Microservice cluster global call chain construction method and system based on edge computing

Abstract

The present invention proposes a method and system for constructing a global call chain of a microservice cluster based on edge computing, and relates to the field of communication technology. The method includes: obtaining a microservice cluster topology map, locating edge nodes based on data sources; collecting call chain data in the microservice cluster topology map, and transmitting it to edge nodes; the edge nodes use a trained decision tree model to filter out high-priority data in the call chain data for edge computing, build local call chains, and upload them to the central node; the high-priority data includes abnormal requests and high-latency requests; the central node aggregates local call chains to form a global call chain. The present invention deploys a lightweight machine learning model on the edge node, and based on historical data analysis, filters the data that constitutes the local call chain, and then the central node receives and merges the local call chains from each edge node to form a complete global call chain view, effectively reducing the burden of data transmission.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular to a method and system for constructing a global call chain of a microservice cluster based on edge computing. Background Art

[0002] The microservice architecture meets the complex requirements of the power information system for high scalability, flexibility, and stability through the decomposition of highly cohesive and low-coupled service units. However, as the number of microservices increases, the call relationship between the various services in the system becomes more complex. In order to ensure the efficient operation of the system, it is necessary to grasp the call links between these services in real time and intuitively display the relationship between each service by building a service topology diagram. This not only helps the operation and maintenance management of the system, but also can quickly locate the root cause of the problem when the system is abnormal.

[0003] The existing solution mainly tracks the call information on the microservice link, transmits it to the information collection module through http or gRPC, and then connects the corresponding microservices of the related upstream and downstream calls in real time to form a directed graph. However, this method relies on centralized processing and analysis during the construction process, resulting in centralized consumption of computing resources and network delay problems, and there is a large delay in the data transmission process, which greatly increases the transmission cost.

[0004] Edge computing is an emerging computing model that can reduce the burden on the central nodes of the power information system and improve the response speed and data processing capabilities. However, the inventors found that the current method of using edge computing to build a microservice call chain has a lot of invalid and redundant data on the edge nodes due to the imperfect data filtering mechanism, which greatly hinders data transmission and increases the load on the central node. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a method and system for constructing a global call chain of a microservice cluster based on edge computing. By deploying a lightweight machine learning model on the edge node, the data that constitutes the local call chain is screened based on historical data analysis, and then the central node receives and merges the local call chains from each edge node to form a complete global call chain view, which effectively reduces the burden of data transmission, thereby efficiently providing a comprehensive and real-time service call relationship view for the operation and maintenance management of the power information system, greatly reducing computing resource consumption and network delay problems.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for constructing a global call chain of a microservice cluster based on edge computing, comprising:

[0008] Obtain the microservice cluster topology map and locate edge nodes based on the data source;

[0009] Collect call chain data in the microservice cluster topology diagram and transmit it to the edge node;

[0010] The edge node uses the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; the high-priority data includes abnormal requests and high-latency requests;

[0011] The central node aggregates local call chains to form a global call chain.

[0012] Preferably, the locating edge nodes based on data sources is specifically: deploying edge nodes inside data sources, and the data sources include Internet of Things devices, user terminals and local data centers.

[0013] Preferably, the collection of call chain data in the microservice cluster topology diagram and transmission to the edge node is specifically as follows:

[0014] Based on the call chain data collected by the central node and edge node in the microservice cluster topology diagram, the microservice call relationship is obtained and a preliminary directed graph is constructed;

[0015] According to the preliminary directed graph, the call chain data between microservice units is collected using proxy tools or tracing codes embedded in microservices and transmitted to the edge nodes.

[0016] Preferably, the call chain data includes logs, requests, response times and error information.

[0017] Preferably, the edge node uses the trained decision tree model to filter out high-priority data in the call chain data for edge computing, builds a local call chain, and uploads it to the central node; specifically including:

[0018] Deploy the trained decision tree model on edge nodes;

[0019] The edge node uses the trained decision tree model to analyze and predict the call chain data, performs edge computing on the data predicted to be high priority, and builds it into a local call chain and uploads it to the central node.

