Fault prediction methods, devices, electronic equipment, and storage media

Abstract

This application discloses a fault prediction method, device, electronic device, and storage medium, relating to the field of data processing technology. By combining physical topology data and virtual topology data to determine hybrid topology characteristics, it fully considers the correlation between server physical components and virtual resources. Fault prediction is achieved by using prediction models and graph convolution. This not only accurately predicts faults in advance but also lays the foundation for subsequent root cause localization, effectively shortening fault repair time, reducing false alarm rate, and improving operation and maintenance efficiency.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and more particularly to a fault prediction method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the rapid development of information technology, server scale continues to expand and architecture becomes increasingly complex, placing increasingly stringent demands on intelligent and precise operation and maintenance (O&M). Server O&M technology has become a key technology for ensuring system reliability. In server O&M, a relatively traditional approach is commonly used, relying on preset threshold alarm mechanisms. This involves real-time monitoring of single indicators such as server processor utilization, memory usage, and disk I / O, triggering alarms when these indicators exceed set thresholds. Simultaneously, fault diagnosis often relies on rule engines, where O&M experts configure alarm rules and fault judgment logic based on experience. After an alarm is generated, O&M personnel must manually screen and analyze relevant indicator data to determine if it is a real fault and attempt to locate the cause.

[0003] The aforementioned operation and maintenance process is difficult to cover complex nonlinear fault scenarios, resulting in low operation and maintenance efficiency. Summary of the Invention

[0004] This application provides a fault prediction method, apparatus, electronic device, and storage medium to at least solve the problem of low operation and maintenance efficiency in related technologies.

[0005] This application provides a fault prediction method, including:

[0006] Obtain multiple runtime data points from the target server;

[0007] Multiple operational data are preprocessed and modeled to obtain physical topology data, virtual topology data, and spatiotemporal features;

[0008] Based on physical and virtual topology data, determine the hybrid topology characteristics;

[0009] Based on spatiotemporal features and hybrid topological features, the fault prediction results are determined through a prediction model and graph convolution.

[0010] This application also provides a fault prediction device, comprising:

[0011] The acquisition module is used to acquire multiple runtime data from the target server;

[0012] The first determining module is used to preprocess and model multiple operational data to obtain physical topology data, virtual topology data, and spatiotemporal features;

[0013] The second determining module is used to determine the hybrid topology features based on physical topology data and virtual topology data;

[0014] The prediction module is used to determine the fault prediction result based on spatiotemporal features and hybrid topological features, through a prediction model and graph convolution.

[0015] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for implementing the steps of any of the above-described fault prediction methods when executing the computer program.

[0016] This application also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of any of the above-described fault prediction methods.

[0017] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of any of the above-described fault prediction methods.

[0018] This application utilizes multiple operational data sets to determine physical topology data, virtual topology data, and spatiotemporal characteristics. It then dynamically fuses the physical and virtual topology data, and based on the spatiotemporal and hybrid topology characteristics, determines the fault prediction result through a predictive model and graph convolution. This approach, combining physical and virtual topology data to determine hybrid topology characteristics, fully considers the correlation between server physical components and virtual resources. By leveraging predictive models and graph convolution to achieve fault prediction, it not only accurately predicts faults in advance but also lays the foundation for subsequent root cause localization, effectively shortening fault repair time, reducing false alarm rates, and improving operational efficiency. Attached Figure Description

[0019] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the structure of a fault prediction system provided in an embodiment of this application;

[0021] Figure 2 A flowchart illustrating a fault prediction method provided in an embodiment of this application;

[0022] Figure 3 A flowchart illustrating another fault prediction method provided in an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of a fault prediction device provided in an embodiment of this application;

[0024] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0026] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0027] With the rapid development of information technology, server scale continues to expand and architecture becomes increasingly complex, placing increasingly stringent demands on intelligent and precise operation and maintenance (O&M). Server O&M technology has become a key technology for ensuring system reliability. In server O&M, a relatively traditional approach is commonly used, relying on preset threshold alarm mechanisms. This involves real-time monitoring of single indicators such as server processor utilization, memory usage, and disk I / O, triggering alarms when these indicators exceed set thresholds. Simultaneously, fault diagnosis often relies on rule engines, where O&M experts configure alarm rules and fault judgment logic based on experience. After an alarm is generated, O&M personnel must manually screen and analyze relevant indicator data to determine if it is a real fault and attempt to locate the cause.

