System upgrading method, device and equipment, readable storage medium and program product

Abstract

The invention discloses a system upgrading method and device, equipment, a readable storage medium and a program product, and relates to the technical field of computers.The method comprises the steps that multiple target components in a to-be-upgraded system are determined, and the to-be-upgraded system comprises a hardware layer, a software layer and a service layer; the target component is one of a component to which a hardware layer belongs, a component to which a software layer belongs and a component to which a business layer belongs; based on the target parameter information of each target component in the at least one target component and an association relationship between the at least two target components obtained by carrying out topological mapping on the to-be-upgraded system, constructing a first digital twinborn model comprising the at least one target component; and based on the first digital twinborn model, determining upgrading priorities of the at least two target components, and upgrading the to-be-upgraded system based on the upgrading priorities of the at least two target components. According to the method, the interaction and dependency relationship among a plurality of components can be captured and managed in the system upgrading process, and the data consistency and the system stability are guaranteed.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a system upgrade method, apparatus, device, readable storage medium, and program product. Background Art

[0002] The upgrade and optimization of large-scale information technology (IT) systems faces challenges such as complex inter-component dependencies, difficulty ensuring data consistency during the upgrade process, and inadequate real-time monitoring and feedback mechanisms. These issues often increase the risk of failure during IT system upgrades, especially in complex systems involving multiple distributed components and multi-layered topologies. This can lead to system performance degradation, data loss, and an inability to respond to business needs in a timely manner. Consequently, digital twin-based system optimization and management methods are being considered as key technical approaches for large-scale IT systems. This technology uses a data-driven approach, combining real-time monitoring, predictive analysis, and optimized decision-making to achieve intelligent management and continuous improvement of the system. However, digital twin-based system optimization and management methods primarily focus on status monitoring of single components or devices and lack comprehensive modeling of the complex dependencies across the entire system. This makes it impossible to fully capture and manage the interactions and dependencies between multiple components during the system upgrade process, thereby impacting data consistency and system stability. Summary of the Invention

[0003] The embodiments of the present application provide a system upgrade method, apparatus, device, readable storage medium, and program product that can comprehensively capture and manage the interactions and dependencies between multiple components during a system upgrade process, thereby ensuring data consistency and system stability during the upgrade process.

[0004] In a first aspect, an embodiment of the present application provides a system upgrade method, comprising:

[0005] Determining multiple target components in a system to be upgraded, the system to be upgraded comprising a hardware layer, a software layer, and a business layer, the target component being one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer;

[0006] Based on target parameter information of each target component in at least one of the target components and an association relationship between at least two of the target components obtained by performing topological mapping on the system to be upgraded, constructing a first digital twin model, wherein the first digital twin model includes at least two of the target components;

[0007] Determine the upgrade priorities of at least two of the target components based on the first digital twin model, and upgrade the system to be upgraded based on the upgrade priorities of at least two of the target components.

[0008] Optionally, determining the upgrade priorities of at least two target components based on the first digital twin model, and upgrading the system to be upgraded based on the upgrade priorities of the at least two target components, includes:

[0009] Obtain a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during the system upgrade;

[0010] Based on the characteristic information of the business layer, predict the target state information of the hardware layer through the twin reverse inference loss function, and determine the target loss value corresponding to the twin reverse inference loss function based on the target state information;

[0011] Obtaining a third score for the first digital twin model based on the first score, the second score, and the target loss value;

[0012] determining a first upgrade priority for each of the target components based on the third score;

[0013] determining a second upgrade priority for each target component based on the first upgrade priority and the third score of each target component;

[0014] The target components are upgraded in sequence according to the second upgrade priority of each target component.

[0015] Optionally, before determining the multiple target components in the system to be upgraded, the method further includes:

[0016] Identifying multiple target components in the system to be upgraded;

[0017] Determining, based on the total number of the plurality of nodes to be verified included in the plurality of target components, the importance weight of each of the nodes to be verified, the uncertainty parameter of the data corresponding to each of the nodes to be verified, and the falsity parameter of the data;

[0018] The security of the multiple target components is verified based on the security parameter value corresponding to each of the nodes to be verified at the first moment.

[0019] Optionally, the target parameter information includes at least one of component status information, configuration parameters, performance indicator data, and behavior information;

[0020] After determining multiple target components in the system to be upgraded, and before constructing the first digital twin model based on target parameter information of each target component in at least one of the target components and an association relationship between at least two target components obtained by topological mapping the system to be upgraded, the method further includes at least one of the following:

[0021] collecting component status information of the at least one target component through an integrated monitoring tool;

[0022] Parsing the configuration file of the system to be upgraded based on an automated management tool to extract the configuration parameters of at least one layer among the hardware layer, the software layer, and the business layer;

[0023] Collecting the performance indicator data of the system to be upgraded based on a performance monitoring tool;

[0024] The behavior information is extracted through the system log of the system to be upgraded.

[0025] Optionally, the association relationship between at least two of the target components includes a first association relationship, a second association relationship, and a third association relationship;

[0026] Among them, the first association relationship is used to represent the physical connection relationship between each of the target components belonging to hardware in the at least one target component, the second association relationship is used to represent the logical dependency relationship between each of the target components belonging to software in the at least one target component, and the third association relationship is used to represent the interaction relationship between the software layer and the business layer.

[0027] Optionally, obtaining the first score includes:

[0028] Determining the first score corresponding to the software layer based on the number of data sources of the software layer, the truth values ​​of log files and monitoring data corresponding to components of the software layer, and the weights corresponding to the components of the software layer;

[0029] The target parameter information includes the truth value of the log file corresponding to the component of the software layer and the truth value of the monitoring data.

[0030] Optionally, obtaining the second score includes:

[0031] Determining the second score corresponding to the business layer based on the total number of components belonging to the business layer, status information of each target component in all component combinations in the business layer, a truth value of business data of each target component in at least one component combination, and a data similarity parameter between two target components in at least one component combination;

[0032] The business layer includes at least one component combination, and the component combination includes any two different target components in the business layer.

[0033] Optionally, the target state information of the hardware layer is predicted based on the feature information of the business layer through a twin back reasoning loss function, including:

[0034] The feature information of the business layer is obtained from the first digital twin model.

[0035] The feature vectors corresponding to the plurality of target components in the software layer of the first digital twin model are extracted.

[0036] The first state information of the hardware layer is predicted based on the feature information of the business layer and the feature vectors corresponding to the plurality of target components in the software layer.

[0037] The target state information of the hardware layer is predicted by back reasoning through the twin back reasoning loss function based on the first state information and historical state information.

[0038] The historical state information includes historical state information of the hardware layer, historical feature information of the business layer, and historical feature vectors of the software layer.

[0039] Optionally, the target state information of the hardware layer is predicted based on the first state information and the historical state information by back reasoning through the twin back reasoning loss function, including:

[0040] The initial loss value of the twin back reasoning loss function is determined based on the first state information, the historical state information of the hardware layer, the historical feature information of the business layer, and the historical feature vectors of the software layer.

[0041] The target weight matrix corresponding to the at least one target component is determined based on the initial loss value, a preset adjustment factor, and a basic weight matrix of the at least one target component.

[0042] The second digital twin model is obtained by adjusting the first digital twin model based on the target weight matrix.

[0043] The target operation is performed according to the second digital twin model until the error value between the predicted second state information of the hardware layer and the historical state information of the hardware layer is less than a preset value, and the predicted state information of the hardware layer is determined as the target state information.

[0044] Among them, the target operation includes: obtaining the characteristic information of the business layer from the second digital twin model; extracting the characteristic vectors corresponding to the multiple target components in the software layer of the second digital twin model; predicting the third state information based on the characteristic information of the business layer and the characteristic vectors corresponding to the multiple target components in the software layer; based on the third state information and the historical state information, reverse deduction is performed through the twin reverse deduction loss function to predict the second state information.

[0045] Optionally, determining the first upgrade priority of each target component based on the third score includes:

[0046] Determining the first upgrade priority of each target component based on the third score, the importance parameter corresponding to each target component, the traffic sensitivity and the historical failure rate of each target component;

[0047] The importance parameter is used to characterize the importance of the target component in performing real-time information feedback, and the target parameter information includes the flow sensitivity and historical failure rate of each target component.

[0048] Optionally, the method further includes:

[0049] Determining a target rate for updating the first digital twin model based on the change rates of the hardware layer, the software layer, and the business layer in the first digital twin model, the time constant for responding to input signals, external input data, and the connection weights between any two layers among the hardware layer, the software layer, and the business layer;

[0050] Determining a target domain of the first digital twin model through a machine learning classification network based on the target rate at which the first digital twin model is updated and the target rates of a preset number of digital twin models before the first digital twin model is updated;

[0051] The target component in the first digital twin model and components in other digital twin models having the target domain are upgraded.

