, reaching 0.85, indicating that the business processes, data relationships, and transaction characteristics within this service are relatively tight and suitable as an independent microservice. The cohesion degree of the order management service is relatively low, only 0.52, indicating that the business functions of this service are relatively independent and can be considered to be split into smaller-grained microservices.

[0201] After constructing the microservice coupling degree matrix and cohesion degree matrix, the team further calculated the service contribution value V of each service i . Taking the production management service as an example, its business contribution degree B i is 0.78, reflecting the key position of this service in the entire production process; the performance contribution degree P i is 0.72, indicating that this service has a greater role in improving the overall performance of the system; the data contribution degree D i is 0.68, indicating that this service is relatively important for data processing. According to the formula After weighted synthesis, the final service contribution value V of the production management service i is 0.75. Similarly, the team calculated the service contribution values of other services, which are: quality control service 0.72, equipment management service 0.68, inventory management service 0.53, and order management service 0.48.

[0202] With the above basic data, the team began to construct a microservice evaluation model. First, according to the overall scale of the system, development workload, and business coverage, the team determined the following three indicators: the total number of microservice modules M = 18, the microservice code scale S = 320,000 lines, and the microservice function points F = 275. Then, the team established the service granularity balance equation G:

[0203]

[0204] According to the calculation, the service granularity balance coefficient G of the current solution is 0.35, slightly lower than the expected target range of 0.4 - 0.6.

[0205] Next, the team established the business value evaluation equation B:

[0206]

[0207] According to the calculation, the business value coefficient B of the current solution is 0.68, which basically meets the expected target of 0.6 - 0.8.

[0208] The team also established the resource cost equation E:

[0209]

[0210]

[0211] According to the calculation, the resource consumption coefficient E of the current solution is 0.55, which is slightly higher than the expected target of 0.4 - 0.5.

[0212] Finally, the team established the data consistency equation D:

[0213]

[0214] According to the calculation, the data consistency coefficient D of the current solution is 0.72, which basically meets the expected target of 0.6 - 0.8.

[0215] Through the comprehensive analysis of the above 4 evaluation equations, the team found the following problems with the current microservice splitting solution:

[0216] 1. The balance of service granularity is slightly lacking. It is necessary to appropriately increase the total number of microservice modules, and at the same time adjust the code scale and function points to improve the overall balance.

[0217] 2. The resource consumption is slightly higher than expected. It is necessary to further optimize the CPU occupancy of computing resources, the MEM occupancy of storage resources, and the NET occupancy of network resources to reduce the operation and maintenance costs of the system.

[0218] 3. The data consistency basically meets the standard, but it is still necessary to further reduce the data update frequency, synchronization delay, and conflict ratio to improve the data processing efficiency.

[0219] Based on the above analysis, the team optimized and iterated the microservice splitting solution. First, by increasing the total number of microservice modules M to 20, appropriately adjusting the code scale S to 350,000 lines and the function points F to 300, the service granularity balance coefficient G was increased to 0.45, meeting the expected target. Second, by optimizing the CPU occupancy of computing resources, the MEM occupancy of storage resources, and the NET occupancy of network resources, the resource consumption coefficient E was reduced to 0.48, meeting the expected target. Finally, by reducing the data update frequency f, synchronization delay l, and conflict ratio r, the data consistency coefficient D was increased to 0.75, further improving the data processing efficiency of the system.

[0220] After multiple rounds of iterative optimization, the team finally determined the following microservice splitting plan:

[0221] 1. Production Management Microservice: Service contribution value V i = 0.75, which is a key service for the first batch of deployments; responsible for functions such as production plan management, production process monitoring, and equipment scheduling; has a relatively high coupling degree and a strong cohesion degree with the Quality Control Microservice and the Equipment Management Microservice;

[0222] 2. Quality Control Microservice: Service contribution value V i = 0.72, which is a sub-key deployment service; responsible for functions such as process parameter monitoring, product quality inspection, and non-conforming product handling; has a relatively high coupling degree and a strong cohesion degree with the Production Management Microservice;

[0223] 3. Equipment Management Microservice: Service contribution value V i = 0.68, which is a sub-key deployment service; responsible for functions such as equipment status monitoring, equipment fault diagnosis, and equipment maintenance plan; has a relatively high coupling degree and a strong cohesion degree with the Production Management Microservice;

[0224] 4. Inventory Management Microservice: Service contribution value V i = 0.53, which is a regular deployment service; responsible for functions such as raw material procurement, inventory query, and inbound / outbound management; has a relatively low coupling degree and cohesion degree with other services;

[0225] 5. Order Management Microservice: Service contribution value V i = 0.48, which is a regular deployment service; responsible for functions such as order acceptance, order tracking, and shipping management; has a relatively low coupling degree and cohesion degree with other services.

[0226] According to the above plan, the team first deployed the Production Management Microservice and the Quality Control Microservice with the highest service contribution values, and collected performance data, resource data, and reliability data.

[0227] In terms of performance data, the average interface response time t of the Production Management Microservice is 80ms, the highest is 105ms, and the lowest is 60ms; the average system throughput th is 3500 units / minute, the highest is 4100 units / minute, and the lowest is 2800 units / minute; the average number of concurrent users u is 180 people, the highest is 230 people, and the lowest is 130 people; the average number of service call links ch is 12, the highest is 17, and the lowest is 8. The performance data of the Quality Control Microservice also basically meets the expected goals.

[0228] In terms of resource data, the average CPU utilization rate (cpu) of the production management microservice is 72%, the highest is 85%, and the lowest is 55%; the average memory occupancy rate (memm) is 68%, the highest is 80%, and the lowest is 50%; the average disk I / O load (io) is 65%, the highest is 75%, and the lowest is 45%; the average network bandwidth utilization rate (bw) is 55%, the highest is 70%, and the lowest is 40%. The resource utilization data of the quality control microservice also basically meets the expected goals.

