Ehr system service coordination processing method and system based on xinchuang technology stack

Abstract

The application relates to the technical field of human resource information processing, and particularly discloses an eHR system service cooperative processing method and system based on an Xinchuang technology stack, which comprises the following steps: generating a health state vector based on the interface response time, connection success rate and disconnection frequency of the Xinchuang middleware, and calculating a dynamic fuse threshold; combing the core service dependency relationship of the eHR system to form a plurality of service dependency chains, determining the priority weight of each service dependency chain based on the business influence degree score result, and generating a three-dimensional correlation table in combination with the Xinchuang resource demand of each service link of each service dependency chain; calculating the real-time scheduling priority of each service dependency chain based on the three-dimensional correlation table, and allocating Xinchuang resources to each service dependency chain based on the real-time scheduling priority; when any service link initiates a cross-dependency chain calling request, determining the calling logic based on the health state vector and the dynamic fuse threshold until the calling is successful; and ensuring that the cooperative processing among the services of the eHR system is efficient and stable.

Description

Technical Field

[0001] This invention relates to the field of human resources information processing technology, and in particular to a method and system for collaborative processing of eHR system services based on the domestic IT innovation technology stack. Background Technology

[0002] In the era of digital office, enterprise human resource management (eHR) systems play a crucial role in the efficient operation of enterprises. With the continuous development of information technology, the domestic IT innovation technology stack has gradually emerged, aiming to achieve independent control over information technology, safeguard national information security, and promote industrial innovation and upgrading. Applying the domestic IT innovation technology stack to eHR systems can provide enterprises with a more secure, reliable, and independently controllable human resource management solution. With its unique advantages, such as a secure and reliable underlying architecture and an independently controllable technical system, the domestic IT innovation technology stack brings new development opportunities for eHR system service collaboration. Based on the domestic IT innovation technology stack, the eHR system service collaboration processing method and system have broad application prospects in the current context of accelerated digital transformation and increasing importance of data security. It is expected to become the mainstream direction for the future development of enterprise eHR systems, driving enterprise human resource management towards a more intelligent, efficient, and secure direction.

[0003] However, due to the interdependence of various service components and the volatile state of the domestic IT middleware, existing technologies struggle to achieve efficient collaborative processing between services. Specifically, this manifests as the inability to scientifically assess the health of the domestic IT middleware, manage service dependencies, and allocate resources rationally, leading to chaotic service scheduling and call failures. This solution achieves efficient collaboration of eHR system services through real-time monitoring, relationship analysis, priority determination, and optimization of resource allocation and invocation logic.

[0004] Therefore, this invention proposes a service collaborative processing method and system for eHR systems based on the domestic IT innovation technology stack. Summary of the Invention

[0005] This invention provides a method and system for collaborative processing of eHR system services based on the domestic IT innovation technology stack, effectively achieving efficient collaboration of eHR system services based on the domestic IT innovation technology stack. By collecting key indicators of the domestic IT innovation middleware in real time to determine its health status and circuit breaker thresholds, system stability is ensured; service dependencies are analyzed, and priority weights are determined using the analytic hierarchy process to generate a three-dimensional association table, achieving reasonable resource allocation; cross-dependency chain call logic is determined based on middleware status and thresholds, improving call success rate. The system service process is comprehensively optimized to ensure the stable and efficient operation of the eHR system.

[0006] This invention provides a service collaboration processing method for an eHR system based on the domestic IT innovation technology stack, comprising:

[0007] The status monitoring agent collects the interface response time, connection success rate, and disconnection frequency of the domestic IT middleware in real time, and generates a health status vector. The dynamic circuit breaker threshold is then calculated based on the health status vector.

[0008] The core service dependencies of the eHR system were analyzed to form multiple service dependency chains. The analytic hierarchy process (AHP) was used to score the business impact of each service dependency chain. Based on the scoring results, the priority weight of each service dependency chain was determined. In conjunction with the information technology innovation resource requirements of each service link of each service dependency chain, a three-dimensional relationship table was generated.

[0009] The real-time scheduling priority of each service dependency chain is calculated based on the three-dimensional association table, and the information technology innovation resources are allocated to each service dependency chain based on the real-time scheduling priority.

[0010] When any service component initiates a cross-dependency chain call request, the call logic is determined based on the health status vector and dynamic circuit breaker threshold until the call is successful.

[0011] Optionally, the dynamic circuit breaker threshold is calculated based on the health status vector, including:

[0012] Based on the influence weights of various indicators in the health status vector on various eHR businesses, an influence factor matrix is ​​constructed.

[0013] The business impact vector is obtained by multiplying the health status vector by the impact factor matrix;

[0014] Normalize the L2 norm of the business impact vector to obtain the health status of the domestic IT innovation middleware;

[0015] The dynamic circuit breaker threshold is determined based on the health status of the domestic IT middleware.

[0016] Optionally, the real-time scheduling priority of each service dependency chain is calculated based on the three-dimensional association table, including:

[0017] Based on the service demand resource quantity corresponding to the information technology innovation resource demand of each service link in each service dependency chain in the three-dimensional association table, the multi-dimensional resource demand value of each service dependency chain is determined, and a multi-dimensional resource demand vector is constructed based on the multi-dimensional resource demand value of each service dependency chain. At the same time, a multi-dimensional resource availability vector is constructed based on the available resource data of the information technology innovation resource pool.

[0018] The cosine similarity between the multidimensional resource demand vector and the multidimensional resource availability vector of each service dependency chain is used as the basic matching degree.

[0019] A pre-defined clustering algorithm is used to divide the basic matching degree of all service dependency chains into multiple clusters. Based on the Euclidean distance between the basic matching degree of each service dependency chain and the cluster center of its respective cluster, the cluster correction coefficient of each service dependency chain is determined.

[0020] The resource requirement matching degree of each service dependency chain is determined based on the basic matching degree and cluster correction coefficient of each service dependency chain.

[0021] Based on the resource demand matching degree of each service dependency chain and the priority weight of each service dependency chain in the three-dimensional association table, the real-time scheduling priority of each service dependency chain is calculated.

[0022] Optionally, based on the basic matching degree of each service dependency chain and the Euclidean distance to the cluster center of its respective cluster, a cluster correction coefficient for each service dependency chain is determined, including:

[0023] Normalize the Euclidean distance between all service dependency chains within the same cluster and the cluster center of the cluster to obtain the normalized distance of each service dependency chain.

[0024] Based on the correction coefficient function and the normalized distance of each service dependency chain, the preliminary correction coefficient of each service dependency chain is calculated.

[0025] A knowledge graph of eHR service dependencies is constructed by using service links as nodes, edges as dependencies, and labeling the strength of the associations.

[0026] The eHR service dependency knowledge graph is analyzed based on the shortest path method, and the inter-chain correlation degree between any two service dependency chains within the same cluster is calculated.

[0027] Based on the initial correction coefficients of each service dependency chain and the inter-chain correlation between any two service dependency chains, the intra-cluster consistency coefficient of each service dependency chain is calculated.

[0028] The initial correction coefficients are calibrated based on the intra-cluster consistency coefficients of each service dependency chain to obtain the cluster correction coefficients for each service dependency chain.