[0028] The aforementioned operation and maintenance process is difficult to cover complex nonlinear fault scenarios, resulting in low operation and maintenance efficiency.

[0029] To address the aforementioned technical issues, this application provides a fault prediction method. This method uses multiple operational data sets to determine physical topology data, virtual topology data, and spatiotemporal features. It then dynamically fuses the physical and virtual topology data. Based on the spatiotemporal features and the hybrid topology features, a prediction model and graph convolution are used to determine the fault prediction result. This approach, combining physical and virtual topology data to determine hybrid topology features, fully considers the relationship between server physical components and virtual resources. By leveraging a prediction model and graph convolution to achieve fault prediction, it not only accurately predicts faults in advance but also lays the foundation for subsequent root cause localization, effectively shortening fault repair time, reducing false alarm rates, and improving operational efficiency.

[0030] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Below, we will provide examples of specific application environment architectures or specific hardware architectures that the execution of the fault prediction method depends on.

[0032] Figure 1 This is a schematic diagram of a fault prediction system provided in an embodiment of this application. Please refer to... Figure 1 , Figure 1 This can include a fault prediction system. A fault prediction system can include a data acquisition layer, a feature fusion layer, and a fault prediction and diagnosis layer.

[0033] The data acquisition layer can be used to comprehensively and in real-time acquire multi-source data from server operation.

[0034] The data acquisition layer can include physical metrics, virtual topology, and log data.

[0035] For example, the data acquisition layer can collect data from physical sensors, such as CPU / memory / disk metrics, collect temperature / power consumption / error counts through the BMC, and collect topology information of deployment units or devices through Kubernetes.

[0036] The feature fusion layer can be used to construct topological relationships, build relationships between physical topological data and virtual topological data, and perform entropy weight fusion between physical topological data and virtual topological data.

[0037] The feature fusion layer can include computing physical topology data and virtual topology data, computing hybrid topology features, and computing spatiotemporal features.

[0038] The fault prediction and diagnosis layer can achieve high-precision, long-term fault prediction data through prediction models and graph convolution. It can also use the causal discovery-graph neural network engine to determine the root cause node based on real-time collected data after a fault is triggered, and match it with a preset knowledge base to generate a root cause report.

[0039] It can send fault prediction data and root cause reports to user terminals.

[0040] Figure 2 This is a flowchart illustrating a fault prediction method provided in an embodiment of this application. The execution entity in this embodiment can be a processor. The processor can be implemented in software or through a combination of software and hardware. Please refer to... Figure 2 The method includes:

[0041] S201. Obtain multiple runtime data from the target server.

[0042] Multiple operational data may include data from physical sensors, topology information of multiple deployment units or devices, log data, etc., without limitation.

[0043] The target server can be monitored and processed through the monitoring component to obtain multiple operational data of the target server.

[0044] S202. Preprocess and model multiple operational data to obtain physical topology data, virtual topology data, and spatiotemporal features.

[0045] Preprocessing includes at least one of missing value interpolation processing, outlier removal processing, and dynamic adjustment of preset coefficients processing.

[0046] For example, when the missing rate of a certain running data is less than 5%, linear interpolation is performed on the running data; when the missing rate is greater than 5%, an alarm is triggered.

[0047] When an abnormal value of a certain running data is detected, check the consistency of the data in its dependent nodes.

[0048] Dynamic adjustment of preset coefficients can be achieved by using the three sigma principle to remove outliers and automatically adjusting the sigma coefficients based on historical fluctuations.

[0049] Physical topology data can be used to describe the physical connections between hardware entities, reflecting the real physical relationships between various components in a server.

[0050] Virtual topology data can be used to describe the logical dependencies between software entities and reflect the interaction topology of virtualized components such as containers and services.

[0051] Spatiotemporal features can be used to describe the dimensional metric changes between historical log data and real-time log data.

[0052] Multiple operational data can be preprocessed separately to obtain multiple preprocessed data. These multiple preprocessed data can then be modeled to obtain physical topology data, virtual topology data, and spatiotemporal features.