[0052] Optionally, the method further includes:

[0053] In a case where a plurality of the target components belonging to the target domain are in the process of being upgraded, determining verification values ​​of all target components belonging to the target domain based on verification weights and verification values ​​of neighboring components in neighboring domains of the target domain; wherein the neighboring components are components in neighboring domains of the target domain;

[0054] Based on the verification value, consistency effects of the plurality of target components during the upgrade process are determined.

[0055] In a second aspect, an embodiment of the present application further provides an encryption device, comprising:

[0056] a component determination module, configured to determine a plurality of target components in a system to be upgraded, wherein the system to be upgraded includes a hardware layer, a software layer, and a business layer, wherein the target component is one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer;

[0057] a model construction module, configured to construct a first digital twin model based on target parameter information of each target component in at least one of the target components and an association relationship between at least two of the target components obtained by topological mapping the system to be upgraded, wherein the first digital twin model includes at least two of the target components;

[0058] A component upgrade module is used to determine the upgrade priorities of at least two target components based on the first digital twin model, and to upgrade the system to be upgraded based on the upgrade priorities of at least two target components.

[0059] Optionally, the component upgrade module includes:

[0060] A first acquisition submodule is configured to acquire a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during system upgrades;

[0061] An information prediction submodule, configured to predict the target state information of the hardware layer through a twin reverse deduction loss function based on the characteristic information of the business layer, and determine a target loss value corresponding to the twin reverse deduction loss function based on the target state information;

[0062] a second acquisition submodule, configured to obtain a third score for the first digital twin model based on the first score, the second score, and the target loss value;

[0063] a first determining submodule, configured to determine a first upgrade priority of each target component based on the third score;

[0064] a second determining submodule, configured to determine a second upgrade priority of each target component based on the first upgrade priority and the third score of each target component;

[0065] The component upgrade submodule is used to upgrade each of the target components in sequence according to the second upgrade priority of each of the target components.

[0066] In a third aspect, an electronic device is provided, which includes a processor, a memory, and a program stored in the memory and executable on the processor, and when the program is executed by the processor, the steps of the system upgrade method according to any one of the first aspect are implemented.

[0067] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the system upgrade method according to any one of the first aspect are implemented.

[0068] In a fifth aspect, a computer program product is provided, which is stored in a storage medium, and when the computer program product is executed by at least one processor, the steps of the system upgrade method according to any one of the first aspect are implemented.

[0069] In the embodiments of the present application, a plurality of target components belonging to the hardware layer, the software layer and the business layer in the system to be upgraded are determined, and then a first digital twin model including at least two target components is constructed according to the target parameter information of each target component in at least one of the target components and the association relationship between the at least two target components obtained by topological mapping of the system to be upgraded. Thus, the interaction and dependency relationship between the plurality of target components in the system to be upgraded are captured into the first digital twin model, which is managed by the first digital twin model, so that the dependency relationship between the layers in the upgrading process is kept consistent, the synchronization between the digital twin model and the actual system is ensured, and thus the first digital twin model can also be flexibly adjusted and dynamically optimized when facing a complex multi-level topology. In addition, each target component is upgraded according to the upgrade priority of each target component in the first digital twin model similar to the system to be upgraded, so as to ensure the consistency and stability of data in the component upgrading process, reduce the upgrade conflict, and reduce the failure risk. BRIEF DESCRIPTION OF DRAWINGS

[0070] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the description of the embodiments of the present application will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0071] Figure 1 is one of the flowcharts of the system upgrade method in the embodiments of the present application;

[0072] Figure 2 is another flowchart of the system upgrade method in the embodiments of the present application;

[0073] Figure 3 yes Figure 2 A schematic diagram of a process for predicting and adjusting the status information of a target component;

[0074] Figure 4 This is a structural diagram of a system upgrade device in an embodiment of the present application;

[0075] Figure 5 It is a schematic diagram of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0077] The embodiments of the present application provide a system upgrade method, device, electronic device, storage medium and program product. The embodiments of the present application are described in detail below through specific embodiments and their application scenarios in conjunction with the accompanying drawings.

[0078] See Figure 1 , Figure 1 This is one of the flowcharts of a system upgrade method in an embodiment of the present application, which specifically includes the following steps:

[0079] Step 101: Determine multiple target components in a system to be upgraded, where the system to be upgraded includes a hardware layer, a software layer, and a business layer. The target component is one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer.

[0080] It should be noted that by comprehensively analyzing and identifying the system to be upgraded, multiple target components in the system to be upgraded can be determined. These target components may be components that need to be included in the subsequent digital twin model. Specifically, the target components may be components belonging to the hardware layer, components belonging to the software layer, and components belonging to the business layer. For example, the hardware layer components in the system to be upgraded include servers, storage devices, network devices, etc.; the software layer components in the system to be upgraded include operating systems, databases, middleware, applications, etc.; and the business layer components include microservices, application programming interfaces (APIs), etc.

[0081] Optionally, before determining the multiple target components in the system to be upgraded, the method further includes:

[0082] Identifying the multiple target components in the system to be upgraded;

[0083] Determining, based on the total number of the plurality of nodes to be verified included in the plurality of target components, the importance weight of each of the nodes to be verified, the uncertainty parameter of the data corresponding to each of the nodes to be verified, and the falsity parameter of the data;

[0084] The security of the multiple target components is verified based on the security parameter value corresponding to each of the nodes to be verified at the first moment.

[0085] In some embodiments, the multiple target components identified in the system to be upgraded can be decentralized to ensure the authenticity and security of the data of each distributed component. It can be understood that independent computing units or network units in a distributed system, such as hardware nodes such as servers, network devices, storage devices, or logical nodes such as microservices and containers in software. The target component can be an entity to be managed in the system to be upgraded. As the smallest unit of function or structure in the system, the target component can include multiple nodes. For example, a database component can be composed of multiple storage nodes, or a microservice component can include multiple instance nodes. In decentralized verification, each node needs to participate in the security calculation of data interaction to ensure the authenticity of its data. Therefore, in the embodiment of the present application, security verification can be performed on multiple nodes to be verified corresponding to multiple target components.

[0086] Specifically, the security parameter value of each node to be verified can be determined by the total number of multiple nodes to be verified contained in multiple target components, the importance weight of each node to be verified, the uncertainty parameter and falsehood parameter of the relevant data of each node to be verified, and the specific time measurement.

[0087] In a specific embodiment, decentralized verification can be introduced to verify the authenticity and security of distributed component data. The security of data interaction is calculated through a multi-node decentralized verification formula, which can be expressed by the following formula:

[0088]

[0089] Wherein, N represents the total number of multiple nodes to be verified, and the total number of nodes to be verified can be determined by the number of multiple target components; C n represents the importance weights of n nodes to be verified; T(t) represents the authenticity measure of the data related to the node to be verified at time t; I(i) represents the uncertainty parameter of the data related to the i-th node to be verified; F(f) represents the falsity parameter of the data related to the node to be verified.

[0090] Through the above formula, the data security of all nodes to be verified corresponding to multiple target components is comprehensively evaluated by weighted summation. In the embodiment of the present application, the product of T(t), I(i), and F(f) can reflect the credibility of the data at a specific time and scenario, and combined with the weight C of each node to be verified, the data security of all nodes to be verified is comprehensively evaluated by weighted summation. n , and finally the overall safety index B(t,i,f,C) is obtained.

[0091] During the upgrade process, component dependencies can be simulated through digital twin models, combined with decentralized verification to ensure the security of the upgrade operation. For example, in a microservices architecture, each service instance of the target component acts as an independent node, and decentralized verification can be used to prevent malicious interference in inter-service communication.

[0092] Step 102: Based on the target parameter information of each target component in at least one of the target components and the association relationship between at least two of the target components obtained by topological mapping of the system to be upgraded, a first digital twin model is constructed, where the first digital twin model includes at least two of the target components.

[0093] In the above steps, target parameter information of each target component in the system to be upgraded can be collected in advance or in real time. The parameter information can specifically include state information, configuration parameters, performance indicators, behavior patterns, logs, and other data of the target component. In the embodiment of the present application, the association relationship between at least two target components in multiple layers of the system to be upgraded can also be obtained through multi-level topological mapping, such as the actual physical connection relationship between hardware, the logical dependency relationship between software components, the interaction relationship between the software layer and the upper-level business services, etc.

[0094] Based on this, the embodiment of the present application can complete the construction of the first digital twin model based on the existing target component association relationship and the target parameter information of the corresponding target component. The first digital twin model can have a similar structure to the system to be upgraded and can accurately reflect the status, configuration and performance of the system to be upgraded. Among them, through multi-level topological mapping including the physical layer, logical layer and application layer, the constructed first digital twin model can accurately characterize the dependency relationship between hardware, software and business layers.