[0020] Preferably, the training process of the decision tree model includes:

[0021] Obtain historical call chain data of the edge node, wherein the historical call chain data includes high priority data and other request data;

[0022] Annotating the historical call chain data to distinguish high-priority data from other request data;

[0023] At each decision tree node, a preferred feature is selected from multiple features included in the call chain data to divide the data, wherein the preferred feature is a feature that can achieve the best effect of data segmentation according to a preset evaluation standard; the data is divided by recursively selecting the best feature according to the evaluation standard to construct a decision tree; wherein each node in the decision tree represents a test of a certain feature, each branch represents a value of a feature, and the final leaf node contains the prediction result;

[0024] The constructed decision tree is trained using the annotated historical call chain data to obtain a trained decision tree model, so that the model can predict whether the data is high-priority data based on the characteristics of the call chain data.

[0025] Preferably, the central node aggregates local call chains to form a global call chain, which specifically includes:

[0026] The central node receives local call chains from different edge nodes, wherein the local call chains include ParentSpan ID information;

[0027] According to the Parent Span ID information, the calling relationship between services or operations is identified, and a corresponding directed edge is generated, where the direction of the directed edge is from the caller to the callee;

[0028] For each directed edge, a corresponding call chain path is constructed based on the identified call relationship;

[0029] Integrate multiple directed edge data to form a global call chain;

[0030] When a new local call chain arrives, the central node dynamically updates the global call chain.

[0031] In a second aspect, the present invention provides a microservice cluster global call chain construction system based on edge computing, comprising:

[0032] Initialization module, used to obtain the microservice cluster topology map and locate edge nodes based on the data source;

[0033] The call chain data acquisition module is used to collect the call chain data in the microservice cluster topology diagram and transmit it to the edge node;

[0034] The local call chain construction module is used for edge nodes to use the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; the high-priority data includes abnormal requests and high-latency requests;

[0035] The global call chain building module is used by the central node to aggregate local call chains to form a global call chain.

[0036] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in a method for constructing a global call chain of a microservice cluster based on edge computing as described in the first aspect.

[0037] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in a method for constructing a global call chain of a microservice cluster based on edge computing described in the first aspect are implemented.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) The present invention achieves efficient processing of local data and greatly optimizes resource utilization efficiency by deploying edge nodes near the source of data generation and reasonably allocating resources for them. By using Jaeger, Zipkin and other agents or embedding tracking codes, the call chain data between microservice instances is comprehensively collected, including key information such as logs, requests, and response time, providing a solid foundation for subsequent analysis. The deployment of lightweight machine learning models (decision trees) on edge nodes to analyze call chain data effectively reduces the burden of data transmission, and preliminarily constructs local call chains. Filtering out high-priority data based on decision trees not only reduces the overhead of data transmission, but also can quickly identify possible performance bottlenecks or abnormal situations. The central node forms a complete global call chain view by merging the data of edge nodes, and can be updated in real time to ensure that the global view always reflects the latest system status. In addition, the persistently stored Span data and service topology map provide strong support for subsequent operation and maintenance work, further improve the response speed and stability of the system, and bring significant benefits to the operation and maintenance of the power information system.

[0040] (2) The present invention can accurately distinguish high-priority data (such as abnormal requests and high-latency requests) from other data through a decision tree model, effectively focus on key information, and avoid irrelevant data interfering with edge computing, thereby improving the efficiency and pertinence of edge computing and reducing unnecessary waste of resources. Secondly, the model is trained based on historical call chain data so that the model has an accurate grasp of data characteristics and more reliable predictions. In a complex microservice cluster environment, this screening method can quickly locate links that may have problems, help to handle abnormal situations in a timely manner, ensure the stable operation of microservices, and improve the performance and reliability of the entire microservice cluster. At the same time, building a local call chain and uploading it to the central node also provides strong support for the accurate construction of the global call chain.

[0041] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their description are used to explain the present invention but do not constitute a limitation of the present invention.