[0053] Optionally, multiple runtime data can be preprocessed and modeled to obtain physical topology data, virtual topology data, and spatiotemporal features in the following ways: preprocessing multiple runtime data to obtain multiple preprocessed data; modeling multiple physical nodes and multiple physical connections to obtain physical topology data; modeling multiple virtual nodes and multiple virtual dependencies to obtain virtual topology data; and modeling log data to obtain spatiotemporal features.

[0054] The preprocessing includes at least one of missing value interpolation processing, outlier removal processing, and dynamic adjustment of preset coefficients processing. The preprocessed data includes multiple physical nodes, multiple physical connection relationships, multiple virtual nodes, multiple virtual dependency relationships, and log data.

[0055] Alternatively, the physical topology data G_{physical} can be obtained by modeling multiple physical nodes and multiple physical connections using the following formula:

[0056] G_{physical} = (V_p, E_p)

[0057] Here, V_p can represent a physical node, and E_p can represent a physical connection.

[0058] Alternatively, multiple virtual nodes and multiple virtual dependencies can be modeled using the following formula to obtain the virtual topology data G_{virtual}:

[0059] G_{virtual} = (V_v, E_v)

[0060] Here, V_v can represent a virtual node, and E_v can represent a virtual dependency.

[0061] S203. Determine the hybrid topology characteristics based on physical topology data and virtual topology data.

[0062] Hybrid topology features can be dynamic graph structures generated by fusing physical and virtual topologies through entropy weighting, in order to quantify the risk of cross-layer fault propagation.

[0063] Optionally, physical topology data and virtual topology data can be input into a preset two-layer graph model of physical and virtual topology to determine the hybrid topology characteristics.

[0064] Optionally, the hybrid topology features can be determined based on physical topology data and virtual topology data in the following manner: calculate the physical topology entropy value based on the physical topology data; calculate the virtual topology entropy value based on the virtual topology data; perform entropy weight fusion on the physical topology data and virtual topology data to determine the target weight; and determine the hybrid topology features based on the target weight, the physical topology entropy value, and the virtual topology entropy value.

[0065] It should be noted that the hybrid topology features can be determined according to any feasible implementation method, and the embodiments of this application do not limit this.

[0066] S204. Based on spatiotemporal characteristics and hybrid topological characteristics, the fault prediction results are determined through prediction models and graph convolution.

[0067] Fault prediction results can be determined in real time by deploying a quantum inference model interface embedded in the baseboard management controller.

[0068] Optionally, feature enhancement can be performed on spatiotemporal features and hybrid topological features through graph convolution. The enhanced spatiotemporal features and hybrid topological features are then input into the prediction model to determine the fault prediction result.

[0069] Optionally, based on spatiotemporal features and hybrid topological features, topological spatiotemporal features are determined by graph convolution; the topological spatiotemporal features are then input into the prediction model to determine the fault prediction result.

[0070] It should be noted that the fault prediction result can be determined according to any feasible implementation method, and the embodiments of this application do not limit this.

[0071] Optionally, the method further includes generating a root cause report, which can be generated by: acquiring real-time operational data in response to received fault information; generating a causal capability graph based on the real-time operational data; determining the node contribution of multiple nodes based on the causal capability graph using a preset analysis model; identifying the root cause node among the multiple nodes based on the contribution of the multiple nodes; and generating a root cause report based on the root cause node and a preset knowledge base.

[0072] The multiple nodes include multiple physical nodes and multiple virtual nodes, and the fault information can be information corresponding to various abnormalities or faults that may occur during the operation of the system.

[0073] Causal capability graphs can be used to represent the causal relationships between various components in a system.

[0074] Optionally, a causal capability graph can be generated based on real-time operational data in the following way: calculate a causal discovery matrix based on a preset causal discovery algorithm, and generate a causal capability graph based on the causal discovery matrix.

[0075] Node contribution can be used to represent the degree to which each node (including physical and virtual nodes) contributes to the current fault. The higher the contribution, the more likely the node is to be the source of the fault or a key influencing factor.