[0095] Optionally, the target parameter information includes at least one of component status information, configuration parameters, performance indicator data, and behavior information;

[0096] After determining multiple target components in the system to be upgraded, and before constructing the first digital twin model based on target parameter information of each target component in at least one of the target components and an association relationship between at least two target components obtained by topological mapping the system to be upgraded, the method further includes at least one of the following:

[0097] collecting component status information of the at least one target component through an integrated monitoring tool;

[0098] Parsing the configuration file of the system to be upgraded based on an automated management tool to extract the configuration parameters of at least one layer among the hardware layer, the software layer, and the business layer;

[0099] Collecting the performance indicator data of the system to be upgraded based on a performance monitoring tool;

[0100] The behavior information is extracted through the system log of the system to be upgraded.

[0101] In some embodiments, after determining multiple target components in the system to be upgraded, target parameter information of each target component may be collected. Specifically, this may be done in the following manner:

[0102] (1) Collect the status information Sta of each target component through integrated monitoring tools (such as Prometheus) i ;

[0103] (2) Extract key event data Et from the log of the system to be upgraded i , and can update the status and behavior information Bw in the first digital twin model in real time i , collect log data Lr i ;

[0104] (3) Based on automated management tools (such as Ansible), the configuration files of the system to be upgraded are parsed to extract data such as the number of CPU cores, memory capacity, operating system version, and database connection parameters. Furthermore, after extraction, unified field mapping and normalization can be performed to ensure the consistency and availability of configuration parameters from different sources. In addition, for dynamically changing parameters (such as network bandwidth, memory allocation strategy, etc.), the latest values ​​can be obtained through real-time monitoring tools and synchronized to the first digital twin model, thereby achieving real-time data updates;

[0105] (4) The performance indicators in the above parameter information can be collected in real time based on performance monitoring tools such as Prometheus for the system to be upgraded. The performance indicators can specifically include CPU utilization, memory usage, network throughput, response time, etc. In addition, the collected raw data can be denoised and normalized to eliminate abnormal fluctuations and extract stable features.

[0106] In this way, the embodiment of the present application utilizes automated tools to scan and collect data from the system to be upgraded, accurately extracting and analyzing multidimensional data from the hardware, software, and business layers of large-scale IT systems to ensure that the data parameter information in the first digital twin model is consistent with the actual system to be upgraded. Furthermore, based on the dynamic multi-level association relationship topology of the first digital twin model, the status of each component during the upgrade process is monitored and optimized in real time, according to the importance, dependencies, real-time feedback, and historical data of the components, thereby reducing the risk of conflicts during the system upgrade and improving system stability.

[0107] Optionally, the association relationship between at least two of the target components includes a first association relationship, a second association relationship, and a third association relationship;

[0108] Among them, the first association relationship is used to represent the physical connection relationship between each of the target components belonging to hardware in the at least one target component, the second association relationship is used to represent the logical dependency relationship between each of the target components belonging to software in the at least one target component, and the third association relationship is used to represent the interaction relationship between the software layer and the business layer.

[0109] In some specific embodiments, the above-mentioned association relationship may include three levels of association relationships, and can be constructed from three levels. For the first association relationship, it can be understood as the dependency relationship between hardware in the physical layer, which is used to describe the actual physical connection relationship between hardware, such as the physical connection between the server and the storage device, the physical topology between the network devices, etc. For the second association relationship, it can be understood as the dependency relationship between software in the logical layer, which is used to describe the interaction and calling relationship between software components, such as the data transmission path between the database and the application program, etc. For the third association relationship, it can be understood as the dependency relationship in the application layer, focusing on the call and interaction between the upper-level business services and the underlying software functions, such as the call link of the microservice interface. Therefore, the embodiment of the present application can cover the association between each target component in the system to be upgraded through multi-level dependency modeling.

[0110] It's worth noting that updating relationships can be achieved through automated scanning tools. By periodically or in real time scanning the component states and interaction paths of the system to be upgraded, new or changed dependencies can be automatically captured, ensuring that the topology in the digital twin model is consistent with the actual system. For example, when a new microservice is deployed in the system to be upgraded or the call path of an existing application is adjusted, ServiceNow can automatically detect the change and synchronize the updated relationship with the digital twin model.

[0111] Step 103: Determine the upgrade priorities of at least two target components based on the first digital twin model, and upgrade the system to be upgraded based on the upgrade priorities of at least two target components.

[0112] It is understandable that in large-scale IT system upgrades, it is difficult to upgrade all components at the same time in certain actual situations. Therefore, the upgrade priority of each target component can be determined based on the importance, correlation, and real-time data of the target components. In this way, the target components to be upgraded first and the target components to be processed later can be determined, thereby optimizing the upgrade path and reducing the occurrence of component priority-related conflicts during the upgrade process.

[0113] In some embodiments, a priority judgment operator can be used to sort target components based on their importance, relationships, real-time feedback, and historical data to determine the upgrade order. This allows priority processing of target components with high importance and complex relationships, thereby optimizing the upgrade path.

[0114] Optionally, determining the upgrade priorities of at least two target components based on the first digital twin model, and upgrading the system to be upgraded based on the upgrade priorities of the at least two target components, includes:

[0115] Obtain a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during the system upgrade;

[0116] Based on the characteristic information of the business layer, predict the target state information of the hardware layer through the twin reverse inference loss function, and determine the target loss value corresponding to the twin reverse inference loss function based on the target state information;

[0117] Obtaining a third score for the first digital twin model based on the first score, the second score, and the target loss value;

[0118] determining a first upgrade priority for each of the target components based on the third score;

[0119] determining a second upgrade priority for each target component based on the first upgrade priority and the third score of each target component;

[0120] The target components are upgraded in sequence according to the second upgrade priority of each target component.

[0121] See Figure 2 , Figure 2 FIG2 is a second flow chart of a system upgrade method in an embodiment, wherein the system upgrade method specifically includes the following steps:

[0122] Step 201: Determine at least two target components in a system to be upgraded, where the at least one target component includes at least one of the following: a component belonging to a hardware layer, a component belonging to a software layer, and a component belonging to a business layer;

[0123] Step 202: Based on target parameter information of each target component in the at least one target component and an association relationship between the at least two target components obtained by performing topological mapping on the system to be upgraded, construct a first digital twin model, where the first digital twin model includes the at least one target component.

[0124] Step 203: Obtain a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during the system upgrade;

[0125] Step 204: Obtain a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during the system upgrade;

[0126] Step 205: Based on the characteristic information of the business layer, predict the target state information of the hardware layer through the twin reverse inference loss function, and determine the target loss value corresponding to the twin reverse inference loss function based on the target state information;

[0127] Step 206: Obtain a third score for the first digital twin model based on the first score, the second score, and the target loss value; and determine a first upgrade priority for each target component based on the third score.

[0128] Step 207: Determine a second upgrade priority for each target component based on the first upgrade priority and the third score of each target component.

[0129] Step 208: Upgrade each of the target components in sequence according to the second upgrade priority of each target component.

[0130] In some implementations, during large-scale IT system upgrades, it's necessary to ensure real-time synchronization between physical components and digital twin models. To this end, software-layer data fusion scoring and business-layer consistency scoring can be used to assess data accuracy and consistency during the upgrade process. Furthermore, by inferring business- and software-layer features, potential hardware-layer issues can be identified in advance. This, combined with hardware prediction accuracy, business consistency, and software synchronization quality, provides a more comprehensive basis for component prioritization.

[0131] It should be noted that the first and second scores mentioned above are key indicators for evaluating the data quality of the software and business layers during system upgrades. The first score is used to evaluate the integration of the data contained in the software layer, that is, by integrating multi-source data (such as logs, monitoring, and configuration files), the accuracy, completeness, and real-time nature of the software layer data are determined. The second score is used to measure the consistency of the business data contained in the business layer during the system upgrade, that is, by comparing the data status of business nodes (such as response time and user visits) to determine the stability of the business data before and after the upgrade.

[0132] It is worth mentioning that the embodiment of this application uses a digital twin reverse iterative algorithm network to characterize the hardware layer in the digital twin model by reversely verifying the second score of the consistency of the business layer and the first score of the data fusion of the software layer. In other words, in large-scale IT systems, for data that is difficult to represent at the business and software layers, a reverse deduction method can be used to construct a digital twin model of the hardware layer, and the derived hardware layer status can be compared with the actual measured hardware status to identify potential performance issues or system optimization points.

[0133] Therefore, the embodiment of the present application can supplement the influence of the business layer and the software layer on the actual hardware layer by based on the Multi-Service Provisioning Platform with Protection and Management (MSPPM) network architecture. Specifically, considering that the business layer has only one output, but the software layer may have multiple nodes, and the hardware layer is a single output scenario. Feature information can be obtained from the business layer, wherein the business layer information may include a first score and feature information, and the feature information may specifically include business traffic, user request volume, response time, etc. The software layer information may include a second score, CPU usage, memory usage, number of threads, etc. Further, the hardware layer status information Yj can be predicted based on the feature information of the business layer and the feature vector of the software layer. In the embodiment of the present application, in the first digital twin model, considering that the data transmission and interaction between the business layer, software layer and hardware layer are more important, the tightly coupled twin relationship between the business layer and the software layer can be improved through MSPPM, and the feedback and information transmission between the business layer and the software layer can be further utilized to finally make a more accurate prediction of the status of the hardware layer.