[0029] Optionally, the initial correction coefficients are calibrated based on the intra-cluster consistency coefficients of each service dependency chain to obtain the cluster correction coefficients for each service dependency chain, including:

[0030] When the intra-cluster consistency coefficient of the service dependency chain does not exceed the consistency coefficient threshold, the preliminary correction coefficient is calibrated based on the preset calibration formula to obtain the cluster correction coefficient of each service dependency chain.

[0031] When the intra-cluster consistency coefficient of a service dependency chain exceeds the consistency coefficient threshold, the preliminary correction coefficient of each service dependency chain is used as the cluster correction coefficient of each service dependency chain.

[0032] Optionally, resources for information technology innovation are allocated to each service dependency chain based on real-time scheduling priorities, including:

[0033] A two-dimensional scheduling basis vector is formed based on the real-time scheduling priority and resource demand matching degree of each service dependency chain. The two-dimensional basis vector is then mapped to the three-dimensional resource allocation space through affine transformation to obtain the three-dimensional spatial mapping point of each service dependency chain.

[0034] Define a planar function in the three-dimensional resource allocation space, and train the resource allocation hyperplane parameters using the support vector machine algorithm to obtain the resource allocation hyperplane function;

[0035] Based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, the resource allocation parameters of each service dependency chain are determined.

[0036] Based on the resource allocation parameters of each service dependency chain, allocate domestic IT innovation resources to each service dependency chain.

[0037] Optionally, resource allocation parameters for each service dependency chain are determined based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, including:

[0038] Based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, the ideal value of the relative allocation ratio of each service dependency chain relative to various types of information technology innovation resources is calculated.

[0039] Determine the boundary values ​​of the allocation ratio of each service dependency chain relative to various types of information technology innovation resources;

[0040] Based on the ideal value and boundary value of the relative allocation ratio of each service dependency chain to various types of information technology innovation resources, the resource allocation parameters of each service dependency chain are determined.

[0041] Optionally, the invocation logic is determined based on the health status vector and the dynamic circuit breaker threshold, including:

[0042] Obtain a large number of historical cross-dependency chain call examples, and generate a feature vector for each historical cross-dependency chain call example based on the health of the domestic middleware and the dynamic circuit breaker threshold of each historical cross-dependency chain call example. Generate the current feature vector based on the health of the domestic middleware and the dynamic circuit breaker threshold corresponding to the health status vector of the current service dependency chain.

[0043] In all historical cross-dependency chain call examples, select all historical cross-dependency chain call examples whose cosine similarity between the corresponding feature vector and the current feature vector is not less than a preset similarity threshold, and use them as reference historical cross-dependency chain call examples;

[0044] Based on the cosine similarity between the feature vectors of each reference history cross-dependency chain call example and the current feature vector, the relative reference weight of each reference history cross-dependency chain call example is calculated.

[0045] Based on the relative reference weights of each historical cross-dependency chain call example, the call results of all historical cross-dependency chain call examples are assigned values ​​and summed in a weighted manner to obtain the current predicted call success rate;

[0046] The call logic is determined based on the current predicted call success rate.

[0047] Optionally, the invocation logic is determined based on the current predicted invocation success rate, including:

[0048] If the current predicted call success rate is greater than the first success rate threshold, the normal call will be executed based on the preset retry interval as the call logic;

[0049] If the current predicted success rate of the call is not greater than the first success rate threshold but is greater than the second success rate threshold, the exponential backoff retry will be executed as the call logic.

[0050] If the current predicted call success rate is not greater than the second success rate threshold, the circuit breaker will be triggered directly and the backup service will be used as the call logic.

[0051] This invention provides an eHR system service collaborative processing system based on the domestic IT innovation technology stack, comprising:

[0052] The threshold determination module is used to collect the interface response time, connection success rate, and disconnection frequency of the domestic IT innovation middleware in real time based on the status monitoring agent, and generate a health status vector, and calculate the dynamic circuit breaker threshold based on the health status vector.

[0053] The relationship analysis module is used to analyze the core service dependencies of the eHR system, forming multiple service dependency chains. The analytic hierarchy process is used to score the business impact of each service dependency chain. Based on the scoring results, the priority weight of each service dependency chain is determined. Combined with the information technology innovation resource requirements of each service link of each service dependency chain, a three-dimensional relationship table is generated.

[0054] The resource allocation module is used to calculate the real-time scheduling priority of each service dependency chain based on the three-dimensional association table, and to allocate domestic IT innovation resources to each service dependency chain based on the real-time scheduling priority.

[0055] The cross-chain invocation module is used to determine the invocation logic based on the health status vector and dynamic circuit breaker threshold when any service link initiates a cross-dependency chain invocation request, until the invocation is successful.

[0056] The beneficial effects of this invention compared to existing technologies are as follows: By collecting key indicators of the domestically developed middleware in real time through a status monitoring agent to generate a health status vector and calculate a dynamic circuit breaker threshold, the health status of the middleware can be dynamically assessed, ensuring the stability of system operation. The core service dependencies of the eHR system are analyzed to form a service dependency chain. The Analytic Hierarchy Process (AHP) is used to score and determine priority weights, and a three-dimensional association table is generated based on the domestically developed resource requirements of each service link, which helps to clearly understand the relationships between services and resource needs. Real-time scheduling priorities are calculated and domestically developed resources are allocated based on the three-dimensional association table, achieving reasonable resource allocation and improving the overall operating efficiency of the system. When a service link initiates a cross-dependency chain call request, the call logic is determined based on the health status vector and the dynamic circuit breaker threshold, ensuring the reliability of the call and guaranteeing efficient and stable collaborative processing between various services of the eHR system, meeting the needs of enterprise human resource management.

[0057] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0058] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0059] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0060] Figure 1 This is a flowchart of the eHR system service collaborative processing method based on the domestic IT innovation technology stack in an embodiment of the present invention;

[0061] Figure 2 This is a flowchart illustrating the dynamic circuit breaker threshold calculation in an embodiment of the present invention.

[0062] Figure 3 This is a flowchart of the real-time scheduling priority calculation in an embodiment of the present invention. Detailed Implementation

[0063] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0064] like Figure 1 As shown, this invention provides an implementation method for a service collaborative processing method of an eHR system based on the domestic IT innovation technology stack, including:

[0065] The status monitoring agent collects the interface response time, connection success rate, and disconnection frequency of the domestic IT middleware in real time, and generates a health status vector. The dynamic circuit breaker threshold is then calculated based on the health status vector.

[0066] The core service dependencies of the eHR system were analyzed to form multiple service dependency chains. The analytic hierarchy process (AHP) was used to score the business impact of each service dependency chain. Based on the scoring results, the priority weight of each service dependency chain was determined. In conjunction with the information technology innovation resource requirements of each service link of each service dependency chain, a three-dimensional relationship table was generated.

[0067] The real-time scheduling priority of each service dependency chain is calculated based on the three-dimensional association table, and the information technology innovation resources are allocated to each service dependency chain based on the real-time scheduling priority.

[0068] When any service component initiates a cross-dependency chain call request, the call logic is determined based on the health status vector and dynamic circuit breaker threshold until the call is successful.