[0076] The node contribution can be calculated using the following formula. :

[0077]

[0078] in, It can represent the target node in both physical and virtual nodes. It can represent a node The set of neighboring nodes, This can represent the weight coefficient of neighbor node j. It can represent the loss function for nodes. The partial derivatives reflect the value at the node. The rate of change of the error (loss value) between the system's predicted result and the actual fault condition when the characteristics of the fault change slightly.

[0079] A root cause node can refer to the node where the root cause of a failure lies.

[0080] By comparing the contributions of each node, the node with the highest contribution can be selected as the root cause node.

[0081] The preset knowledge base can include system knowledge, failure modes, solutions, and other content.

[0082] Root cause reports can record the diagnostic process of a fault, detailed information about the root cause, possible solutions, etc.

[0083] This embodiment provides a fault prediction method that acquires multiple operational data points from a target server; preprocesses and models these data points to obtain physical topology data, virtual topology data, and spatiotemporal features; determines hybrid topology features based on the physical and virtual topology data; and determines the fault prediction result based on the spatiotemporal features and hybrid topology features using a prediction model and graph convolution. This method, by combining physical and virtual topology data to determine hybrid topology features, fully considers the relationship between physical components and virtual resources of the server. By using a prediction model and graph convolution to achieve fault prediction, it not only accurately predicts faults in advance but also lays the foundation for subsequent root cause localization, effectively shortening fault repair time, reducing false alarm rates, and improving operational efficiency.

[0084] Below, in conjunction with Figure 3 The specific process of determining hybrid topology features based on physical and virtual topology data is explained.

[0085] Figure 3 This is a flowchart illustrating another fault prediction method provided in an embodiment of this application. Based on the above embodiments, see [link to relevant documentation]. Figure 3 The method includes:

[0086] S301. Calculate the physical topology entropy value based on the physical topology data.

[0087] Physical topology entropy can be used to measure the uncertainty of hardware connection structures.

[0088] A higher physical topology entropy value indicates a more complex physical topology.

[0089] The physical topology entropy value can be calculated based on the physical topology data using a preset algorithm.

[0090] S302. Calculate the virtual topology entropy value based on the virtual topology data.

[0091] Virtual topology entropy can be used to measure the uncertainty of service dependencies.

[0092] A higher virtual topology entropy value indicates a more complex virtual call chain.

[0093] The virtual topology entropy value can be calculated based on the virtual topology data using a preset algorithm.

[0094] S303. Perform entropy weight fusion on physical topology data and virtual topology data to determine the target weight.

[0095] Target weights can be used to improve the fault detection rate in server hardware and container drift scenarios.

[0096] Optionally, the target weight can be determined by entropy weight fusion of physical topology data and virtual topology data in the following manner: determine the physical topology complexity based on the physical topology data; determine the virtual topology complexity based on the virtual topology data; determine the total topology complexity based on the physical topology complexity and the virtual topology complexity; and determine the ratio of the virtual topology complexity to the total topology complexity as the target weight.

[0097] Alternatively, the target weight α can be determined by entropy weight fusion of physical topology data and virtual topology data using the following formula:

[0098]

[0099] Among them, entropy(physical_graph) can represent physical topological complexity;

[0100] entropy(virtual_graph) can represent virtual topological complexity.

[0101] S304. Determine the hybrid topology features based on the target weight, physical topology entropy value, and virtual topology entropy value.

[0102] Optionally, a first product value between the target weight and the physical topology entropy value can be determined; a first difference value between the preset value and the target weight can be determined; a second product value between the first difference value and the virtual topology entropy value can be determined; and the sum of the first product value and the second product value can be determined as the hybrid topology feature.

[0103] Alternatively, the hybrid topology feature `hybrid_matrix` can be determined as follows:

[0104] hybrid_matrix = α×adj(physical_graph) + (1-α)×adj(virtual_graph)

[0105] Where α can represent the target weight;

[0106] adj(physical_graph) can represent the physical topological entropy value;

[0107] adj(virtual_graph) can represent the virtual topological entropy value.

[0108] The implementation details of each step in this application embodiment can be found in the description of the corresponding steps or operations in the above method embodiments; repeated content will not be repeated.