[0134] Therefore, based on the above-mentioned Lossfan calculated by generating a better digital twin hardware layer state through the MSPPM network architecture final , V(σ) is the consistency score of the business layer and the data fusion score of the software layer. A comprehensive twin model information score is proposed to express the overall information comprehensive score of this twin component:

[0135]

[0136] Among them, S(σ) represents the overall information comprehensive score of all target components as twin components, that is, the third score; V(σ) represents the second score; F(σ) represents the first score; Lossfan final represents the target loss value; σ represents the state parameters or configuration parameters of the system to be upgraded during the upgrade process.

[0137] In this way, by obtaining the overall information comprehensive score S(σ) of the twin components, the priority judgment operator completes the order arrangement of all component upgrades. Specifically, the embodiment of the present application discloses a twin component update formula, which is used to schedule large-scale IT system upgrades, that is, each time a component is updated, it is recalculated. Each time the reordering calculation is performed, the overall information comprehensive score of the twin components is considered. That is, when upgrading a large-scale IT system, the stability item of the twin model is given priority to reduce uncertainty factors. The component upgrade order determination formula can be expressed as:

[0138] Ps(i) t+1 =Ps(i)t *(1+rand(0,σ 2 ))*(1+S(σ))

[0139] Among them, Ps(i) t+1 The priority at time step t+1, that is, the second priority; Ps(i) t represents the priority at time step t, i.e. the first priority; S(σ) represents the third score.

[0140] In this embodiment, after each component upgrade, S(σ) is recalculated and the priority is updated to ensure that components with high stability are upgraded first. By prioritizing high-scoring components, the risk of conflicts and system fluctuations during the upgrade process are reduced.

[0141] Optionally, obtaining the first score includes:

[0142] Determining the first score corresponding to the software layer based on the number of data sources of the software layer, the truthfulness values ​​of log files corresponding to components of the software layer, the truthfulness values ​​of monitoring data, and the weights corresponding to the components of the software layer;

[0143] The target parameter information includes the truth value of the log file corresponding to the component of the software layer and the truth value of the monitoring data.

[0144] In some specific embodiments, the first score corresponding to the software layer may be determined by the following formula:

[0145]

[0146] Where F(σ) represents the first score of the software layer; M represents the number of data sources of the software layer; τ j (P(σ) represents the truth value of the log file corresponding to the i-th target component in the software layer; τ j Q(σ) represents the truth value of the monitoring data in the software layer; w j Represents the weight corresponding to the target component in the software layer; j represents the number or sequence number of the data source, which is the index variable of the summation symbol, and each j corresponds to an independent data source.

[0147] Therefore, through the above first scoring, it is possible to make upgrade-related decisions based on more comprehensive and accurate information in the software layer, and by integrating data, it is possible to ensure that the status information of the software layer can be more directly judged based on the software layer information.

[0148] Optionally, obtaining the second score includes:

[0149] determining the second score corresponding to the service layer based on a total number of components in the service layer, state information of each target component in all component combinations in the service layer, a truth value of service data of each target component in at least one component combination, and a data similarity parameter between two target components in the at least one component combination.

[0150] The service layer includes at least one component combination, and the component combination includes any two different target components in the service layer.

[0151] In some embodiments, after data fusion in the software layer, the data consistency of the service layer can be verified by the following service consistency verification formula to ensure the accuracy of data in the system upgrade process. The formula can be specifically represented as:

[0152]

[0153] V(σ) represents the consistency score of the service layer, i.e., the second score, which is used to measure the consistency and integrity of service data in the system upgrade process.

[0154] N represents the total number of relevant business nodes in the system, i.e., the number of target components related to the business, which is used to calculate the consistency relationship between all business components.

[0155] P i (σ) and P j (σ) represent the state information of the i th and j th business components, respectively. The state information can include request response time, user access volume, task processing delay, etc.

[0156] τ i P(σ) and τ j (P(σ) represent the truth value of service data of the i th and j th target components, respectively. Similar to the log truth value, it is used to reflect the reliability and integrity of the service data in the collection process.

[0157] Sim(P i (σ),P j (σ) represents the data similarity between the i th and j th business components, which is used to measure their consistency. The similarity can be calculated by a distance function, distribution feature similarity, etc.

[0158] 1-|(τ i (P(σ)-τ j (P(σ))| represents the difference between the truth values of service data of the i th and j th business components. By taking the reciprocal, it can reflect the situation that the smaller the difference is, the higher the consistency is.

[0159] Therefore, the above embodiment can determine the second score of the business layer, and the second score can be used to detect business data inconsistencies in advance, such as log loss, monitoring delays, etc., to avoid upgrade failures.

[0160] Optionally, the predicting the target state information of the hardware layer by using a twin reverse inference loss function based on the feature information of the business layer includes:

[0161] Acquire characteristic information of the business layer from the first digital twin model;

[0162] Extracting feature vectors corresponding to a plurality of target components in the software layer of the first digital twin model;

[0163] Predicting first state information of the hardware layer based on the feature information of the business layer and feature vectors corresponding to the plurality of target components in the software layer;

[0164] Based on the first state information and historical state information, reverse inference is performed through the twin reverse inference loss function to predict the target state information of the hardware layer;

[0165] The historical status information includes the historical status information of the hardware layer, the historical feature information of the service layer, and the historical feature vector of the software layer.

[0166] In some specific embodiments, Figure 3 As shown, the following steps are included:

[0167] Step 301: Acquire feature information of the business layer from the first digital twin model or the second digital twin model;

[0168] Step 302: extracting feature vectors corresponding to multiple target components in the software layer of the first digital twin model or the second digital twin model;

[0169] Step 303: Predict and obtain first state information of the hardware layer based on the feature information of the business layer and the feature vectors corresponding to the plurality of target components in the software layer;

[0170] Step 304: Based on the third state information and the historical state information, reverse inference is performed using the twin reverse inference loss function to predict the second state information.

[0171] Step 305: Calculate and obtain an error value between the second state information of the hardware layer and the historical state information of the hardware layer;

[0172] Step 306: Determine whether the error value is less than a preset value; if so, execute step 310; otherwise, execute step 307;

[0173] Step 307: Determine an initial loss value of the twin reverse inference loss function based on the first state information, the historical state information of the hardware layer, the historical feature information of the business layer, and the historical feature vector of the software layer;

[0174] Step 308: Determine a target weight matrix corresponding to the at least one target component based on the initial loss value, the preset adjustment factor, and the basic weight matrix of the at least one target component;

[0175] Step 309: Adjust the first digital twin model based on the target weight matrix to obtain a second digital twin model;

[0176] Step 310: Determine the predicted state information of the hardware layer as the target state information.

[0177] In the above embodiment, based on the global characteristics of the business layer, the business layer characteristic information is generated, which can be expressed as F yewu Through multi-agents, multiple nodes in the software layer can be processed and feature vectors F of multiple nodes in the software layer can be extracted. ruan-1 、F ruan-2 ,...,F ruan-n Then, through the deduction network, the status of the business layer and software layer is passed step by step, and a preliminary result of predicting the status of the hardware layer is generated, which can be specifically expressed as:

[0178] Yj pred =f(F yewu ,F ruan-1 、F ruan-2 ,...,F ruan-n )

[0179] Among them, Yj pred Predict the first state information of the hardware layer for the Siamese network.

[0180] Furthermore, the twin reverse inference loss function in the embodiment of the present application can be specifically expressed as:

[0181]

[0182] Among them, Lossfan is the twin reverse inference loss value; Yj real 、F yewu-real 、F yewu-i-real They represent the historical state information extracted from the real data source, the historical feature information of the business layer, and the historical feature vector of the software layer respectively; α, β, γ represent Yj real 、Fyewu-real 、F yewu-i-real The corresponding weight.

[0183] Therefore, the above embodiment collects and integrates the above feature information in real time through the business layer module of the digital twin model to form a feature vector of the business layer. Then, through the multi-agent method (Multi-Agent) or automated scanning tools (such as ServiceNow), the feature vectors of multiple target components in the software layer are extracted. In this way, based on the business layer feature information and the software layer feature vector, the feature information of the business layer and the software layer is input through the deduction network f(·) to predict the first state information of the hardware layer (such as CPU usage, memory occupancy, etc.). Among them, the deduction network can use a neural network, a regression model or a rule-based algorithm to simulate the impact of the business layer and the software layer on the hardware layer. Finally, reverse deduction can be performed based on the historical state information, and the hardware layer state information can be finally predicted through multiple iterative optimizations.