[0069] In this embodiment, the status monitoring agent is a module embedded in the domestic IT innovation middleware. It collects three core indicators of the domestic IT innovation middleware in real time at a frequency of 10 seconds / time, namely interface response time, connection success rate, and disconnection frequency, and transmits them to the eHR service collaboration layer through SM2 algorithm encryption.

[0070] In this embodiment, domestically developed middleware such as Eastcom and Kingdee Apusic are key components of the eHR system's service collaboration architecture, used to implement functions such as service invocation and data transfer.

[0071] In this embodiment, the interface response time refers to the time taken from sending a request to receiving a response when one service in the eHR system calls another service through the domestic IT innovation middleware, and the unit is ms.

[0072] In this embodiment, the connection success rate refers to the percentage of successful connections established between eHR system services through the domestic IT middleware out of the total number of connections.

[0073] In this embodiment, the disconnection frequency refers to the number of times the connection of the domestic IT middleware is interrupted per unit time (per minute).

[0074] In this embodiment, the dynamic circuit breaker threshold is a value that is dynamically adjusted based on the health of the domestic IT middleware, and is used to control the number of service call retries.

[0075] In this embodiment, the core service dependencies of the eHR system are analyzed to form multiple service dependency chains. This involves organizing and analyzing the interdependencies between the various core services in the eHR system, clarifying which services can only be performed after other services are completed, thus forming a series of sequential service chains. For example, attendance statistics → payroll calculation → individual income tax declaration → payroll disbursement is a typical service dependency chain, where payroll calculation depends on the results of attendance statistics, and individual income tax declaration depends on the data from payroll calculation, etc.

[0076] In this embodiment, the Analytic Hierarchy Process (AHP) is used to score the business impact of each service dependency chain, constructing a three-level scoring index system. The first-level indicators include business urgency (weight 0.4), the scope of affected employees (weight 0.3), and the core process relevance (weight 0.3). For the second-level indicators, business urgency is further subdivided into "response timeliness requirements" and "time window constraints"; the scope of affected employees is subdivided into "number of people involved" and "organizational level"; and the core process relevance is subdivided into "importance of process nodes" and "data relevance." Then, a 1-9 scale is used to compare and score each sub-indicator pairwise. After consistency checks, a comprehensive score (out of 100) is calculated for each service dependency chain to assess its importance to the business.

[0077] In this embodiment, the scoring result is a specific score obtained by scoring the business impact of each service dependency chain using the analytic hierarchy process (AHP), with a maximum score of 100. This score reflects the relative importance of each service dependency chain in the business; for example, the "Attendance → Payroll → Individual Income Tax" dependency chain scores 80 points, and the "Employee Information → Performance" dependency chain scores 50 points.

[0078] In this embodiment, the priority weight of each service dependency chain is determined based on the scoring results, and a three-dimensional association table is generated by combining the information technology innovation resource requirements of each service link in each service dependency chain. Specifically, the priority weight is obtained by dividing the scoring result by 100 (within the range of [0,1]), such as a priority weight of 0.8 for a score of 80. At the same time, the information technology innovation resource requirements of each service link in each dependency chain are marked, such as the payroll calculation service requiring ≥4 CPU cores and ≥8GB of memory, and the individual income tax declaration service requiring ≥100MB / s database IO, etc. Then, a three-dimensional association table is generated with the dependency chain identifier, priority weight, and resource requirement details as fields.

[0079] In this embodiment, a service dependency chain refers to a series of interconnected services in the eHR system that have a sequential order. The output of one service serves as the input of the next, together completing a specific business process in the eHR system. For example, a chain like "attendance statistics → payroll calculation → individual income tax declaration → salary payment".

[0080] In this embodiment, the real-time scheduling priority is a value calculated based on the priority weight of the service dependency chain and the resource demand matching degree. Using the formula P=W×R (W is the priority weight, and R is the resource demand matching degree), the resource scheduling module allocates domestically developed resources to service dependency chains with higher P values ​​based on this real-time scheduling priority, ensuring the smooth operation of high-priority service dependency chains and improving the service collaboration efficiency of the eHR system.

[0081] In this embodiment, a service link refers to a specific individual service in a service dependency chain, which is the basic unit that makes up the service dependency chain. For example, in the service dependency chain of "attendance statistics → payroll calculation → individual income tax declaration → salary payment", attendance statistics, payroll calculation, individual income tax declaration, and salary payment are all service links, and each service link has its specific functions and resource requirements.

[0082] like Figure 2 As shown, in order to determine the dynamic circuit breaker threshold that reflects the impact of middleware on business, a dynamic circuit breaker threshold calculation based on health status vectors is proposed, including:

[0083] Based on the influence weights of various indicators in the health status vector on various eHR businesses, an influence factor matrix is ​​constructed.

[0084] The business impact vector is obtained by multiplying the health status vector by the impact factor matrix;

[0085] Normalize the L2 norm of the business impact vector to obtain the health status of the domestic IT innovation middleware;

[0086] The dynamic circuit breaker threshold is determined based on the health status of the domestic IT middleware.

[0087] In this embodiment, an influence factor matrix is ​​constructed based on the impact weights of various indicators in the health status vector on different types of eHR services. Specifically, the health status vector consists of standardized values ​​of interface response time, connection success rate, and disconnection frequency. For different types of eHR services, such as real-time queries, batch calculations, and data synchronization, the degree of influence of each indicator varies. By analyzing and determining the impact weights of each indicator on various services, a 3×3 matrix is ​​formed. For example, the impact weight of interface response time on real-time query services is set to 0.6, on batch calculation services to 0.3, and on data synchronization services to 0.1; the impact weight of connection success rate on real-time query services is set to 0.2, on batch calculation services to 0.5, and on data synchronization services to 0.3; the impact weight of disconnection frequency on real-time query services is set to 0.3, on batch calculation services to 0.2, and on data synchronization services to 0.5. Combining these elements constructs the influence factor matrix.

[0088] .

[0089] In this embodiment, the L2 norm of the business impact vector is normalized to obtain the health score of the domestic IT innovation middleware. First, the health status vector (1×3) and the impact factor matrix (3×3) are multiplied by V×M to obtain a 3-dimensional business impact vector. Then, the L2 norm of the business impact vector is calculated and normalized by dividing it by √3 to obtain the health score of the domestic IT innovation middleware, which ranges from [0,1].

[0090] In this embodiment, a dynamic circuit breaker threshold is determined based on the health status of the domestic IT middleware. Specifically, a circuit breaker threshold function is defined. Let's calculate, where H is the middleware health. First, calculate the first derivative of the function. ,when (Right now When the health status is determined to be in a rapid decline phase, the circuit breaker threshold is set to an integer. And trigger an emergency warning; when (Right now When the circuit breaker threshold is an integer, the circuit breaker threshold is taken as an integer. ;when (Right now When the circuit breaker threshold is an integer, the circuit breaker threshold is taken as an integer. Furthermore, the second derivative is calculated simultaneously during threshold adjustment. Based on the property that the second derivative is always positive, an exponential smoothing method is used to predict the health trend over the next 5 minutes and adjust the threshold 5 seconds in advance. In this way, the circuit breaker threshold is dynamically adjusted according to the middleware health, making the service call retry mechanism more reasonable and avoiding excessive retries that increase the load when the middleware is unstable.