[0229] In terms of reliability data, the average service availability (a) of the production management microservice is 97%, and the lowest is 94%; the average fault recovery time (rt) is 15 minutes, the longest is 25 minutes, and the shortest is 10 minutes; the data consistency detection result (dc) is excellent; the average call success rate (sr) is 93%, and the lowest is 88%; the average retry policy score (rp) is 85 points, the highest is 95 points, and the lowest is 75 points. The reliability data of the quality control microservice also basically meets the expected goals.

[0230] According to the above evaluation results, the overall performance of the production management microservice and the quality control microservice meets the expectations and satisfies the target thresholds of performance, resources, and reliability. Subsequently, the team deployed the device management microservice with the second-highest service contribution value and continuously evaluated and optimized it in the same way as the previous two services.

[0231] During the subsequent deployment process, the team also introduced service performance compensation parameters to dynamically adjust the deployment strategies of subsequent services. For example, for the inventory management microservice with a relatively low service contribution value but excellent performance, its deployment priority can be appropriately increased; while for the order management microservice with a relatively high service contribution value but low performance indicators, more resource configurations need to be added to improve its overall performance. Through this flexible deployment strategy, a highly optimized microservice architecture system is finally formed, meeting the business needs of the enterprise in the field of intelligent manufacturing.

[0232] It should be noted that the variables involved in the present invention are explained in detail in Tables 3 and 4 as follows:

[0233] Table 3 Variable Explanation Table (1)

[0234]

[0235]

[0236] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. A method for splitting industrial production data business microservices, characterized in that: The method comprises the following steps: collecting business flow data, service call data, and data interaction data of the industrial production data system, and the data is used for subsequent analysis and evaluation; constructing a microservice coupling matrix and a microservice cohesion matrix containing multi-dimensional evaluation indicators; Calculate the service contribution value that reflects the business value based on business flow data; Establish a microservice evaluation model and calculate the initial microservice splitting plan based on the service contribution value, microservice coupling matrix and microservice cohesion matrix; Construct a microservice optimization evaluation equation group composed of multiple equations and perform iterative optimization to obtain the optimized microservice splitting plan; Deploy the first batch of microservices and collect performance data, resource data, and reliability data for evaluation; Establish an operation evaluation model that includes multiple evaluation dimensions; use the operation evaluation model to calculate the scores of various indicators of the first batch of microservices and determine whether they meet the preset thresholds. If so, complete the deployment of the remaining microservices to form the final microservice architecture system.

2. The method for splitting industrial production data business microservices according to claim 1, characterized in that: The business flow data is used to obtain business response indicators, system throughput indicators, and business integrity indicators that reflect the system business performance; the service call data is used to obtain computing resource occupancy, storage resource occupancy, and network resource occupancy that reflect the system resource usage; the data interaction data is used to obtain data update frequency, data synchronization delay, data conflict ratio, and data consistency level that reflect the system data synchronization efficiency.

3. The method for splitting industrial production data business microservices according to claim 2, characterized in that: The microservice coupling matrix evaluates the dependency between microservices through service call frequency, data interaction volume, and interface complexity, and the microservice cohesion matrix evaluates the tightness within microservices through business similarity, data consistency requirements, and transaction integrity requirements.

4. The method for splitting industrial production data business microservices according to claim 3 is characterized in that: The service contribution value reflects the degree of support for business objectives through the business contribution degree, reflects the degree of improvement in system performance through the performance contribution degree, and reflects the importance of data processing through the data contribution degree.

5. The method for splitting industrial production data business microservices according to claim 4, characterized in that: The microservice evaluation model reflects the overall scale of the system through the total number of microservice modules, reflects the development workload through the scale of microservice codes, and reflects the business coverage through the number of microservice function points.

6. The method for splitting industrial production data business microservices according to claim 5, characterized in that: The service granularity balance equation, business value evaluation equation, resource cost equation, and data consistency equation in the microservice optimization evaluation equation group are respectively used to evaluate the balance of microservice division, system value benefit, operation and maintenance cost, and data synchronization efficiency; the microservice optimization evaluation equation group is used to calculate the service granularity balance coefficient, business value coefficient, resource consumption coefficient, and data consistency coefficient as the optimization basis.

7. The method for splitting industrial production data business microservices according to claim 6, characterized in that: The service granularity balance equation evaluates the partition balance through the total number of microservice modules, the scale of microservice code, and the number of microservice function points; the business value evaluation equation evaluates the system benefit through the service contribution value, business response index, system throughput index, and business integrity index; the resource cost equation evaluates the operation and maintenance cost through the computing resource occupancy, storage resource occupancy, and network resource occupancy; the data consistency equation evaluates data efficiency through data update frequency, data synchronization delay, data conflict ratio, and data consistency level.

8. The method for splitting industrial production data business microservices according to claim 7, characterized in that: The performance evaluation function in the operation evaluation model evaluates system performance through interface response time, system throughput, number of concurrent users, and service call link data; the resource evaluation function evaluates resource usage through CPU utilization, memory occupancy, disk I / O load, network bandwidth utilization, and resource scaling records; the reliability evaluation function evaluates system reliability through service availability data, fault recovery time, data consistency detection results, call success rate, and retry strategy records.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and when the program instructions are executed in a computer, they are used to execute the industrial production data business microservice splitting method according to any one of claims 1 to 8.

10. An industrial production data business microservice splitting system, characterized in that: A computer-readable storage medium comprising the computer-readable storage medium of claim 9.

Images