[0184] In this way, the present embodiment combines characteristic information from the business and software layers with historical data to dynamically predict hardware status, ensuring accurate results. Furthermore, the model can dynamically adjust based on real-time data to adapt to operational changes in complex systems. Furthermore, error thresholds can be used to prevent upgrade failures caused by excessive prediction errors, ensuring system stability.

[0185] Optionally, performing reverse deduction based on the first state information and the historical state information through the twin reverse deduction loss function to predict target state information of the hardware layer includes:

[0186] Determining an initial loss value of the twin reverse inference loss function based on the first state information, the historical state information of the hardware layer, the historical feature information of the business layer, and the historical feature vector of the software layer;

[0187] Determining a target weight matrix corresponding to the at least one target component based on the initial loss value, the preset adjustment factor, and the basic weight matrix of the at least one target component;

[0188] Adjusting the first digital twin model based on the target weight matrix to obtain a second digital twin model;

[0189] Executing a target operation according to the second digital twin model until an error value between the predicted second state information of the hardware layer and the historical state information of the hardware layer is less than a preset value, and determining that the predicted state information of the hardware layer is the target state information;

[0190] Among them, the target operation includes: obtaining the characteristic information of the business layer from the second digital twin model; extracting the characteristic vectors corresponding to the multiple target components in the software layer of the second digital twin model; predicting the third state information based on the characteristic information of the business layer and the characteristic vectors corresponding to the multiple target components in the software layer; based on the third state information and the historical state information, reverse deduction is performed through the twin reverse deduction loss function to predict the second state information.

[0191] It should be noted that the above-mentioned first status information can be the status data of the current hardware layer (such as CPU usage, memory usage, etc.). The historical status information can be the historical status data of the hardware layer, such as the operation records over the past period of time. The business layer historical feature information, that is, the feature vector of the business layer, such as business consistency score, traffic, response time, etc. The software layer historical feature vector, that is, the node feature vector of the software layer, such as CPU usage, memory usage, number of threads, etc.

[0192] In some specific embodiments, see Figure 3 Based on the initial loss value of the twin reverse deduction, the deviation between the hardware layer prediction and the actual state is quantified by the initial loss value. The twin information transmission of the business layer, software layer, and hardware layer can be dynamically adjusted to optimize the first digital twin model and upgrade the interaction strategy of the business layer, software layer, and hardware layer in the first digital twin model. Among them, the weight matrix can be dynamically adjusted according to the loss value to reduce the weight of high-error components and increase the weight of low-error components. The target weight matrix can be calculated and expressed by the following formula:

[0193] W jadjust =W jbase *(1-γ1*Lossfan);

[0194] Among them, Wj adjust Represents the target weight matrix after dynamic adjustment; Wj base Represents the basic weight matrix; γ1 represents the adjustment factor to adjust the loss function, that is, the preset adjustment factor.

[0195] Furthermore, based on Yj pred The prediction results are reversed and the actual status information Yj of the hardware layer is directly obtained real , calculating the error between the two and using this error to correct the hardware layer in the first digital twin model. For example, in the cloud platform, after adjusting the weight matrix, the digital twin model recalculates the dependencies between components, reducing the weights of components with high errors and increasing the weights of components with low errors, thereby optimizing the model's predictive capabilities. This weight adjustment allows the model to better align with the actual system state and improves predictive accuracy.

[0196] In one specific embodiment, feature information of the business layer and the software layer (e.g., business consistency score, software layer node feature vector) is extracted from the second digital twin model. Feature vectors of multiple target components in the software layer (e.g., CPU usage, memory occupancy) are extracted. Based on the feature information and the software layer feature vectors, third state information of the hardware layer (e.g., CPU usage, memory occupancy) is predicted. Based on the third state information and the historical state information, the loss value is calculated again through the twin back reasoning loss function to predict the second state information of the hardware layer. Finally, if the error value of the predicted second state information and the historical state information is less than a preset threshold (e.g., ∈ = 0.01), the iteration is stopped and the target state information is output. Otherwise, the foregoing steps are repeated until the error meets the condition.

[0197] In this way, the embodiments of the present application ensure that the prediction result of the hardware layer state information is highly consistent with the actual state through multiple iterations and optimization, so that the digital twin model can be dynamically adjusted according to real-time data to adapt to the running changes of a complex system.

[0198] Optionally, the determining of the first upgrade priority of each target component based on the third score comprises:

[0199] determining the first upgrade priority of each target component based on the third score, an importance parameter corresponding to each target component, a traffic sensitivity of each target component, and a historical failure rate of each target component.

[0200] The importance parameter is used to represent the importance degree of real-time information feedback of the target component, and the target parameter information includes the traffic sensitivity and the historical failure rate of each target component.

[0201] In some specific embodiments, a large-scale IT system upgrade component priority determination operator is provided:

[0202]

[0203] wherein Ps(i) represents the first priority of the target component i; S(i) represents the initial priority of the target component i before upgrade (which can be preset according to the basic attributes of the component such as importance, dependency relationship, historical failure rate, etc.); S i (σ) represents the overall information comprehensive score of all twin components i of the target component; Cy i Rf(i) represents the dynamic adjustment parameter of the target component i, which can be determined in advance according to the capacity, load capacity or business characteristics of the target component i; Rf i Fl(i) represents the importance of real-time feedback of component twin information; Fl i Lr(i) represents the traffic sensitivity (from Lri Hg represents the historical failure rate of IT systems (from Bw i characterization).

[0204] Optionally, the method further includes:

[0205] Determining a target rate for updating the first digital twin model based on the change rates of the hardware layer, the software layer, and the business layer in the first digital twin model, the time constant for responding to input signals, external input data, and the connection weights between any two layers among the hardware layer, the software layer, and the business layer;

[0206] Determining a target domain of the first digital twin model through a machine learning classification network based on the target rate at which the first digital twin model is updated and the target rates of a preset number of digital twin models before the first digital twin model is updated;

[0207] The target component in the first digital twin model and components in other digital twin models having the target domain are upgraded.

[0208] In some embodiments, such as Figure 4 As shown in the figure, the Liquid Recurrent Model (LRM) can be used to capture the time-varying dependencies between different layers, thereby achieving dynamic dependency updates between components during large-scale IT system upgrades.

[0209] Specifically, the target domain of the first digital twin model can be determined by the following formula, which is specifically expressed as:

[0210]

[0211] Among them, x r-total is the comprehensive rate of change and update of the twin model, x r1 、x r2 、x r3 Ti is the hardware layer rate, software layer rate, and business layer rate of the current twin model change. r1 、Ti r2 、Ti r3 is the time constant of the twin model hardware layer, software layer and business layer’s response to the input signal, u r1 、u r2 、u r3 For external input to the hardware layer, software layer, and business layer of the twin model during update, w ij1 、w ij2 、w ij3are the connection weights between the two components of the twin model hardware layer, software layer, and business layer, respectively, and f is a nonlinear function.

[0212] Where N(r)=SVM(x r-total-1 ,x r-total-2 ,x r-total-3 ,...,x r-total-10 );x r-total-1 ,x r-total-2 ,x r-total-3 ,...,x r-total-10 It represents the value of the comprehensive update rate of the first ten times of the twin model during the upgrade; SVM is a machine learning classification network; N(r) represents the target domain of the r-th twin model.

[0213] It is understandable that the domains of the components in the digital twin model can be determined based on system complexity, categorizing the twin models into ten or twenty types. This approach can also simplify the complex relationships between digital twins. After determining the domain, use twin models within the same neighborhood and try to upgrade twin components within the same neighborhood together to reduce upgrade conflicts caused by cross-neighborhood dependencies.

[0214] Therefore, the embodiments of the present application can achieve accurate response to the system status by dynamically calculating the update target rate of the digital twin model, combining historical data with real-time parameters (such as change rate, time constant, external input and hierarchical connection weight); using machine learning classification networks to divide components into predefined fields to improve the targeted nature of upgrade strategies; and finally, centrally upgrade components within the same field to reduce cross-field dependency conflicts and reduce system risks. At the same time, through dynamic adjustment and field division, it optimizes upgrade efficiency and ensures system stability and consistency.

[0215] Optionally, the method further includes:

[0216] In a case where a plurality of target components belonging to the target domain are in the process of being upgraded, determining verification values ​​of all target components belonging to the target domain based on verification weights and verification values ​​of neighboring components in neighboring domains of the target domain; wherein the neighboring components are components in neighboring domains of the target domain;

[0217] Based on the verification value, consistency effects of the plurality of target components during the upgrade process are determined.