[0091] like Figure 3 As shown, in order to provide a basis for the rational allocation of information technology innovation resources, a method is proposed to calculate the real-time scheduling priority of each service dependency chain based on a three-dimensional association table, including:

[0092] Based on the service demand resource quantity corresponding to the information technology innovation resource demand of each service link in each service dependency chain in the three-dimensional association table, the multi-dimensional resource demand value of each service dependency chain is determined, and a multi-dimensional resource demand vector is constructed based on the multi-dimensional resource demand value of each service dependency chain. At the same time, a multi-dimensional resource availability vector is constructed based on the available resource data of the information technology innovation resource pool.

[0093] The cosine similarity between the multidimensional resource demand vector and the multidimensional resource availability vector of each service dependency chain is used as the basic matching degree.

[0094] A pre-defined clustering algorithm is used to divide the basic matching degree of all service dependency chains into multiple clusters. Based on the Euclidean distance between the basic matching degree of each service dependency chain and the cluster center of its respective cluster, the cluster correction coefficient of each service dependency chain is determined.

[0095] The resource requirement matching degree of each service dependency chain is determined based on the basic matching degree and cluster correction coefficient of each service dependency chain.

[0096] Based on the resource demand matching degree of each service dependency chain and the priority weight of each service dependency chain in the three-dimensional association table, the real-time scheduling priority of each service dependency chain is calculated.

[0097] In this embodiment, the available resource data of the domestic IT innovation resource pool refers to the relevant information of various resources that are currently actually available for use in the resource pool that supports the collaborative service of the eHR system under the domestic IT innovation environment. This includes specific values ​​of resources such as the number of CPU cores, memory capacity, database IO rate, and network bandwidth. These data are dynamic and will change as the various services in the eHR system use and release resources.

[0098] In this embodiment, the service requirement resource quantity refers to the specific value of the information technology innovation resource requirement marked for each service link of each service dependency chain in the three-dimensional association table. For example, the payroll calculation service requires ≥4 CPU cores and ≥8GB of memory. Here, 4 CPU cores and 8GB of memory are the service requirement resource quantity of the payroll calculation service link. Different service links have different service requirement resource quantities according to their business function characteristics.

[0099] In this embodiment, based on the resource requirements corresponding to the information technology innovation resource requirements of each service link in each service dependency chain in the three-dimensional association table, the multi-dimensional resource requirement value of each service dependency chain is determined, and a multi-dimensional resource requirement vector is constructed based on the multi-dimensional resource requirement value of each service dependency chain. This involves extracting resource requirement information for each service link in each service dependency chain from the three-dimensional association table, such as the number of CPU cores, memory capacity, database I / O rate, network bandwidth, etc., and integrating these different types of resource requirement values ​​together to form a multi-dimensional resource requirement value. Using these multi-dimensional resource requirement values ​​as elements, a vector is constructed in a certain order, which is the multi-dimensional resource requirement vector. For example, if the resource requirement of a certain link in a service dependency chain is 4 CPU cores, 8GB of memory, and 100MB / s database I / O, then the corresponding multi-dimensional resource requirement vector may be [4,8,100] (assuming only these three resources are considered and the order is fixed).

[0100] In this embodiment, a multi-dimensional resource availability vector is constructed based on the available resource data of the domestic IT innovation resource pool. This involves obtaining various currently available resource data from the domestic IT innovation resource pool, such as the number of CPU cores, memory capacity, database I / O rate, and network bandwidth. These available resource values ​​are then integrated according to the same order and rules as the multi-dimensional resource demand vector to construct a vector representing the current resource availability of the domestic IT innovation resource pool. For example, if the currently available resources in the domestic IT innovation resource pool are 6 CPU cores, 10GB of memory, and 120MB / s database I / O, then the multi-dimensional resource availability vector might be [6, 10, 120] (assuming only these three resources are considered and the order is consistent with the multi-dimensional resource demand vector).

[0101] In this embodiment, a preset clustering algorithm is used to divide the basic matching degree of all service dependency chains into multiple clusters. The preset clustering algorithm is K-means clustering. The basic matching degree data calculated for all service dependency chains is used as input, and this algorithm divides this basic matching degree data into different categories, i.e., multiple clusters. The purpose of clustering is to group service dependency chains with similar basic matching degrees into one category, so that different cluster correction coefficients can be determined for different categories of service dependency chains, thus more reasonably calculating the resource demand matching degree. For example, service dependency chains with basic matching degrees between 0.8 and 1.0 might be divided into high-matching-degree clusters, those between 0.6 and 0.8 into medium-matching-degree clusters, and those between 0.4 and 0.6 into low-matching-degree clusters, etc.

[0102] In this embodiment, the resource demand matching degree of each service dependency chain is determined based on the basic matching degree and the cluster correction coefficient of each service dependency chain. This is achieved by multiplying the basic matching degree and the cluster correction coefficient of each service dependency chain after obtaining the basic matching degree and the cluster correction coefficient of the corresponding cluster, thereby obtaining a resource demand matching degree that more accurately reflects the resource adaptation of each service dependency chain to the current information technology innovation resource pool.

[0103] In this embodiment, based on the resource demand matching degree of each service dependency chain and the priority weight of each service dependency chain in the three-dimensional association table, the real-time scheduling priority of each service dependency chain is calculated. The formula P=W×R (where W is the priority weight of each service dependency chain in the three-dimensional association table determined based on business impact score, and R is the resource demand matching degree of each service dependency chain) is used to calculate the real-time scheduling priority P for each service dependency chain. Based on this real-time scheduling priority, the resource scheduling module prioritizes allocating domestically developed resources to service dependency chains with higher P values, ensuring that high-priority service dependency chains with good resource compatibility receive resources first, thus guaranteeing efficient service collaboration within the eHR system.

[0104] like Figure 3As shown, in order to calibrate the initial correction coefficients to obtain cluster correction coefficients that more accurately reflect service dependencies, a method is proposed to determine the cluster correction coefficients for each service dependency chain based on the basic matching degree of each service dependency chain and the Euclidean distance of the cluster center of its respective cluster. This method includes:

[0105] Normalize the Euclidean distance between all service dependency chains within the same cluster and the cluster center of the cluster to obtain the normalized distance of each service dependency chain.

[0106] Based on the correction coefficient function and the normalized distance of each service dependency chain, the preliminary correction coefficient of each service dependency chain is calculated.

[0107] A knowledge graph of eHR service dependencies is constructed by using service links as nodes, edges as dependencies, and labeling the strength of the associations.

[0108] The eHR service dependency knowledge graph is analyzed based on the shortest path method, and the inter-chain correlation degree between any two service dependency chains within the same cluster is calculated.

[0109] Based on the initial correction coefficients of each service dependency chain and the inter-chain correlation between any two service dependency chains, the intra-cluster consistency coefficient of each service dependency chain is calculated.