[0043] Figure 1 A schematic diagram of a method for constructing a global call chain of a microservice cluster based on edge computing provided by an embodiment of the present invention;

[0044] Figure 2 The main flow chart of a method for building a global call chain of a microservice cluster based on edge computing provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] Embodiment 1

[0047] like Figure 1 As shown, this embodiment discloses a method for building a global call chain of a microservice cluster based on edge computing, including the following steps:

[0048] S1: Obtain the microservice cluster topology map and locate edge nodes based on the data source;

[0049] S2: Collect call chain data in the microservice cluster topology diagram and transmit it to the edge node;

[0050] S3: The edge node uses the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; the high-priority data includes abnormal requests and high-latency requests;

[0051] S4: The central node aggregates local call chains to form a global call chain.

[0052] Next, combine Figure 2 , a method for building a global call chain of a microservice cluster based on edge computing disclosed in this embodiment is described in detail.

[0053] Specifically, in S1, the microservice cluster topology diagram includes information such as the distribution location of microservice units in the cluster, the connection relationship between each microservice unit, and the possible hierarchical structure of the entire cluster (such as the setting of central nodes, edge nodes, etc.).

[0054] Edge nodes are deployed inside data sources, such as IoT devices, user terminals, or local data centers. At the same time, resources are configured for edge nodes according to the load conditions at different locations, such as allocation of processors, storage, and network bandwidth, to ensure that they can process local data efficiently.

[0055] S2 specifically includes:

[0056] S201: Based on the call chain data collected by the central node and the edge node in the microservice cluster topology diagram, the microservice call relationship is obtained and a preliminary directed graph is constructed;

[0057] S202: According to the preliminary directed graph, call chain data between microservice units is collected using a proxy tool or a tracing code embedded in the microservice, and transmitted to an edge node.

[0058] In S201, a microservice call relationship is constructed based on the call chain data collected by the central node and the edge nodes to generate a preliminary directed graph, in which a node represents a microservice unit (in fact, a microservice can be composed of multiple container groups deployed on multiple nodes).

[0059] In S202, for the collection of call chain data, Jaeger, Zipkin and other agents can be used or tracking code can be embedded in the microservice code to collect the call chain data between the corresponding microservice instances during the operation process, including logs, requests, response time and error information.

[0060] Specifically, you can choose appropriate proxy tools, such as OpenTelemetry, Jaeger, Zipkin, etc., according to the system architecture and requirements. These proxy tools can usually capture call chain data automatically or semi-automatically. Use container technology (such as Docker) to package and deploy edge computing applications to edge nodes, and use automation tools (such as Kubernetes) to manage the deployment, expansion, and upgrade of edge nodes. Install the appropriate operating system and middleware to ensure that edge nodes can run data collection and processing tools.

[0061] As a specific implementation method, the tracing library (Tracer) embedded in the microservice by Jaeger is used to collect the call chain data of each service. Each service call is marked as a "Span". In this system, each service call is regarded as an independent unit and is marked as a "Span". This "Span" is like a record encapsulation of a service call activity. It contains a lot of key information about this call, including service name, start time, duration, context information (such as Trace ID, Span ID, Parent Span ID), etc.

[0062] Among them, Trace ID is an ID used to uniquely identify a complete distributed transaction or business process. In a complex microservice architecture, multiple services may call and collaborate with each other to complete a complete business operation. For example, a user's order may involve the collaborative work of multiple microservices such as order service, inventory service, and payment service. Trace ID is like a globally unique "label" for the entire business process. Through this ID, all service calls related to the business process can be linked in series, making it easy to track and analyze the execution of the entire business process.

[0063] Each "Span" has a unique Span ID (Span ID), which is used to uniquely identify a specific service call within the same "Trace" (the complete business process identified by the TraceID). In other words, in a business process consisting of multiple service calls, each individual service call has its own unique Span ID, through which the corresponding service call activity can be accurately found among the many service calls in the business process.

[0064] Jaeger uses parent-child relationships to represent the call initiator and the callee. Each Span has a Span ID and can have one or more child spans. Specifically, the Parent Span ID is used when there is a call relationship between services. For example, when service A calls service B, the "Span" generated by the call operation of service A is the parent Span, and the "Span" generated by the call operation of service B is the child Span. The child Span will contain a Parent Span ID, which points to the Span ID of the parent Span. It clearly indicates which parent Span represents the service call that initiated the service call corresponding to the child Span, thereby establishing a hierarchical relationship between calls between services, making it easier to understand the structure and flow of the entire call chain.