[0109] This embodiment provides a fault prediction method that calculates physical topology entropy based on physical topology data; calculates virtual topology entropy based on virtual topology data; performs entropy weight fusion on the physical and virtual topology data to determine a target weight; and determines hybrid topology characteristics based on the target weight, physical topology entropy, and virtual topology entropy. By comprehensively considering the influence of physical and virtual topologies, system performance can be optimized more effectively. Furthermore, by analyzing the changing trends of hybrid topology characteristics, the future state of the system can be predicted, allowing for advance planning and adjustments, thus improving operational efficiency.

[0110] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0111] Figure 4 This is a schematic diagram of a fault prediction device provided in an embodiment of this application. Please refer to... Figure 4 The fault prediction device 400 includes an acquisition module 401, a first determination module 402, a second determination module 403, and a prediction module 404.

[0112] The acquisition module 401 is used to acquire multiple operational data from the target server;

[0113] The first determining module 402 is used to preprocess and model multiple operational data to obtain physical topology data, virtual topology data and spatiotemporal features;

[0114] The second determining module 403 is used to determine the hybrid topology features based on physical topology data and virtual topology data;

[0115] The prediction module 404 is used to determine the fault prediction result based on the spatiotemporal features and mixed topological features, through the prediction model and graph convolution.

[0116] In one possible implementation, the second determining module 403 is specifically used for:

[0117] Calculate the physical topology entropy value based on the physical topology data;

[0118] Calculate the virtual topology entropy value based on the virtual topology data;

[0119] Entropy weight fusion is performed on physical topology data and virtual topology data to determine the target weight;

[0120] Hybrid topology features are determined based on target weights, physical topology entropy values, and virtual topology entropy values.

[0121] In one possible implementation, the second determining module 403 is specifically used for:

[0122] Determine the physical topology complexity based on the physical topology data;

[0123] Determine the virtual topology complexity based on the virtual topology data;

[0124] The total topology complexity is determined based on the physical topology complexity and the virtual topology complexity.

[0125] The ratio of virtual topological complexity to total topological complexity is determined as the target weight.

[0126] In one possible implementation, the second determining module 403 is specifically used for:

[0127] Determine the first product value between the target weight and the physical topology entropy value;

[0128] Determine the first difference between the preset value and the target weight;

[0129] Determine the second product value between the first difference and the virtual topological entropy value;

[0130] The sum of the first and second product values ​​is determined as the hybrid topological feature.

[0131] In one possible implementation, the prediction module 404 is specifically used for:

[0132] Based on spatiotemporal features and hybrid topological features, topological spatiotemporal features are determined through graph convolution;

[0133] The topological spatiotemporal features are input into the prediction model to determine the fault prediction results.

[0134] In one possible implementation, the first determining module 402 is specifically used for:

[0135] Multiple running data are preprocessed to obtain multiple preprocessed data. The preprocessing includes at least one of missing value interpolation processing, outlier removal processing, and dynamic adjustment of preset coefficient processing. The multiple preprocessed data includes multiple physical nodes, multiple physical connection relationships, multiple virtual nodes, multiple virtual dependency relationships, and log data.

[0136] Modeling multiple physical nodes and multiple physical connections yields physical topology data.

[0137] Multiple virtual nodes and multiple virtual dependencies are modeled to obtain virtual topology data;

[0138] The log data is modeled and processed to obtain spatiotemporal features.

[0139] In one possible implementation, the apparatus further includes a generation module 405, which is specifically used for:

[0140] In response to received fault information, acquire real-time operational data;

[0141] Generate a causal capability diagram based on real-time operational data;

[0142] Based on the causal capability diagram, the node contribution of multiple nodes is determined by a preset analysis model. These multiple nodes include multiple physical nodes and multiple virtual nodes.

[0143] The root cause node is determined from among multiple nodes based on their contributions.

[0144] Generate a root cause report based on root cause nodes and a pre-defined knowledge base.

[0145] For a description of the features in the embodiment corresponding to the fault prediction device, please refer to the relevant description of the embodiment corresponding to the fault prediction method, which will not be repeated here.

[0146] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. Figure 5 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the electronic device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus.

[0147] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to execute the above-described fault prediction method embodiment.

[0148] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0149] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0150] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0151] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0152] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above-described fault prediction method embodiments when running.