[0110] The initial correction coefficients are calibrated based on the intra-cluster consistency coefficients of each service dependency chain to obtain the cluster correction coefficients for each service dependency chain.

[0111] In this embodiment, normalizing the Euclidean distance between all service dependency chains within the same cluster and the cluster center of the cluster is achieved by taking the ratio of a single Euclidean distance to the maximum value among all Euclidean distances as the normalized distance of each service dependency chain.

[0112] In this embodiment, the correction coefficient function is a distance-based function used to adjust the resource demand matching degree based on the distance between the service dependency chain and the cluster center. Its function form is as follows: ,in This is the normalized distance between the service dependency chain and the cluster center. The purpose of this function is to ensure that the closer a service dependency chain is to the cluster center, the closer its correction coefficient is to 1, resulting in a more significant improvement in resource demand matching and thus a more reasonable reflection of the resource adaptation of each service dependency chain.

[0113] In this embodiment, preliminary correction coefficients for each service dependency chain are calculated based on the correction coefficient function and the normalized distance of each service dependency chain. The normalized distance of each service dependency chain is then used to calculate the correction coefficients. Substitute into the correction coefficient function Thus, the preliminary correction coefficients were obtained.

[0114] In this embodiment, an eHR service dependency knowledge graph is constructed using service links as nodes, edges as dependencies, and the strength of these dependencies labeled. In the eHR system, each service link is considered a node, the dependencies between service links are represented as edges, and each edge is labeled with a dependency strength (range 0-10). For example, if the "Attendance Statistics" service link and the "Payroll Calculation" service link are dependent, and their relationship is close, the dependency strength of this edge can be labeled as 8. Constructing the knowledge graph in this way allows for a clear visualization of the dependencies and their closeness between service links in the eHR system.

[0115] In this embodiment, the eHR service dependency knowledge graph is analyzed based on the shortest path method, and the inter-chain correlation degree between any two service dependency chains within the same cluster is calculated. The shortest path method is an algorithm for finding the shortest path between two nodes in a graph structure, and it is applied here to the eHR service dependency knowledge graph. For any two service dependency chains within the same cluster, the algorithm calculates the shortest path length between the service link nodes they contain, and combines this with the edge correlation strength to calculate the inter-chain correlation degree between the two service dependency chains. For example, service dependency chain A contains nodes a1, a2, and a3, and service dependency chain B contains nodes b1, b2, and b3. The shortest path method is used to find the shortest path from a1 to b1, and the sum of the edge correlation strengths on the path divided by the path length (that is, the total number of smallest unit paths traversed) is the inter-chain correlation degree.

[0116] In this embodiment, based on the initial correction coefficient of each service dependency chain and the inter-chain correlation between any two service dependency chains, the intra-cluster consistency coefficient of each service dependency chain is calculated:

[0117] The consistency coefficient within a cluster of each service dependency chain is calculated by taking the sum of the products of the absolute differences between the inter-chain correlation coefficients of the two service dependency chains and the initial correction coefficients of those two service dependency chains, and the sum of the inter-chain correlation coefficients of the two service dependency chains within the same cluster.

[0118] To obtain cluster correction coefficients that better reflect reality and thus optimize resource scheduling, a method is proposed to calibrate the initial correction coefficients based on the intra-cluster consistency coefficients of each service dependency chain, thereby obtaining the cluster correction coefficients for each service dependency chain, including:

[0119] When the intra-cluster consistency coefficient of the service dependency chain does not exceed the consistency coefficient threshold, the preliminary correction coefficient is calibrated based on the preset calibration formula to obtain the cluster correction coefficient of each service dependency chain.

[0120] When the intra-cluster consistency coefficient of a service dependency chain exceeds the consistency coefficient threshold, the preliminary correction coefficient of each service dependency chain is used as the cluster correction coefficient of each service dependency chain.

[0121] In this embodiment, the consistency coefficient threshold is set to 0.9 to determine whether the initial correction coefficient of the service dependency chain needs to be calibrated.

[0122] When the intra-cluster consistency coefficient of the service dependency chain Time-triggered coefficient calibration, preset calibration formula is ,in This is the average correction factor within the cluster. These are the initial correction factors, obtained after calibration. This refers to the cluster correction coefficient for each service dependency chain. When At that time, the initial correction coefficient of each service dependency chain is used as the cluster correction coefficient of each service dependency chain.

[0123] To achieve a scientific and rational allocation of domestic IT innovation resources, a method is proposed to allocate domestic IT innovation resources to each service dependency chain based on real-time scheduling priorities, including:

[0124] A two-dimensional scheduling basis vector is formed based on the real-time scheduling priority and resource demand matching degree of each service dependency chain. The two-dimensional basis vector is then mapped to the three-dimensional resource allocation space through affine transformation to obtain the three-dimensional spatial mapping point of each service dependency chain.

[0125] Define a planar function in the three-dimensional resource allocation space, and train the resource allocation hyperplane parameters using the support vector machine algorithm to obtain the resource allocation hyperplane function;

[0126] Based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, the resource allocation parameters of each service dependency chain are determined.

[0127] Based on the resource allocation parameters of each service dependency chain, allocate domestic IT innovation resources to each service dependency chain.

[0128] In this embodiment, a two-dimensional scheduling basis vector is formed based on the real-time scheduling priority and resource demand matching degree of each service dependency chain. Real-time scheduling priority and resource demand matching degree are two key indicators for measuring the resource allocation priority of service dependency chains. These two indicators are combined into a two-dimensional vector, meaning each service dependency chain corresponds to a two-dimensional scheduling basis vector. ,in Indicates real-time scheduling priority. This indicates the degree of matching between resource demand and requirements.

[0129] In this embodiment, affine transformation is used to map each two-dimensional dependency vector to a three-dimensional resource allocation space, obtaining the three-dimensional spatial mapping points of each service dependency chain. Affine transformation is a geometric transformation that can map points in two-dimensional space to three-dimensional space while preserving properties such as linear parallelism. Here, for each two-dimensional scheduling dependency vector... Perform a specific affine transformation to convert it into a point in three-dimensional space. For example, through formula , z= This maps two-dimensional vectors to three-dimensional space, resulting in the three-dimensional spatial mapping points for each service dependency chain.

[0130] In this embodiment, a planar function is defined in the three-dimensional resource allocation space, and the resource allocation hyperplane parameters are trained using a support vector machine algorithm to obtain the resource allocation hyperplane function. A planar function is defined in the three-dimensional resource allocation space. ,in , , These are parameters to be determined. Support Vector Machine (SVM) is a machine learning algorithm that is trained using a large amount of known data samples (which may be historical resource allocation data and corresponding service dependency chain information) to find the hyperplane parameters that best distinguish different categories (e.g., which service dependency chains should be allocated resources first, and which can be allocated later). During training, the algorithm continuously adjusts... , , The value of is determined so that the hyperplane can accurately divide service dependency chains of different priorities. The resulting resource allocation hyperplane function is used to determine the relationship between the position of each service dependency chain in three-dimensional space and the resource allocation priority. For example, points above the hyperplane correspond to service dependency chains that receive priority in resource allocation.