[0065] The tracing library of each service reports the collected Span data to the Jaeger agent. This agent is usually deployed next to each node or service instance.

[0066] This embodiment uses proxy tools or tracking codes embedded in microservices to collect call chain data between microservice units, which can achieve comprehensive and accurate collection. No matter what type of call, it can be effectively recorded without missing any link, ensuring the integrity and accuracy of the data. At the same time, it can obtain data in real time, and instantly capture new call situations during the operation of microservices, so as to dynamically grasp the operation status. When a problem occurs, the call process can be quickly traced back based on the collected call chain data, and the fault point can be accurately located, providing strong support for optimizing microservice performance and troubleshooting.

[0067] S3 specifically includes:

[0068] S301: deploying the trained decision tree model on the edge node;

[0069] S302: The edge node uses the trained decision tree model to analyze and predict the call chain data, performs edge computing on the data predicted to be high priority, and constructs it into a local call chain, which is uploaded to the central node.

[0070] Among them, the training process of the decision tree model includes:

[0071] S3021: Acquire historical call chain data of the edge node, where the historical call chain data includes high priority data and other request data;

[0072] S3022: marking the historical call chain data to distinguish high-priority data from other request data;

[0073] S3023: At each decision tree node, select a preferred feature from multiple features included in the call chain data to partition the data, wherein the preferred feature is a feature that can achieve the best effect of data segmentation according to a preset evaluation standard; recursively select the best feature according to the evaluation standard to partition the data, and construct a decision tree; wherein each node in the decision tree represents a test of a certain feature, each branch represents a value of a feature, and the final leaf node contains a prediction result;

[0074] S3024: Train the constructed decision tree using the annotated historical call chain data to obtain a trained decision tree model, so that the model can predict whether the data is high priority data based on the characteristics of the call chain data.

[0075] Specifically, in the process of building a microservice call chain in edge computing, the trained decision tree model plays an important role.

[0076] First, the decision tree model is trained. The historical call chain data containing high-priority data (such as abnormal requests, high-latency requests) and other request data are obtained from the edge nodes, and are annotated to distinguish different types of data. At each node of the decision tree, the preferred features that can achieve the best data segmentation effect are selected from multiple features of the call chain data according to the preset evaluation criteria (such as information gain, etc.) to divide the data. The decision tree is constructed by recursively selecting the best features. Its nodes represent feature tests, branches are feature values, and leaf nodes contain prediction results. The model is trained using the annotated historical data so that it can predict whether the data is high priority based on the features.

[0077] Then, the trained decision tree model is deployed on the edge node. The edge node uses the model to analyze and predict the call chain data and obtain high-priority data. After that, the edge node performs edge computing on these high-priority data and uses these processed high-priority data to build a local call chain. It should be understood that building a local call chain based on edge computing is something that can be achieved by those skilled in the art.

[0078] In this process, the edge node can capture the Span data generated by the local microservice through the proxy, parse and extract key information (such as Trace ID and Span ID), and build a local call chain in the form of a directed graph based on the parent-child relationship. At the same time, the edge node can perform data preprocessing, including filtering irrelevant data, compressing data size, aggregating statistical information, etc., to reduce the burden of data transmission, and can also independently complete the preliminary analysis of the local call chain, such as detecting anomalies or performance bottlenecks. Define priority levels according to business needs, mark specific events (such as anomalies, timeouts, system failures) as high priority, and the edge node will prioritize uploading high-priority or abnormal call chain data to the central node. Low-priority data can choose not to upload or delay uploading. The central node is responsible for the construction and maintenance of the global view to enhance system stability.