[0218] In some other embodiments, a phased multi-hop neighborhood verification formula can be proposed based on the twin model to confirm that relevant components and data in the domain remain consistent during the upgrade process. Specifically, it can be expressed as:

[0219]

[0220] Among them, Fdl k (r) is the overall verification value of the domain where the twin component r is located in the IT system upgrade stage k; Fd k (r,dv) represents the multi-hop neighborhood set related to the twin component r in the upgrade stage k; ρ n is the verification weight of component n in the unified domain, which is used to reflect the importance of the component in the neighborhood; Val n (k) is the verification value of component n in stage k, which represents the data consistency and state synchronization of the component with other components in the neighborhood (which can be compared through the database, etc.); K represents the total number of stages that the upgraded system needs to go through, that is, the entire upgrade process is divided into K stages; k represents the index of the current upgrade stage, which is used to identify the current processing stage.

[0221] Understandably, Fdl k (r) represents the overall verification value of the neighborhood where component r is located in the upgrade phase k. This value is a sum value used to measure the data consistency and state synchronization of all components in the neighborhood in this phase. k When (r) reaches a certain threshold, it indicates that the relevant components in the neighborhood are in good condition and the data are consistent; conversely, if the value is low, it indicates that there are potential problems that require further inspection and processing.

[0222] In addition, Fd k (r, dv) is the set of multi-hop neighborhoods of component r in stage k. This set includes all components in the same neighborhood as component r at the current stage, as well as components that were adjacent when the order was determined in step A3. This allows for a more comprehensive capture of the state of the twin component r.

[0223] In this way, the embodiment of the present application dynamically optimizes the verification evaluation of the target domain components by introducing the verification weights and verification values ​​of the neighborhood components in the adjacent fields, thereby improving the accuracy of the consistency judgment of the component status during the upgrade process; at the same time, based on the verification value, the consistency effect between components is quantified, potential conflicts or anomalies can be discovered in advance, the upgrade risks caused by cross-domain dependencies are reduced, and the system stability and coordination of the upgrade process are ensured.

[0224] Please refer to Figure 4 , Figure 4 : is a structural diagram of a system upgrade device in an embodiment of the present application. The system upgrade device 400 specifically includes:

[0225] Component determination module 401, configured to determine multiple target components in a system to be upgraded, wherein the system to be upgraded includes a hardware layer, a software layer, and a business layer, and the target component is one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer;

[0226] A model construction module 402 is configured to construct a first digital twin model based on target parameter information of each target component in at least one of the target components and an association relationship between at least two of the target components obtained by topological mapping the system to be upgraded, wherein the first digital twin model includes at least two of the target components;

[0227] The component upgrade module 403 is used to determine the upgrade priorities of at least two target components based on the first digital twin model, and upgrade the system to be upgraded based on the upgrade priorities of at least two target components.

[0228] Optionally, the component upgrade module includes:

[0229] A first acquisition submodule is configured to acquire a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during system upgrades;

[0230] An information prediction submodule, configured to predict the target state information of the hardware layer through a twin reverse deduction loss function based on the characteristic information of the business layer, and determine a target loss value corresponding to the twin reverse deduction loss function based on the target state information;

[0231] a second acquisition submodule, configured to obtain a third score for the first digital twin model based on the first score, the second score, and the target loss value;

[0232] A first determining submodule, configured to determine a first upgrade priority of each target component based on the third score;

[0233] a second determining submodule, configured to determine a second upgrade priority of each target component based on the first upgrade priority and the third score of each target component;

[0234] The component upgrade submodule is used to upgrade each of the target components in sequence according to the second upgrade priority of each of the target components.

[0235] Optionally, the component determination module is further configured to:

[0236] Identifying the multiple target components in the system to be upgraded;

[0237] Determining, based on the total number of the plurality of nodes to be verified included in the plurality of target components, the importance weight of each of the nodes to be verified, the uncertainty parameter of the data corresponding to each of the nodes to be verified, and the falsity parameter of the data;

[0238] The security of the multiple target components is verified based on the security parameter value corresponding to each of the nodes to be verified at the first moment.

[0239] Optionally, the target parameter information includes at least one of component status information, configuration parameters, performance indicator data, and behavior information;

[0240] The system upgrade device 400 further includes at least one of the following:

[0241] a first acquisition module, configured to acquire component status information of the at least one target component through an integrated monitoring tool;

[0242] a second acquisition module, configured to parse a configuration file of the system to be upgraded based on an automated management tool, and extract the configuration parameters of at least one of the hardware layer, the software layer, and the business layer;

[0243] A third collection module is used to collect the performance indicator data of the system to be upgraded based on a performance monitoring tool;

[0244] The fourth collection module is used to extract the behavior information through the system log of the system to be upgraded.

[0245] Optionally, the association relationship between at least two of the target components includes a first association relationship, a second association relationship, and a third association relationship;

[0246] Among them, the first association relationship is used to represent the physical connection relationship between each of the target components belonging to hardware in the at least one target component, the second association relationship is used to represent the logical dependency relationship between each of the target components belonging to software in the at least one target component, and the third association relationship is used to represent the interaction relationship between the software layer and the business layer.

[0247] Optionally, the second acquisition submodule includes:

[0248] a first determining unit, configured to determine the first score corresponding to the software layer based on the number of data sources of the software layer, the truth values ​​of log files corresponding to components of the software layer, the truth values ​​of monitoring data, and the weights corresponding to the components of the software layer;

[0249] The target parameter information includes a degree of truth of a log file corresponding to a component in the software layer and a degree of truth of monitoring data.

[0250] Optionally, the second obtaining sub-module comprises:

[0251] The second determining unit is configured to determine the second score corresponding to the business layer based on a total number of components in the business layer, state information of each target component in all component combinations in the business layer, a degree of truth of business data of each target component in at least one component combination, and a data similarity parameter between two target components in the at least one component combination.

[0252] The business layer comprises at least one component combination, and the component combination comprises any two different target components in the business layer.

[0253] Optionally, the information predicting sub-module comprises:

[0254] The first obtaining unit is configured to obtain feature information of the business layer from the first digital twin model.

[0255] The information extracting unit is configured to extract feature vectors corresponding to a plurality of target components in the software layer of the first digital twin model.

[0256] The first predicting unit is configured to predict first state information of the hardware layer based on the feature information of the business layer and the feature vectors corresponding to the plurality of target components in the software layer.

[0257] The second predicting unit is configured to predict target state information of the hardware layer by performing back propagation based on the first state information and historical state information and the twin back propagation loss function.

[0258] The historical state information comprises historical state information of the hardware layer, historical feature information of the business layer, and historical feature vectors of the software layer.

[0259] Optionally, the second predicting unit is configured to:

[0260] determine an initial loss value of the twin back propagation loss function based on the first state information, historical state information of the hardware layer, historical feature information of the business layer, and historical feature vectors of the software layer.

[0261] determine a target weight matrix corresponding to the at least one target component based on the initial loss value, a preset adjustment factor, and a basic weight matrix of the at least one target component.

[0262] Adjusting the first digital twin model based on the target weight matrix to obtain a second digital twin model;

[0263] Executing a target operation according to the second digital twin model until an error value between the predicted second state information of the hardware layer and the historical state information of the hardware layer is less than a preset value, and determining that the predicted state information of the hardware layer is the target state information;

[0264] Among them, the target operation includes: obtaining the characteristic information of the business layer from the second digital twin model; extracting the characteristic vectors corresponding to the multiple target components in the software layer of the second digital twin model; predicting the third state information based on the characteristic information of the business layer and the characteristic vectors corresponding to the multiple target components in the software layer; based on the third state information and the historical state information, reverse deduction is performed through the twin reverse deduction loss function to predict the second state information.

[0265] Optionally, the first determining submodule includes:

[0266] a third determining unit, configured to determine the first upgrade priority of each target component based on the third score, an importance parameter corresponding to each target component, and a traffic sensitivity and a historical failure rate of each target component;

[0267] The importance parameter is used to characterize the importance of the target component in performing real-time information feedback, and the target parameter information includes the flow sensitivity and historical failure rate of each target component.

[0268] Optionally, the system upgrading apparatus 400 further includes:

[0269] a fourth determining unit, configured to determine a target rate for updating the first digital twin model based on the change rates of the hardware layer, the software layer, and the business layer in the first digital twin model, a time constant for responding to an input signal, external input data, and a connection weight between any two layers of the hardware layer, the software layer, and the business layer;

[0270] a fifth determining unit, configured to determine a target domain of the first digital twin model through a machine learning classification network based on the target rate for updating the first digital twin model and the target rates of a preset number of digital twin models before the first digital twin model is updated;

[0271] A component upgrading unit is used to upgrade the target component in the first digital twin model and components in other digital twin models having the target domain.

[0272] Optionally, the system upgrading apparatus 400 further includes:

[0273] a component verification module, configured to determine, when a plurality of target components within the target domain are in the process of being upgraded, verification values ​​of all target components within the target domain based on verification weights and verification values ​​of neighboring components in neighboring domains of the target domain; wherein the neighboring components are components in neighboring domains of the target domain;

[0274] The effect determination module is used to determine the consistency effects of the multiple target components during the upgrade process based on the verification value.

