Maintenance methods and devices for configuration management database

Abstract

This invention relates to the field of database management, and more particularly to a method and apparatus for maintaining a configuration management database. The disclosed apparatus includes: an NLP model, a time series analysis model, a graph neural network, a database, and an interaction module. The NLP model extracts CI information from the data and generates standardized configuration items. The time series analysis model predicts resource usage trends or failure risks. The graph neural network establishes a CI relationship model to analyze dependencies and their impact range. The database stores user data, and the interaction module supports human-computer interaction between the user and the device. This invention extracts CI information from text using an NLP model and combines semantic understanding with a graph neural network to capture dependencies within the information, overcoming the limitation of traditional tools that can only handle structured data.

Description

Technical Field

[0001] This invention relates to the field of database management, and more particularly to a method and apparatus for maintaining a configuration management database. Background Technology

[0002] Configuration Management Database (CMDB) is a core system used to store and manage all configuration items and their relationships in an IT system. It provides a centralized database to describe the structure, state, and dependencies of IT infrastructure, typically including configuration items, their attributes, and their dependencies.

[0003] The maintenance of existing configuration management databases mainly relies on manual collection, static scripts, or automated discovery tools based on certain rules. However, with changes in enterprise architecture, manual intervention is prone to data errors due to operational oversights or inconsistent standards. Furthermore, in large databases, describing CI relationships through predefined relation types makes it difficult to capture multi-layered dependencies in complex environments. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides the following technical solution:

[0005] The configuration management database maintenance device includes: NLP model, time series analysis model, graph neural network, database, and interaction module.

[0006] Specifically, the NLP model is used to extract CI information from the data and generate standardized configuration items; the time series analysis model is used to predict resource usage trends or failure risks; the graph neural network is used to establish CI relationship models, analyze dependencies and impact scope; the database is used to store user data; and the interaction module is used to support human-computer interaction between users and the device.

[0007] As an improvement to the above technical solution, the NLP model extracts CI information from the data and generates standardized configuration items, including the following steps:

[0008] The spaCy library is used to perform text splitting on the categorized and labeled data.

[0009] The BERT model is used to extract entities from the split questions with fine granularity, locate CI information fields in the text, and generate standardized configuration items based on the CI information.

[0010] As an improvement to the above technical solution, the database integrates an automatic asset discovery module. The automatic asset discovery module discovers asset information by acquiring data packet traffic and performs dynamic updates by establishing CI information for the assets.

[0011] A method for maintaining a configuration management database, applied to the configuration management database maintenance apparatus as described in any of the foregoing technical solutions, includes the following steps:

[0012] S10: Data preprocessing.

[0013] S20: Extract CI information from the processed data using an NLP model, and generate standardized configuration items based on the CI information.

[0014] S30: Based on standardized configuration items, a CI relationship model is established through a graph neural network to analyze dependencies and the scope of influence.

[0015] S40: Analyze resource indicator data through time series analysis models and combine them with CI relationship models to predict resource usage trends or failure risks.

[0016] As an improvement to the above technical solution, step S10 includes the following steps:

[0017] S11: Pre-collection of data.

[0018] S12: Clean up redundant, missing, or abnormal data in the collected data.

[0019] S13: Classify and label the processed data to provide supervised data for model training.

[0020] As an improvement to the above technical solution, the pre-collection of data includes the following steps:

[0021] S111: Collect time-series data and resource metrics data through the Prometheus monitoring system.

[0022] S112: Collect system alarm log information through the Nagios model.

[0023] S113: Collect system log information using the ELK tool.

[0024] As an improvement to the above technical solution, step S20 includes the following steps:

[0025] S21: Input the processed data into the NLP model, call the spaCy library to perform text splitting on the classified and labeled data, and use the BERT model to perform fine-grained entity extraction on the split questions to locate CI information fields in the text.

[0026] S22: Through configuration item standardization processing, the corresponding CI information fields are organized and matched to generate standardized configuration items.

[0027] As an improvement to the above technical solution, step S30 includes the following steps:

[0028] S31: Treat each configuration item in the standardized configuration items as node data of the graph, and define the relationship type between CIs as edge data.

[0029] S32: Build a structure graph based on node data, edge data, and the implicit relationships between data.

[0030] S33: Build a GraphSAGE graph neural network and learn the importance of neighbors through an attention mechanism.