[0079] This embodiment uses a trained lightweight machine learning model, the decision tree model, to filter high-priority data, which can effectively filter out a large amount of invalid and redundant data on the edge node. The decision tree model is trained based on historical call chain data, has a clear understanding of the characteristics of high-priority data, and can accurately distinguish key data such as abnormal requests and high-latency requests, avoiding irrelevant data from being mixed into subsequent processes. Secondly, it reduces the burden of invalid and redundant data transmission. Since only high-priority data is screened out for edge computing and transmission, the data transmission condition is greatly improved, and transmission obstacles caused by a large amount of invalid data transmission are avoided, thereby improving data transmission efficiency. In addition, since the edge node effectively filters the data, the amount of data transmitted to the central node is greatly reduced, and the central node does not need to process too much useless data, thereby effectively reducing the load on the central node and improving the performance and stability of the entire microservice call chain system based on edge computing.

[0080] S4 specifically includes:

[0081] S401: The central node receives local call chains from different edge nodes, where the local call chains include Parent Span ID information;

[0082] S402: Identify the calling relationship between services or operations according to the Parent Span ID information, and generate a corresponding directed edge, where the direction of the directed edge is from the caller to the callee;

[0083] S403: For each directed edge, construct a corresponding call chain path according to the identified call relationship;

[0084] S404: Integrate multiple directed edge data to form a global call chain;

[0085] S405: When a new local call chain arrives, the central node dynamically updates the global call chain.

[0086] After receiving the call chain data from multiple edge nodes, the central node is responsible for merging these local data to form a complete global call chain view. When the data uploaded by the edge node indicates that the topology structure has changed, the global view is updated in time. Persistent storage of Span data and service topology diagrams.

[0087] The central node integrates Span data from different service instances. Based on the Parent Span ID information in the Span data, the central node can identify the calling relationship between services or operations and generate corresponding directed edges, with the direction of the edge pointing from the caller to the callee. The call chain corresponding to each Trace usually forms a path, and the data of multiple Traces can be integrated into a complete topology structure, showing all possible call paths between services. When new Span data arrives, the central node can dynamically update the topology to ensure that the global call chain view always reflects the latest system status. If the service structure in the system changes (such as adding or deleting services), the topology map will also be updated accordingly.

[0088] After the central node is integrated, it will obtain Span data and the corresponding service topology diagram. These key information need to be persistently stored in the database to provide data support for the subsequent adjustment of dynamic monitoring strategies, and to facilitate fault tracing when an abnormality occurs in the information system.

[0089] This specific embodiment discloses a method for dynamically constructing service topology in a microservice cluster of an electric power information system. By using the Jaeger agent in conjunction with the edge node, efficient collection and preliminary processing of distributed tracking data can be achieved. The edge node can not only reduce the pressure of data transmission, but also improve the response speed of the system through local analysis, thereby reducing the burden on the central node and ensuring that the system's tracking data can be processed in a timely and effective manner.

[0090] This embodiment creatively uses a decision tree model to screen high-priority data. The decision tree is trained based on historical call chain data and can accurately distinguish high-priority data (such as abnormal requests and high-latency requests) from other data. This can avoid interference from invalid data, focus on key information, improve edge computing efficiency, and reduce resource waste. At the same time, high-priority data is calculated and local call chains are constructed at the edge nodes, reducing the load on the central node. The central node generates a global call chain by processing local call chains, ensuring the efficiency and stability of the system. This method can quickly locate problem links and improve the performance of the entire microservice cluster.

[0091] Embodiment 2

[0092] This embodiment provides a microservice cluster global call chain construction system based on edge computing, including:

[0093] Initialization module, used to obtain the microservice cluster topology map and locate edge nodes based on the data source;

[0094] The call chain data acquisition module is used to collect the call chain data in the microservice cluster topology diagram and transmit it to the edge node;

[0095] The local call chain construction module is used for edge nodes to use the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; the high-priority data includes abnormal requests and high-latency requests;

[0096] The global call chain building module is used by the central node to aggregate local call chains to form a global call chain.

[0097] Embodiment 3

[0098] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the method for building a global call chain of a microservice cluster based on edge computing as described in the first embodiment above are implemented.

[0099] Embodiment 4

[0100] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the method for constructing a global call chain of a microservice cluster based on edge computing as described in the first embodiment above are implemented.