[0275] The system upgrade device 400 provided in the embodiment of the present application can perform the above Figure 1 The method embodiment shown can capture the interactions and dependencies between multiple target components in the system to be upgraded into the first digital twin model, which is managed by the first digital twin model, so that the dependencies between the various levels in the upgrade process remain consistent, ensuring the synchronization of the digital twin model with the actual system, so that even when faced with complex multi-level topologies, the first digital twin model can be used for flexible adjustment and dynamic optimization. In addition, the embodiment of the present application can upgrade each target component according to the upgrade priority of each target component in the first digital twin model that is similar to the system to be upgraded, so as to ensure the consistency and stability of data during the component upgrade process, reduce upgrade conflicts, and reduce the risk of failure. The system upgrade device 300 in the embodiment of the present application has similar implementation principles and technical effects to those of the system upgrade device 300 in the embodiment of the present application. Figure 1 The method embodiment shown is not described in detail here.

[0276] The present application also provides an electronic device. Since the principle of solving the problem of the electronic device is similar to the system upgrade method in the present application, the implementation of the electronic device can refer to Figure 1 or Figure 2 The implementation of the method shown in FIG. 1 is omitted for repetition. Figure 5 As shown, the electronic device of the embodiment of the present application includes: a processor 510, which is used to read the program in the memory 520 and execute the following process:

[0277] Determining multiple target components in a system to be upgraded, the system to be upgraded comprising a hardware layer, a software layer, and a business layer, the target component being one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer;

[0278] Based on target parameter information of each target component in at least one of the target components and an association relationship between at least two of the target components obtained by performing topological mapping on the system to be upgraded, constructing a first digital twin model, wherein the first digital twin model includes at least two of the target components;

[0279] Determine the upgrade priorities of at least two of the target components based on the first digital twin model, and upgrade the system to be upgraded based on the upgrade priorities of at least two of the target components.

[0280] Among them, Figure 5 In the embodiment of the present invention, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits such as one or more processors represented by processor 510 and memory represented by memory 520. The bus architecture may also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are not further described herein. The bus interface provides an interface.

[0281] Optionally, the processor 510 is further configured to read a program in the memory 520 and execute the following steps:

[0282] Obtain a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during the system upgrade;

[0283] Based on the characteristic information of the business layer, predict the target state information of the hardware layer through the twin reverse inference loss function, and determine the target loss value corresponding to the twin reverse inference loss function based on the target state information;

[0284] Obtaining a third score for the first digital twin model based on the first score, the second score, and the target loss value;

[0285] determining a first upgrade priority for each of the target components based on the third score;

[0286] determining a second upgrade priority for each target component based on the first upgrade priority and the third score of each target component;

[0287] The target components are upgraded in sequence according to the second upgrade priority of each target component.

[0288] Optionally, the processor 510 is further configured to read a program in the memory 520 and execute the following steps:

[0289] Identifying the multiple target components in the system to be upgraded;

[0290] Determining, based on the total number of the plurality of nodes to be verified included in the plurality of target components, the importance weight of each of the nodes to be verified, the uncertainty parameter of the data corresponding to each of the nodes to be verified, and the falsity parameter of the data;

[0291] The security of the multiple target components is verified based on the security parameter value corresponding to each of the nodes to be verified at the first moment.

[0292] Optionally, the target parameter information includes at least one of component status information, configuration parameters, performance indicator data, and behavior information;

[0293] The processor 510 is further configured to read a program in the memory 520 and execute at least one of the following steps:

[0294] collecting component status information of the at least one target component through an integrated monitoring tool;

[0295] Parsing the configuration file of the system to be upgraded based on an automated management tool to extract the configuration parameters of at least one layer among the hardware layer, the software layer, and the business layer;

[0296] Collecting the performance indicator data of the system to be upgraded based on a performance monitoring tool;

[0297] The behavior information is extracted through the system log of the system to be upgraded.

[0298] Optionally, the association relationship between at least two of the target components includes a first association relationship, a second association relationship, and a third association relationship;

[0299] Among them, the first association relationship is used to represent the physical connection relationship between each of the target components belonging to hardware in the at least one target component, the second association relationship is used to represent the logical dependency relationship between each of the target components belonging to software in the at least one target component, and the third association relationship is used to represent the interaction relationship between the software layer and the business layer.

[0300] Optionally, the processor 510 is further configured to read a program in the memory 520 and execute the following steps:

[0301] Determining the first score corresponding to the software layer based on the number of data sources of the software layer, the truthfulness values ​​of log files corresponding to components of the software layer, the truthfulness values ​​of monitoring data, and the weights corresponding to the components of the software layer;

[0302] The target parameter information includes a degree of truth of a log file corresponding to a component in the software layer and a degree of truth of monitoring data.

[0303] Optionally, the processor 510 is further configured to read a program in the memory 520, and perform the following steps:

[0304] The second score corresponding to the business layer is determined based on a total number of components in the business layer, state information of each target component in all component combinations in the business layer, a degree of truth of business data of each target component in at least one component combination, and a data similarity parameter between two target components in the at least one component combination.

[0305] The business layer includes at least one component combination, and the component combination includes any two different target components in the business layer.

[0306] Optionally, the processor 510 is further configured to read a program in the memory 520, and perform the following steps:

[0307] The feature information of the business layer is obtained from the first digital twin model.

[0308] Feature vectors corresponding to the plurality of target components in the software layer of the first digital twin model are extracted.

[0309] The first state information of the hardware layer is predicted based on the feature information of the business layer and the feature vectors corresponding to the plurality of target components in the software layer.

[0310] The target state information of the hardware layer is predicted by performing back propagation based on the first state information and historical state information and the twin back propagation loss function.

[0311] The historical state information includes historical state information of the hardware layer, historical feature information of the business layer, and historical feature vectors of the software layer.

[0312] Optionally, the processor 510 is further configured to read a program in the memory 520, and perform the following steps:

[0313] The initial loss value of the twin back propagation loss function is determined based on the first state information, historical state information of the hardware layer, historical feature information of the business layer, and historical feature vectors of the software layer.

[0314] The target weight matrix corresponding to the at least one target component is determined based on the initial loss value, a preset adjustment factor, and a basic weight matrix of the at least one target component.

[0315] Adjusting the first digital twin model based on the target weight matrix to obtain a second digital twin model;

[0316] Executing a target operation according to the second digital twin model until an error value between the predicted second state information of the hardware layer and the historical state information of the hardware layer is less than a preset value, and determining that the predicted state information of the hardware layer is the target state information;

[0317] Among them, the target operation includes: obtaining the characteristic information of the business layer from the second digital twin model; extracting the characteristic vectors corresponding to the multiple target components in the software layer of the second digital twin model; predicting the third state information based on the characteristic information of the business layer and the characteristic vectors corresponding to the multiple target components in the software layer; based on the third state information and the historical state information, reverse deduction is performed through the twin reverse deduction loss function to predict the second state information.

[0318] Optionally, the processor 510 is further configured to read a program in the memory 520 and execute the following steps:

[0319] Determining the first upgrade priority of each target component based on the third score, the importance parameter corresponding to each target component, the traffic sensitivity and the historical failure rate of each target component;

[0320] The importance parameter is used to characterize the importance of the target component in performing real-time information feedback, and the target parameter information includes the flow sensitivity and historical failure rate of each target component.

[0321] Optionally, the processor 510 is further configured to read a program in the memory 520 and perform the following steps:

[0322] Determining a target rate for updating the first digital twin model based on the change rates of the hardware layer, the software layer, and the business layer in the first digital twin model, the time constant for responding to input signals, external input data, and the connection weights between any two layers among the hardware layer, the software layer, and the business layer;

[0323] Determining a target domain of the first digital twin model through a machine learning classification network based on the target rate at which the first digital twin model is updated and the target rates of a preset number of digital twin models before the first digital twin model is updated;

[0324] The target component in the first digital twin model and components in other digital twin models having the target domain are upgraded.

[0325] Optionally, the processor 510 is further configured to read a program in the memory 520 and perform the following steps:

[0326] In a case where a plurality of the target components belonging to the target domain are in the process of being upgraded, determining verification values ​​of all target components belonging to the target domain based on verification weights and verification values ​​of neighboring components in neighboring domains of the target domain; wherein the neighboring components are components in neighboring domains of the target domain;

[0327] Based on the verification value, consistency effects of the plurality of target components during the upgrade process are determined.

[0328] The electronic device provided in the embodiment of the present application can perform the above Figure 1 or Figure 2 The implementation principle and technical effects of the method embodiment shown are similar, and will not be described in detail in this embodiment.

[0329] The present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned Figure 1 or Figure 2 The various processes of the system upgrade method embodiment in the embodiment can achieve the same technical effect, and are not described here in detail to avoid repetition. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0330] The embodiment of the present application further provides a computer program / program product, which is stored in a storage medium and is executed by at least one processor to implement the above Figure 1 The various processes of the system upgrade method embodiment can achieve the same technical effect. To avoid repetition, they will not be described here.