[0031] S34: Calculate node similarity through GNN embedding, discover potential dependencies, and analyze directly connected nodes and indirectly dependent nodes.

[0032] S35: Based on the list of CIs that identify the scope of influence of directly connected nodes and indirectly dependent nodes, as well as the topological paths in the structure graph, obtain the dependencies and scope of influence of the faulty nodes.

[0033] The beneficial effects of this invention are:

[0034] This system leverages NLP models to accurately extract CI attributes (such as version numbers and dependency descriptions) from unstructured texts like maintenance documents and logs. By using NLP models to extract CI information from the text, and combining semantic understanding with graph neural networks to capture dependencies, it overcomes the limitations of traditional tools that can only handle structured data. Automated processes replace manual data entry and other non-automated update methods, making system updates more convenient and reducing failure rates. Attached Figure Description

[0035] Figure 1 This is a schematic block diagram of a maintenance device for a configuration management database according to the present invention;

[0036] Figure 2 This is a flowchart illustrating a method for maintaining a configuration management database according to the present invention. Detailed Implementation

[0037] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0038] The maintenance of existing configuration management databases mainly relies on manual collection, static scripts, or automated discovery tools based on certain rules. However, with changes in enterprise architecture, manual intervention is prone to data errors due to operational oversights or inconsistent standards. Furthermore, in large databases, describing CI relationships through predefined relation types makes it difficult to capture multi-layered dependencies in complex environments.

[0039] To address the above problems, the following implementation method is provided:

[0040] Example 1

[0041] A maintenance device for configuration management database is provided, including: NLP model, time series analysis model, graph neural network, database, and interaction module.

[0042] Specifically, the NLP model is used to extract CI information from the data and generate standardized configuration items; the time series analysis model is used to predict resource usage trends or failure risks; the graph neural network is used to establish CI relationship models, analyze dependencies and impact scope; the database is used to store user data; and the interaction module is used to support human-computer interaction between users and the device.

[0043] The NLP model analyzes and processes the prepared data to extract CI information and generates standardized configuration items based on the CI information. A CI relationship model is established through a graph neural network. The CI relationship model includes: physical connection relationships between devices (such as network cable connection between server and switch), logical dependency relationships (such as application dependency on database service), and deployment relationships (such as virtual machine deployment on physical host). Based on these relationships, the dependency relationships of faulty nodes and the scope of impact can be analyzed.

[0044] When a fault occurs, the time series analysis model can analyze predicted resource metrics (such as disk utilization). If a device's predicted metric exceeds a threshold, a linked graph neural network quickly generates a list, enabling rapid alarm notifications. The interaction module is further implemented using an NLP model. The NLP model performs semantic understanding on the user's input text, analyzes the user's needs, and then generates and executes corresponding action commands based on those needs. The database stores the extracted CI information, typically including configuration items, their attributes, and dependencies.

[0045] In one embodiment, the NLP model extracts CI information from the data and generates standardized configuration items by the following steps:

[0046] The spaCy library is used to perform text splitting on the categorized and labeled data.

[0047] spaCy is a high-performance natural language processing library that can break down long texts into sentence-level fragments. It can be used to segment long text information and reduce the length requirements for semantic recognition.

[0048] The BERT model is used to extract entities from the split questions with fine granularity, locate CI information fields in the text, and generate standardized configuration items based on the CI information.

[0049] The BERT model, acting as a "semantic understanding layer," addresses contextual dependencies and fuzzy matching issues. It can work with the spaCy library to achieve accurate identification of CI information fields.

[0050] In one embodiment, the database integrates an automatic asset discovery module, which discovers asset information by acquiring data packet traffic and performs dynamic updates by establishing CI information for the assets.

[0051] Specifically, when new data appears in the database, the system discovers asset types and IP addresses by acquiring data packet traffic. This information is then organized to create CI information, which is then updated to the system using an update tool. The ServiceNowDiscovery module, one of the core modules of the ServiceNow platform, can be used to implement this update tool. It can automatically discover configuration items (CI) and their relationships in the enterprise IT environment and synchronize this data to the Configuration Management Database (CMDB) in real time.

[0052] Example 2

[0053] To cooperate with Embodiment 1, a method for maintaining a configuration management database is provided, applied to the configuration management database maintenance apparatus as described in any one of Embodiment 1, comprising the following steps:

[0054] S10: Data preprocessing.