[0101] The steps or modules involved in the above embodiments 2 to 4 correspond to those in embodiment 1. For the specific implementation, please refer to the relevant description of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0102] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for constructing a global call chain of a microservice cluster based on edge computing, characterized in that: include: Obtain the microservice cluster topology map and locate edge nodes based on the data source; Collect call chain data in the microservice cluster topology diagram and transmit it to the edge node; The edge node uses the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; the high-priority data includes abnormal requests and high-latency requests; The central node aggregates local call chains to form a global call chain.

2. A method for constructing a global call chain of a microservice cluster based on edge computing as claimed in claim 1, characterized in that: The positioning of edge nodes based on data sources is specifically: deploying edge nodes inside data sources, and the data sources include Internet of Things devices, user terminals, and local data centers.

3. A method for constructing a global call chain of a microservice cluster based on edge computing as claimed in claim 1, characterized in that: The collection of call chain data in the microservice cluster topology diagram and transmission to the edge node is specifically as follows: Based on the call chain data collected by the central node and edge node in the microservice cluster topology diagram, the microservice call relationship is obtained and a preliminary directed graph is constructed; According to the preliminary directed graph, the call chain data between microservice units is collected using proxy tools or tracing codes embedded in microservices and transmitted to the edge nodes.

4. A method for constructing a global call chain of a microservice cluster based on edge computing as claimed in claim 1, characterized in that: The call chain data includes logs, requests, response times and error information.

5. A method for constructing a global call chain of a microservice cluster based on edge computing as claimed in claim 1, characterized in that: The edge node uses the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; specifically, it includes: Deploy the trained decision tree model on edge nodes; The edge node uses the trained decision tree model to analyze and predict the call chain data, performs edge computing on the data predicted to be high priority, and builds it into a local call chain and uploads it to the central node.

6. A method for constructing a global call chain of a microservice cluster based on edge computing as claimed in claim 5, characterized in that: The training process of the decision tree model includes: Obtain historical call chain data of the edge node, wherein the historical call chain data includes high priority data and other request data; Annotating the historical call chain data to distinguish high-priority data from other request data; At each decision tree node, a preferred feature is selected from multiple features included in the call chain data to divide the data, wherein the preferred feature is a feature that can achieve the best effect of data segmentation according to a preset evaluation standard; the data is divided by recursively selecting the best feature according to the evaluation standard to construct a decision tree; wherein each node in the decision tree represents a test of a certain feature, each branch represents a value of a feature, and the final leaf node contains the prediction result; The constructed decision tree is trained using the annotated historical call chain data to obtain a trained decision tree model, so that the model can predict whether the data is high-priority data based on the characteristics of the call chain data.

7. A method for constructing a global call chain of a microservice cluster based on edge computing as claimed in claim 1, characterized in that: The central node aggregates local call chains to form a global call chain, which specifically includes: The central node receives local call chains from different edge nodes, wherein the local call chains include Parent Span ID information; According to the Parent Span ID information, the calling relationship between services or operations is identified, and a corresponding directed edge is generated, where the direction of the directed edge is from the caller to the callee; For each directed edge, a corresponding call chain path is constructed based on the identified call relationship; Integrate multiple directed edge data to form a global call chain; When a new local call chain arrives, the central node dynamically updates the global call chain.

8. A microservice cluster global call chain construction system based on edge computing, characterized in that: include: Initialization module, used to obtain the microservice cluster topology map and locate edge nodes based on the data source; The call chain data acquisition module is used to collect the call chain data in the microservice cluster topology diagram and transmit it to the edge node; The local call chain construction module is used for edge nodes to use the trained decision tree model to filter out high-priority data in the call chain data for edge computing, build a local call chain, and upload it to the central node; the high-priority data includes abnormal requests and high-latency requests; The global call chain building module is used by the central node to aggregate local call chains to form a global call chain.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the steps in the method for building a global call chain of a microservice cluster based on edge computing are implemented as described in any one of claims 1-7.

10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for building a global call chain of a microservice cluster based on edge computing are implemented as described in any one of claims 1-7.

Images