[0331] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection of some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0332] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may be physically included separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0333] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute some steps of the sending and receiving methods described in various embodiments of the present application. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program code.

[0334] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A system upgrade method, characterized in that: include: Determining multiple target components in a system to be upgraded, the system to be upgraded comprising a hardware layer, a software layer, and a business layer, the target component being one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer; Based on target parameter information of each target component in at least one of the target components and an association relationship between at least two of the target components obtained by performing topological mapping on the system to be upgraded, constructing a first digital twin model, wherein the first digital twin model includes at least two of the target components; Determine the upgrade priorities of at least two of the target components based on the first digital twin model, and upgrade the system to be upgraded based on the upgrade priorities of at least two of the target components.

2. The method according to claim 1, characterized in that The determining the upgrade priorities of at least two target components based on the first digital twin model, and upgrading the system to be upgraded based on the upgrade priorities of the at least two target components, includes: Obtain a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during the system upgrade; Based on the characteristic information of the business layer, predict the target state information of the hardware layer through the twin reverse inference loss function, and determine the target loss value corresponding to the twin reverse inference loss function based on the target state information; Obtaining a third score for the first digital twin model based on the first score, the second score, and the target loss value; determining a first upgrade priority for each of the target components based on the third score; determining a second upgrade priority for each target component based on the first upgrade priority and the third score of each target component; The target components are upgraded in sequence according to the second upgrade priority of each target component.

3. The method according to claim 1 or 2, characterized in that Before determining a plurality of target components in the system to be upgraded, the method further includes: Identifying multiple target components in the system to be upgraded; Determining, based on the total number of the plurality of nodes to be verified included in the plurality of target components, the importance weight of each of the nodes to be verified, the uncertainty parameter of the data corresponding to each of the nodes to be verified, and the falsity parameter of the data; The security of the multiple target components is verified based on the security parameter value corresponding to each of the nodes to be verified at the first moment.

4. The method according to claim 1 or 2, characterized in that The target parameter information includes at least one of component status information, configuration parameters, performance indicator data and behavior information; After determining multiple target components in the system to be upgraded, and before constructing the first digital twin model based on target parameter information of each target component in at least one of the target components and an association relationship between at least two target components obtained by topological mapping the system to be upgraded, the method further includes at least one of the following: collecting component status information of the at least one target component through an integrated monitoring tool; Parsing the configuration file of the system to be upgraded based on an automated management tool to extract the configuration parameters of at least one layer among the hardware layer, the software layer, and the business layer; Collecting the performance indicator data of the system to be upgraded based on a performance monitoring tool; The behavior information is extracted through the system log of the system to be upgraded.

5. The method according to claim 1 or 2, characterized in that The association relationship between at least two of the target components includes a first association relationship, a second association relationship, and a third association relationship; Among them, the first association relationship is used to represent the physical connection relationship between each of the target components belonging to hardware in the at least one target component, the second association relationship is used to represent the logical dependency relationship between each of the target components belonging to software in the at least one target component, and the third association relationship is used to represent the interaction relationship between the software layer and the business layer.

6. The method according to claim 2, characterized in that The obtaining of the first score includes: Determining the first score corresponding to the software layer based on the number of data sources of the software layer, the truth values ​​of log files and monitoring data corresponding to components of the software layer, and the weights corresponding to the components of the software layer; The target parameter information includes the truth value of the log file corresponding to the component of the software layer and the truth value of the monitoring data.

7. The method according to claim 2, characterized in that The obtaining of the second score includes: Determining the second score corresponding to the business layer based on the total number of components belonging to the business layer, status information of each target component in all component combinations in the business layer, a truth value of business data of each target component in at least one component combination, and a data similarity parameter between two target components in at least one component combination; The business layer includes at least one component combination, and the component combination includes any two different target components in the business layer.

8. The method according to claim 2 or any one of claims 6 to 7, characterized in that: The predicting of the target state information of the hardware layer by using a twin reverse deduction loss function based on the feature information of the business layer includes: Acquire characteristic information of the business layer from the first digital twin model; Extracting feature vectors corresponding to a plurality of target components in the software layer of the first digital twin model; Predicting first state information of the hardware layer based on the feature information of the business layer and feature vectors corresponding to the plurality of target components in the software layer; Based on the first state information and historical state information, reverse inference is performed through the twin reverse inference loss function to predict the target state information of the hardware layer; The historical status information includes the historical status information of the hardware layer, the historical feature information of the service layer, and the historical feature vector of the software layer.

9. The method according to claim 8, characterized in that The step of performing reverse deduction based on the first state information and the historical state information by using the twin reverse deduction loss function to predict target state information of the hardware layer includes: Determining an initial loss value of the twin reverse inference loss function based on the first state information, the historical state information of the hardware layer, the historical feature information of the business layer, and the historical feature vector of the software layer; Determining a target weight matrix corresponding to the at least one target component based on the initial loss value, the preset adjustment factor, and the basic weight matrix of the at least one target component; Adjusting the first digital twin model based on the target weight matrix to obtain a second digital twin model; Executing a target operation according to the second digital twin model until an error value between the predicted second state information of the hardware layer and the historical state information of the hardware layer is less than a preset value, and determining that the predicted state information of the hardware layer is the target state information; Among them, the target operation includes: obtaining the characteristic information of the business layer from the second digital twin model; extracting the characteristic vectors corresponding to the multiple target components in the software layer of the second digital twin model; predicting the third state information based on the characteristic information of the business layer and the characteristic vectors corresponding to the multiple target components in the software layer; based on the third state information and the historical state information, reverse deduction is performed through the twin reverse deduction loss function to predict the second state information.

10. The method according to claim 2, characterized in that Determining a first upgrade priority of each target component based on the third score includes: Determining the first upgrade priority of each target component based on the third score, the importance parameter corresponding to each target component, the traffic sensitivity and the historical failure rate of each target component; The importance parameter is used to characterize the importance of the target component in performing real-time information feedback, and the target parameter information includes the flow sensitivity and historical failure rate of each target component.

11. The method according to claim 2, characterized in that The method further comprises: Determining a target rate for updating the first digital twin model based on the change rates of the hardware layer, the software layer, and the business layer in the first digital twin model, the time constant for responding to input signals, external input data, and the connection weights between any two layers among the hardware layer, the software layer, and the business layer; Determining a target domain of the first digital twin model through a machine learning classification network based on the target rate at which the first digital twin model is updated and the target rates of a preset number of digital twin models before the first digital twin model is updated; The target component in the first digital twin model and components in other digital twin models having the target domain are upgraded.

12. The method according to claim 11, characterized in that The method further comprises: In a case where a plurality of the target components belonging to the target domain are in the process of being upgraded, determining verification values ​​of all target components belonging to the target domain based on verification weights and verification values ​​of neighboring components in neighboring domains of the target domain; wherein the neighboring components are components in neighboring domains of the target domain; Based on the verification value, consistency effects of the plurality of target components during the upgrade process are determined.

13. A system upgrade device, characterized in that: include: a component determination module, configured to determine a plurality of target components in a system to be upgraded, wherein the system to be upgraded includes a hardware layer, a software layer, and a business layer, wherein the target component is one of a component belonging to the hardware layer, a component belonging to the software layer, and a component belonging to the business layer; a model construction module, configured to construct a first digital twin model based on target parameter information of each target component in at least one of the target components and an association relationship between at least two of the target components obtained by topological mapping the system to be upgraded, wherein the first digital twin model includes at least two of the target components; A component upgrade module is used to determine the upgrade priorities of at least two target components based on the first digital twin model, and to upgrade the system to be upgraded based on the upgrade priorities of at least two target components.

14. The device according to claim 13, characterized in that The component upgrade module includes: A first acquisition submodule is configured to acquire a first score and a second score; wherein the first score is used to evaluate the integration effect of the data included in the software layer, and the second score is used to measure the consistency of the business data included in the business layer during system upgrades; An information prediction submodule, configured to predict the target state information of the hardware layer through a twin reverse deduction loss function based on the characteristic information of the business layer, and determine a target loss value corresponding to the twin reverse deduction loss function based on the target state information; a second acquisition submodule, configured to obtain a third score for the first digital twin model based on the first score, the second score, and the target loss value; a first determining submodule, configured to determine a first upgrade priority of each target component based on the third score; a second determining submodule, configured to determine a second upgrade priority of each target component based on the first upgrade priority and the third score of each target component; The component upgrade submodule is used to upgrade each of the target components in sequence according to the second upgrade priority of each of the target components.

15. An electronic device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the steps of the system upgrade method according to any one of claims 1 to 12 are implemented.

16. A computer-readable storage medium for storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the system upgrade method according to any one of claims 1 to 12 are implemented.

17. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps in the system upgrade method according to any one of claims 1 to 12.

Images