[0055] Specifically, step S10 includes the following steps:

[0056] S11: Pre-collection of data.

[0057] The pre-collection of the data includes the following steps:

[0058] S111: Collect time-series data and resource metrics data through the Prometheus monitoring system.

[0059] The resource metrics data collected by the Prometheus monitoring system typically include CPU utilization, network latency, and other data.

[0060] S112: Collect system alarm log information through the Nagios model.

[0061] Alarm log information is temporary alarm information generated when a fault occurs in the system. This alarm information is usually stored and analyzed so that alarm feedback can be provided more quickly when the same fault problem is encountered again.

[0062] S113: Collect system log information using the ELK tool.

[0063] In the ELK toolkit, Elasticsearch serves as a dedicated log database. When archiving for a long time, it may be transferred to Hadoop HDFS. This tool typically collects log information from the system's daily operations, which can be used to help train neural networks.

[0064] S12: Clean up redundant, missing, or abnormal data in the collected data.

[0065] During the preceding data collection process, a large amount of useless, missing, and abnormal data will inevitably be collected. If all of this data is used for training and analysis, it will not only waste a lot of time but also affect the training results. Therefore, it is necessary to clean and process it. Tools such as Pandas (for small-scale data), Dask (for distributed processing), and PySpark (for large-scale data) in the Python ecosystem can usually be used for this purpose.

[0066] S13: Classify and label the processed data to provide supervised data for model training.

[0067] After data cleaning is completed, the data needs to be classified and labeled. This can usually be done using knowledge graphs, semi-automatic labeling, and other methods. These are all publicly available solutions in the prior art, so they will not be described in detail in this embodiment.

[0068] S20: Extract CI information from the processed data using an NLP model, and generate standardized configuration items based on the CI information.

[0069] Specifically, step S20 includes the following steps:

[0070] S21: Input the processed data into the NLP model, call the spaCy library to perform text splitting on the classified and labeled data, and use the BERT model to perform fine-grained entity extraction on the split questions to locate CI information fields in the text.

[0071] The spaCy library can quickly extract fixed entities and perform correlation matching on text. Therefore, it can quickly decompose long texts into short texts. The BERT model can identify complex entities related to context and deeply understand semantic structure, thus enabling it to quickly extract CI information fields from related short texts. In this step, the extracted information also includes the corresponding implicit relationships.

[0072] S22: Through configuration item standardization processing, the corresponding CI information fields are organized and matched to generate standardized configuration items.

[0073] Standardization of configuration items typically involves unifying naming formats, pattern validation, and error correction. Unifying naming formats aims to pre-set the rule engine, ensure data regularity, and reduce the complexity of subsequent searches. Pattern validation is usually achieved through dictionary constraints, logical validation, and other methods. Matching and error correction aims to correct spelling errors, such as changing "WindwosServer" to "WindowsServer".

[0074] Finally, the CI information fields processed through configuration item standardization are organized and stored in the database as standardized configuration items.

[0075] S30: Based on standardized configuration items, a CI relationship model is established through a graph neural network to analyze dependencies and the scope of influence.

[0076] Specifically, step S30 includes the following steps:

[0077] S31: Treat each configuration item in the standardized configuration items as node data of the graph, and define the relationship type between CIs as edge data.

[0078] S32: Build a structure graph based on node data, edge data, and the implicit relationships between data.

[0079] The structure diagram constructed through steps S31 and S32 contains configuration items and the relationships between them. These data correspond to CI information, thus converting the CI information into a topological structure diagram linked by relationships.

[0080] S33: Build a GraphSAGE graph neural network and learn the importance of neighbors through an attention mechanism.

[0081] By using GraphSAGE graph neural networks to analyze and learn structural graphs, we can understand the relationships between nodes and establish dependency recognition.

[0082] S34: Calculate node similarity through GNN embedding, discover potential dependencies, and analyze directly connected nodes and indirectly dependent nodes.

[0083] By embedding GNNs to compute node similarity, we can obtain the implicit dependencies between some nodes. This allows us to obtain not only directly connected nodes but also indirectly dependent nodes.

[0084] S35: Based on the list of CIs that identify the scope of influence of directly connected nodes and indirectly dependent nodes, as well as the topological paths in the structure graph, obtain the dependencies and scope of influence of the faulty nodes.