[0153] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0154] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described fault prediction method embodiments.

[0155] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in any of the above-described fault prediction method embodiments.

[0156] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0157] The above provides a detailed description of a fault prediction method provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A fault prediction method, characterized in that, include: Obtain multiple operational data from the target server, including multiple physical nodes, multiple physical connection relationships, multiple virtual nodes, multiple virtual dependency relationships, and log data; The multiple operational data are preprocessed and modeled to obtain physical topology data, virtual topology data, and spatiotemporal features. The physical topology data is determined by modeling the multiple physical nodes and the multiple physical connection relationships. The virtual topology data is determined by modeling the multiple virtual nodes and the multiple virtual dependency relationships. The spatiotemporal features are determined by modeling the log data. Calculate the physical topology entropy value based on the physical topology data; Calculate the virtual topology entropy value based on the virtual topology data; Entropy weight fusion is performed on the physical topology data and the virtual topology data to determine the target weight; The hybrid topology features are determined based on the target weight, the physical topology entropy value, and the virtual topology entropy value. Based on the spatiotemporal features and the hybrid topological features, the fault prediction result is determined through a prediction model and graph convolution.

2. The method according to claim 1, characterized in that, Entropy weight fusion is performed on the physical topology data and the virtual topology data to determine the target weight, including: Based on the physical topology data, determine the physical topology complexity; The virtual topology complexity is determined based on the virtual topology data. The total topology complexity is determined based on the physical topology complexity and the virtual topology complexity. The ratio of the virtual topological complexity to the total topological complexity is determined as the target weight.

3. The method according to claim 1, characterized in that, Based on the target weight, the physical topology entropy value, and the virtual topology entropy value, the hybrid topology features are determined, including: Determine the first product value between the target weight and the physical topology entropy value; Determine the first difference between the preset value and the target weight; Determine the second product value between the first difference and the virtual topological entropy value; The sum of the first product value and the second product value is determined as the hybrid topological feature.

4. The method according to any one of claims 1-3, characterized in that, Based on the spatiotemporal features and the hybrid topological features, the fault prediction results are determined through a prediction model and graph convolution, including: Based on the spatiotemporal features and the hybrid topological features, topological spatiotemporal features are determined by graph convolution; The topological spatiotemporal features are input into the prediction model to determine the fault prediction result.

5. The method according to any one of claims 1-3, characterized in that, Preprocessing of the multiple operational data includes: At least one of the following processes is performed on the multiple running data: missing value interpolation processing, outlier removal processing, and dynamic adjustment of preset coefficients, to obtain the preprocessed multiple running data.

6. The method according to any one of claims 1-3, characterized in that, The method further includes: In response to received fault information, acquire real-time operational data; Based on the real-time operational data, a causal capability graph is generated; Based on the causal capability diagram, the node contribution of multiple nodes is determined by a preset analysis model. The multiple nodes include multiple physical nodes and multiple virtual nodes. Based on the contribution of the multiple nodes, the root cause node is determined among the multiple nodes; A root cause report is generated based on the root cause nodes and the preset knowledge base.

7. A fault prediction device, characterized in that, include: The acquisition module is used to acquire multiple runtime data of the target server, including multiple physical nodes, multiple physical connection relationships, multiple virtual nodes, multiple virtual dependency relationships, and log data; The first determining module is used to preprocess and model the multiple running data to obtain physical topology data, virtual topology data, and spatiotemporal features. The physical topology data is determined by modeling the multiple physical nodes and the multiple physical connection relationships. The virtual topology data is determined by modeling the multiple virtual nodes and the multiple virtual dependency relationships. The spatiotemporal features are determined by modeling the log data. The second determining module is used to calculate the physical topology entropy value based on the physical topology data, calculate the virtual topology entropy value based on the virtual topology data, perform entropy weight fusion on the physical topology data and the virtual topology data to determine the target weight, and determine the hybrid topology features based on the target weight, the physical topology entropy value and the virtual topology entropy value. The prediction module is used to determine the fault prediction result based on the spatiotemporal features and the hybrid topological features, through a prediction model and graph convolution.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the fault prediction method as described in any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the fault prediction method as described in any one of claims 1 to 6.

Images