[0131] In this embodiment, domestic IT resources are allocated to each service dependency chain based on the resource allocation parameters of each service dependency chain. After determining the resource allocation hyperplane function, the distance 'd' from the three-dimensional spatial mapping point of each service dependency chain to the hyperplane is calculated. Resources are allocated in ascending order of distance 'd', with smaller distances indicating higher allocation priority. After resource allocation, the density cluster centers of the allocated resource points in the three-dimensional space are calculated. When a new service dependency chain joins, it is preferentially allocated to the nearest cluster center region to reduce resource fragmentation. Based on these rules, the resource allocation parameters for each service dependency chain are determined, such as the specific resource allocation ratio or quantity for CPU cores, memory capacity, database I / O rate, etc. Then, the corresponding resources in the domestic IT resource pool are allocated to each service dependency chain according to these parameters to achieve efficient and reasonable resource allocation among service dependency chains in the eHR system, ensuring the smooth operation of system service collaboration.

[0132] To determine the resource allocation parameters for each service dependency chain used to guide the rational allocation of resources for information technology innovation, a method is proposed that determines the resource allocation parameters for each service dependency chain based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function. This method includes:

[0133] Based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, the ideal value of the relative allocation ratio of each service dependency chain relative to various types of information technology innovation resources is calculated.

[0134] Determine the boundary values ​​of the allocation ratio of each service dependency chain relative to various types of information technology innovation resources;

[0135] Based on the ideal value and boundary value of the relative allocation ratio of each service dependency chain to various types of information technology innovation resources, the resource allocation parameters of each service dependency chain are determined.

[0136] In this embodiment, the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function is typically the distance. The smaller the value, the higher the priority for that service dependency chain to obtain resources. Then, through a certain functional relationship, the distance... This is transformed into an ideal value representing the relative allocation proportion of various information technology innovation resources (such as CPU core count, memory capacity, database I / O rate, etc.). For example, a linear or non-linear function might be used for the transformation. Assuming a linear function is used, if the distance d ranges from [0,10], and the ideal value of the relative allocation proportion ranges from [0,1], the transformation formula can be set as follows: This allows us to calculate the ideal relative allocation ratio of each service dependency chain for each type of information technology innovation resource based on the distance, reflecting the proportion of each type of resource that each service dependency chain should obtain under ideal conditions.

[0137] In this embodiment, boundary values ​​for the allocation ratio of each service dependency chain relative to various types of domestic IT innovation resources are determined. These are limits set to ensure the rationality and feasibility of resource allocation. Considering factors such as the total limit of the domestic IT innovation resource pool, the basic resource requirements of each service dependency chain, and the stability of system operation, a lower limit and an upper limit for the allocation ratio are determined for each type of domestic IT innovation resource and each service dependency chain. For example, for the CPU resource allocation of a certain service dependency chain, the lower limit may be set to 60% of its demand to ensure that the service dependency chain can operate basically, and the upper limit may be set to 120% of its demand to prevent over-allocation of resources from causing waste or affecting other service dependency chains. These lower and upper limits are the allocation ratio boundary values, which constrain the range of resource allocation and avoid unreasonable resource allocation.

[0138] In this embodiment, resource allocation parameters for each service dependency chain are determined based on the ideal and boundary values ​​of the relative allocation ratio of each service dependency chain relative to various types of domestic IT innovation resources. First, the ideal relative allocation ratio is compared with the boundary values. If the ideal relative allocation ratio is within the lower and upper limits, it is directly used as the resource allocation parameter; if the ideal relative allocation ratio is less than the lower limit, the lower limit is used; and if the ideal relative allocation ratio is greater than the upper limit, the upper limit is used. For example, if the ideal relative allocation ratio of a service dependency chain for memory resources is calculated to be 0.8, and the set lower limit for the boundary values ​​is 0.7 and the upper limit is 0.9, then 0.8 is used as the allocation parameter for memory resources for that service dependency chain. This process is repeated to determine the allocation parameters for other types of domestic IT innovation resources for that service dependency chain. By comprehensively considering both ideal allocation conditions and actual constraints, the final resource allocation parameters for each service dependency chain are determined, thereby achieving reasonable and effective allocation of domestic IT innovation resources and ensuring the stable operation of the eHR system service collaboration.

[0139] To provide a quantitative basis for determining the invocation logic, a method based on health status vectors and dynamic circuit breaker thresholds is proposed, including:

[0140] Obtain a large number of historical cross-dependency chain call examples, and generate a feature vector for each historical cross-dependency chain call example based on the health of the domestic middleware and the dynamic circuit breaker threshold of each historical cross-dependency chain call example. Generate the current feature vector based on the health of the domestic middleware and the dynamic circuit breaker threshold corresponding to the health status vector of the current service dependency chain.

[0141] In all historical cross-dependency chain call examples, select all historical cross-dependency chain call examples whose cosine similarity between the corresponding feature vector and the current feature vector is not less than a preset similarity threshold, and use them as reference historical cross-dependency chain call examples;

[0142] Based on the cosine similarity between the feature vectors of each reference history cross-dependency chain call example and the current feature vector, the relative reference weight of each reference history cross-dependency chain call example is calculated.

[0143] Based on the relative reference weights of each historical cross-dependency chain call example, the call results of all historical cross-dependency chain call examples are assigned values ​​and summed in a weighted manner to obtain the current predicted call success rate;

[0144] The call logic is determined based on the current predicted call success rate.

[0145] In this embodiment, acquiring a large number of historical cross-dependency chain call examples refers to collecting numerous instances of cross-chain calls between different service dependency chains during the past operation of the eHR system. These instances cover various business scenarios, different states of the domestic IT middleware, and corresponding call results.

[0146] In this embodiment, a feature vector for each historical cross-dependency chain call example is generated based on the health of the domestically developed middleware and the dynamic circuit breaker threshold. For each historical cross-dependency chain call example, the health of the domestically developed middleware and the dynamic circuit breaker threshold are two key attributes. These two attributes are combined to form a vector, which characterizes the key features of the historical example.

[0147] In this embodiment, a current feature vector is generated based on the health status of the domestically developed middleware and the dynamic circuit breaker threshold corresponding to the health status vector of the current service dependency chain. Similar to generating feature vectors for historical cross-dependency chain call examples, a vector is constructed based on the health status of the domestically developed middleware and the dynamic circuit breaker threshold corresponding to the current service dependency chain. First, the health status of the domestically developed middleware is calculated from the health status vector, and then combined with the currently set dynamic circuit breaker threshold to form a feature vector that can represent the key conditions of the current service dependency chain call.

[0148] In this embodiment, the preset similarity threshold is a pre-set numerical value used to measure the degree of similarity between two vectors. In this embodiment, it is mainly used to determine whether the similarity between the feature vector of a historical cross-dependency chain call example and the current feature vector is high enough to determine whether the historical example can be used as a reference. For example, the preset similarity threshold is set to 0.7.