[0085] When a fault occurs, while locating the faulty node, the directly connected nodes and indirectly dependent nodes can be analyzed immediately. It is generally believed that the synchronization influence rate of directly connected nodes is higher, while the synchronization influence rate of indirectly dependent nodes needs to be calculated and predicted based on GNN. After completing all the influence calculations, the CI information of the nodes and edges included in the affected range is output in the form of a list, and the topological path in the structure graph is organized and displayed.

[0086] S40: Analyze resource indicator data through time series analysis models and combine them with CI relationship models to predict resource usage trends or failure risks.

[0087] Time series models (such as LSTM or Prophet) analyze historical and real-time resource data (such as CPU utilization and disk I / O) to capture periodic patterns and abnormal fluctuations. Simultaneously, a CI (Correlation Interaction) model based on graph neural networks (GNNs) models the correlation between resource metrics and topological dependencies (such as the deployment relationship between servers and applications). This allows for the simultaneous assessment of the cascading impact on upstream and downstream systems (such as database service interruptions triggered by insufficient storage space) when predicting resource trends for individual devices. This dual-model collaborative mechanism can accurately predict resource bottlenecks and dynamically generate fault impact paths, providing a basis for capacity planning and fault self-healing decisions.

[0088] The above embodiments are merely illustrative of the technical solutions of the present invention and are not intended to limit it. Anyone skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A maintenance device for configuring and managing a database, characterized in that, include: An NLP model is used to analyze and process pre-prepared data, extract CI information, and generate standardized configuration items based on the CI information. The specific steps include: S21: Input the processed data into the NLP model, call the spaCy library to perform text splitting on the classified and labeled data, and use the BERT model to perform fine-grained entity extraction on the split questions to locate CI information fields in the text; S22: Through configuration item standardization processing, the corresponding CI information fields are organized and matched to generate standardized configuration items; Time series analysis models are used to predict resource usage trends or failure risks based on historical time series data. Graph neural networks are used to build CI relationship models based on standardized configuration items of CI information, and to analyze the dependencies and influence scope between CI information. Specifically, the following steps are included: S31: Treat each configuration item in the standardized configuration items as node data of the graph, and define the relationship type between CIs as edge data; S32: Build a structural graph based on node data, edge data, and the implicit relationships between data; S33: Build a GraphSAGE graph neural network and learn the importance of neighbors through an attention mechanism; S34: Calculate node similarity through GNN embedding, discover potential dependencies, and analyze directly connected nodes and indirectly dependent nodes; S35: Identify the CI list of impact ranges based on directly connected nodes and indirectly dependent nodes, as well as the topological paths in the structure diagram, to obtain the dependencies and impact ranges of the faulty nodes. A database is used to store user data; The interaction module is used to call the NLP model to perform semantic understanding of the text information input by the user, analyze the user's needs, and generate and execute corresponding action instructions based on the user's needs. The database integrates an automatic asset discovery module, which discovers asset information by acquiring data packet traffic and performs dynamic updates by establishing asset CI information.

2. The maintenance device for the configuration management database according to claim 1, characterized in that: The NLP model extracts CI information from the data and generates standardized configuration items through the following steps: The spaCy library is used to perform text splitting on the categorized and labeled data. The BERT model is used to extract entities from the split questions with fine granularity, locate CI information fields in the text, and generate standardized configuration items based on the CI information.

3. A method for maintaining a configuration management database, applied to the maintenance apparatus for a configuration management database as described in claim 1 or 2, characterized in that, The steps include: S10: Data preprocessing; S20: Extract CI information from the processed data using an NLP model, and generate standardized configuration items based on the CI information; S30: Based on standardized configuration items, a CI relationship model is established through a graph neural network to analyze dependencies and the scope of influence; S40: Analyze resource indicator data through time series analysis models and combine them with CI relationship models to predict resource usage trends or failure risks.

4. The method for maintaining a configuration management database according to claim 3, characterized in that: Step S10 includes the following steps: S11: Pre-collection of data; S12: Clean up redundant, missing, or abnormal data in the collected data; S13: Classify and label the processed data to provide supervised data for model training.

5. The method for maintaining a configuration management database according to claim 4, characterized in that: The preliminary collection of the data includes the following steps: S111: Collect time-series data and resource metrics data through the Prometheus monitoring system; S112: Collect system alarm log information through the Nagios model; S113: Collect system log information using the ELK tool.

Images