[0149] In this embodiment, the relative reference weight of each reference historical cross-dependency chain call example is calculated based on the cosine similarity between the feature vector of each reference historical cross-dependency chain call example and the current feature vector. Cosine similarity is a commonly used method to measure the similarity between two vectors. It reflects their similarity by calculating the cosine value of the angle between the two vectors, with a value ranging from [-1, 1]. The closer the value is to 1, the more similar the two vectors are. For each feature vector of a reference historical cross-dependency chain call example and the current feature vector, the cosine similarity between them is calculated. Then, based on these cosine similarities, the relative reference weight of each reference historical cross-dependency chain call example is determined through a certain calculation method. For example, a normalized method can be used, with the sum of the cosine similarities of all reference historical cross-dependency chain call examples as the denominator and the cosine similarity of each example as the numerator, to calculate the relative reference weight of each example.

[0150] In this embodiment, the call results of all reference historical cross-dependency chain call examples are assigned weighted sums based on the relative reference weights of each example to obtain the current predicted call success rate. A numerical value is assigned to the call result of each reference historical cross-dependency chain call example, for example, 1 for a successful call and 0 for a failed call. Then, the relative reference weight of each example is multiplied by its corresponding call result value, and all products are summed to obtain the current predicted call success rate.

[0151] To ensure the reliability and stability of cross-dependency chain calls, a call logic based on the current predicted call success rate is proposed, including:

[0152] If the current predicted call success rate is greater than the first success rate threshold, the normal call will be executed based on the preset retry interval as the call logic;

[0153] If the current predicted success rate of the call is not greater than the first success rate threshold but is greater than the second success rate threshold, the exponential backoff retry will be executed as the call logic.

[0154] If the current predicted call success rate is not greater than the second success rate threshold, the circuit breaker will be triggered directly and the backup service will be used as the call logic.

[0155] In this embodiment, the first success rate threshold is a pre-set probability value used to distinguish different call logics. When the current predicted call success rate calculated based on historical data and the current state is greater than this first success rate threshold, it indicates that the current call has a high probability of success, and the system will adopt a relatively conventional call method. In this embodiment, the first success rate threshold may be set to 0.8, that is, if the current predicted call success rate is greater than 0.8, the subsequent "execution of normal calls based on preset retry intervals as call logic" will be adopted.

[0156] In this embodiment, executing normal calls based on a preset retry interval is considered the call logic. This means that when the predicted call success rate is high (greater than a first success rate threshold), if the first call fails during the cross-chain call of the service dependency chain, the system will retry at a pre-set fixed time interval. For example, if the preset retry interval is 1 second, then after each failed call, it will wait 1 second before attempting the call again, until the call succeeds. This approach is suitable when the system considers the current call environment relatively stable and the probability of success high. It ensures call success through a simple retry mechanism while avoiding unnecessary consumption of system resources due to overly frequent or complex retry strategies.

[0157] In this embodiment, the second success rate threshold is also a pre-set probability value, which is less than the first success rate threshold. It is used to further refine the call logic. When the current predicted call success rate is not greater than the first success rate threshold but greater than the second success rate threshold, the system will adopt a relatively complex retry strategy. For example, the second success rate threshold may be set to 0.5, that is, when the predicted call success rate is between 0.5 (inclusive) and 0.8, "executing exponential backoff retries as call logic" will be executed.

[0158] In this embodiment, exponential backoff retries are treated as the invocation logic. This means that if the cross-chain call to the service-dependent chain fails on the first attempt, subsequent retry intervals will gradually increase exponentially. For example, the first retry interval is 2 seconds (2... 1 The second retry interval is 4 seconds (2) 2 The third retry interval is 8 seconds (2 3 This process continues (k being the number of retries). This exponential backoff method avoids putting excessive pressure on the system due to frequent retries when the success rate of the call is not very high. As the number of retries increases, the interval becomes longer, giving the system more time to handle potential problems. At the same time, it reduces the continuous occupation of system resources, thus balancing retry efficiency and system resource consumption to a certain extent.

[0159] In this embodiment, directly triggering the circuit breaker and activating the backup service is used as the calling logic. This means that when the current predicted call success rate is not greater than the second success rate threshold, the system considers the probability of the current call success to be extremely low. Continuing to retry may waste system resources and cannot guarantee call success. At this time, the system will immediately trigger the circuit breaker mechanism, suspend the call of the current service dependency chain, and activate the backup service to replace the original call. For example, when the predicted call success rate is less than or equal to 0.5, the current call attempt is directly stopped, and a backup service (such as calling the backup database service or cached data) is used instead, ensuring that eHR system service collaboration can continue and reducing business interruptions caused by the failure of the main service call.

[0160] This invention provides an implementation method for an eHR system service collaborative processing system based on the domestic IT innovation technology stack, comprising:

[0161] The threshold determination module is used to collect the interface response time, connection success rate, and disconnection frequency of the domestic IT innovation middleware in real time based on the status monitoring agent, and generate a health status vector, and calculate the dynamic circuit breaker threshold based on the health status vector.

[0162] The relationship analysis module is used to analyze the core service dependencies of the eHR system, forming multiple service dependency chains. The analytic hierarchy process is used to score the business impact of each service dependency chain. Based on the scoring results, the priority weight of each service dependency chain is determined. Combined with the information technology innovation resource requirements of each service link of each service dependency chain, a three-dimensional relationship table is generated.

[0163] The resource allocation module is used to calculate the real-time scheduling priority of each service dependency chain based on the three-dimensional association table, and to allocate domestic IT innovation resources to each service dependency chain based on the real-time scheduling priority.

[0164] The cross-chain invocation module is used to determine the invocation logic based on the health status vector and dynamic circuit breaker threshold when any service link initiates a cross-dependency chain invocation request, until the invocation is successful.

[0165] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A service collaborative processing method for an eHR system based on the domestic IT innovation technology stack, characterized in that, include: The status monitoring agent collects the interface response time, connection success rate, and disconnection frequency of the domestic IT middleware in real time, and generates a health status vector. The dynamic circuit breaker threshold is calculated based on the health status vector. The core service dependencies of the eHR system were analyzed to form multiple service dependency chains. The analytic hierarchy process (AHP) was used to score the business impact of each service dependency chain. Based on the scoring results, the priority weight of each service dependency chain was determined. In conjunction with the information technology innovation resource requirements of each service link of each service dependency chain, a three-dimensional relationship table was generated. The real-time scheduling priority of each service dependency chain is calculated based on the three-dimensional association table, and the information technology innovation resources are allocated to each service dependency chain based on the real-time scheduling priority. When any service component initiates a cross-dependency chain call request, the call logic is determined based on the health status vector and dynamic circuit breaker threshold until the call is successful. The calculation of real-time scheduling priority for each service dependency chain based on a three-dimensional association table includes: Based on the service demand resource quantity corresponding to the information technology innovation resource demand of each service link in each service dependency chain in the three-dimensional association table, the multi-dimensional resource demand value of each service dependency chain is determined, and a multi-dimensional resource demand vector is constructed based on the multi-dimensional resource demand value of each service dependency chain. At the same time, a multi-dimensional resource availability vector is constructed based on the available resource data of the information technology innovation resource pool. The cosine similarity between the multidimensional resource demand vector and the multidimensional resource availability vector of each service dependency chain is used as the basic matching degree. A pre-defined clustering algorithm is used to divide the basic matching degree of all service dependency chains into multiple clusters. Based on the Euclidean distance between the basic matching degree of each service dependency chain and the cluster center of its respective cluster, the cluster correction coefficient of each service dependency chain is determined. The resource requirement matching degree of each service dependency chain is determined based on the basic matching degree and cluster correction coefficient of each service dependency chain. Based on the resource demand matching degree of each service dependency chain and the priority weight of each service dependency chain in the three-dimensional association table, the real-time scheduling priority of each service dependency chain is calculated. The invocation logic, determined based on the health status vector and dynamic circuit breaker threshold, includes: Obtain a large number of historical cross-dependency chain call examples, and generate a feature vector for each historical cross-dependency chain call example based on the health of the domestic middleware and the dynamic circuit breaker threshold of each historical cross-dependency chain call example. Generate the current feature vector based on the health of the domestic middleware and the dynamic circuit breaker threshold corresponding to the health status vector of the current service dependency chain. In all historical cross-dependency chain call examples, select all historical cross-dependency chain call examples whose cosine similarity between the corresponding feature vector and the current feature vector is not less than a preset similarity threshold, and use them as reference historical cross-dependency chain call examples; Based on the cosine similarity between the feature vectors of each reference history cross-dependency chain call example and the current feature vector, the relative reference weight of each reference history cross-dependency chain call example is calculated. Based on the relative reference weights of each historical cross-dependency chain call example, the call results of all historical cross-dependency chain call examples are assigned and summed in a weighted manner to obtain the current predicted call success rate; The call logic is determined based on the current predicted call success rate.

2. The eHR system service collaborative processing method based on the domestic IT innovation technology stack according to claim 1, characterized in that, The dynamic circuit breaker threshold is calculated based on the health status vector, including: Based on the influence weights of various indicators in the health status vector on various eHR businesses, an influence factor matrix is ​​constructed. The business impact vector is obtained by multiplying the health status vector by the impact factor matrix; Normalize the L2 norm of the business impact vector to obtain the health status of the domestic IT innovation middleware; The dynamic circuit breaker threshold is determined based on the health status of the domestic IT middleware.

3. The eHR system service collaborative processing method based on the domestic IT innovation technology stack according to claim 1, characterized in that, Based on the basic matching degree of each service dependency chain and the Euclidean distance to the cluster center of its respective cluster, the cluster correction coefficient of each service dependency chain is determined, including: Normalize the Euclidean distance between all service dependency chains within the same cluster and the cluster center of the cluster to obtain the normalized distance of each service dependency chain. Based on the correction coefficient function and the normalized distance of each service dependency chain, the preliminary correction coefficient of each service dependency chain is calculated. An eHR service dependency knowledge graph is constructed using service links as nodes, edges as dependencies, and the strength of the association is labeled. The eHR service dependency knowledge graph is analyzed based on the shortest path method, and the inter-chain correlation degree between any two service dependency chains within the same cluster is calculated. Based on the initial correction coefficients of each service dependency chain and the inter-chain correlation between any two service dependency chains, the intra-cluster consistency coefficient of each service dependency chain is calculated. The initial correction coefficients are calibrated based on the intra-cluster consistency coefficients of each service dependency chain to obtain the cluster correction coefficients for each service dependency chain.

4. The eHR system service collaborative processing method based on the domestic IT innovation technology stack according to claim 3, characterized in that, The initial correction coefficients are calibrated based on the intra-cluster consistency coefficients of each service dependency chain to obtain the cluster correction coefficients for each service dependency chain, including: When the intra-cluster consistency coefficient of the service dependency chain does not exceed the consistency coefficient threshold, the preliminary correction coefficient is calibrated based on the preset calibration formula to obtain the cluster correction coefficient of each service dependency chain. When the intra-cluster consistency coefficient of a service dependency chain exceeds the consistency coefficient threshold, the preliminary correction coefficient of each service dependency chain is used as the cluster correction coefficient of each service dependency chain.

5. The eHR system service collaborative processing method based on the domestic IT innovation technology stack according to claim 1, characterized in that, Based on real-time scheduling priorities, domestic IT innovation resources are allocated to each service dependency chain, including: A two-dimensional scheduling basis vector is formed based on the real-time scheduling priority and resource demand matching degree of each service dependency chain. The two-dimensional basis vector is then mapped to a three-dimensional resource allocation space through affine transformation to obtain the three-dimensional spatial mapping point of each service dependency chain. Define a planar function in the three-dimensional resource allocation space, and train the resource allocation hyperplane parameters using the support vector machine algorithm to obtain the resource allocation hyperplane function; Based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, the resource allocation parameters of each service dependency chain are determined. Based on the resource allocation parameters of each service dependency chain, allocate domestic IT innovation resources to each service dependency chain.

6. The eHR system service collaborative processing method based on the domestic IT innovation technology stack according to claim 5, characterized in that, Based on the distance from the 3D spatial mapping points of each service dependency chain to the resource allocation hyperplane function, the resource allocation parameters of each service dependency chain are determined, including: Based on the distance from the three-dimensional spatial mapping point of each service dependency chain to the resource allocation hyperplane function, the ideal value of the relative allocation ratio of each service dependency chain relative to various types of information technology innovation resources is calculated. Determine the boundary values ​​of the allocation ratio of each service dependency chain relative to various types of information technology innovation resources; Based on the ideal value and boundary value of the relative allocation ratio of each service dependency chain to various types of information technology innovation resources, the resource allocation parameters of each service dependency chain are determined.

7. The eHR system service collaborative processing method based on the domestic IT innovation technology stack according to claim 1, characterized in that, The call logic is determined based on the current predicted call success rate, including: If the current predicted call success rate is greater than the first success rate threshold, the normal call will be executed based on the preset retry interval as the call logic; If the current predicted success rate of the call is not greater than the first success rate threshold but greater than the second success rate threshold, the exponential backoff retry will be executed as the call logic. If the current predicted call success rate is not greater than the second success rate threshold, the circuit breaker will be triggered directly and the backup service will be used as the call logic.

8. A service collaboration processing system for eHR systems based on the domestic IT innovation technology stack, characterized in that: The method for performing the eHR system service collaborative processing method based on the domestic IT innovation technology stack according to any one of claims 1 to 7 includes: The threshold determination module is used to collect the interface response time, connection success rate, and disconnection frequency of the domestic IT innovation middleware in real time based on the status monitoring agent, and generate a health status vector, and calculate the dynamic circuit breaker threshold based on the health status vector. The relationship analysis module is used to analyze the core service dependencies of the eHR system, forming multiple service dependency chains. The analytic hierarchy process is used to score the business impact of each service dependency chain. Based on the scoring results, the priority weight of each service dependency chain is determined. Combined with the information technology innovation resource requirements of each service link of each service dependency chain, a three-dimensional relationship table is generated. The resource allocation module is used to calculate the real-time scheduling priority of each service dependency chain based on the three-dimensional association table, and to allocate domestic IT innovation resources to each service dependency chain based on the real-time scheduling priority. The cross-chain invocation module is used to determine the invocation logic based on the health status vector and dynamic circuit breaker threshold when any service link initiates a cross-dependency chain invocation request, until the invocation is successful.

Images