Container migration guiding system and method based on large model and electronic equipment

Abstract

The invention provides a container migration guiding system and method based on a large model and electronic equipment, and the method comprises the steps: obtaining container environment data through the mutual cooperation of an intelligent decision engine module and a migration execution engine module in the container migration guiding system based on the large model; a dynamic migration strategy is generated based on container environment data through a preset large model unit, a pre-check list corresponding to the dynamic migration strategy is generated, resource availability verification, dependency service topology verification and compliance verification are carried out through the pre-check list, and after the verification of the resource availability verification, the dependency service topology verification and the compliance verification are passed, the dynamic migration strategy is generated. And executing the dynamic migration strategy to migrate the request traffic of the original container meeting the container migration condition to the new container. Therefore, the running state of the container and the cloud environment data can be dynamically analyzed, and the corresponding dynamic container migration strategy is adaptively generated to migrate the container so as to adapt to the dynamically changing cloud environment, and the container migration efficiency can be improved.

Description

Technical Field

[0001] This application relates to the field of database system operation and maintenance technology, and in particular to a container migration and bootstrapping system, method and electronic device based on a large model. Background Technology

[0002] In the field of internet technology, containers are a lightweight virtualization technology that uses operating system-level isolation mechanisms to package an application and its dependencies into an independent, portable unit, ensuring that the application can run in the same way in different environments. The application's dependencies include databases, configuration files, runtime environments, and so on.

[0003] In the daily use of containers, it is often necessary to migrate running containers from one host machine (including physical machines or virtual machines) to another. The technology that enables the migration of containers from one host machine to another is called container migration technology. Traditional container migration technologies typically determine whether to migrate a container based on migration decisions with fixed rules, which can lead to poor flexibility and difficulty in adapting to dynamically changing cloud environments. Summary of the Invention

[0004] In view of this, embodiments of this application provide a container migration bootstrapping system, method, and electronic device based on a large model to dynamically analyze container operating status and cloud environment data, and generate container migration strategies that adapt to dynamic changes in the cloud environment.

[0005] In a first aspect, embodiments of this application provide a container migration bootstrapping system based on a large model, wherein the method includes: The intelligent decision engine module is pre-configured with a large model unit. The intelligent decision engine module is used to acquire container environment data, including container environment monitoring metrics, cloud service provider quotes, and container topology. Based on this container environment data, the large model unit generates a dynamic migration strategy. The migration execution engine module is used to obtain the dynamic migration strategy, generate a pre-inspection list corresponding to the dynamic migration strategy, and perform resource availability verification, dependent service topology verification, and compliance verification based on the pre-inspection list. If the resource availability verification, dependent service topology verification, and compliance verification all pass, the dynamic migration strategy is executed to migrate the request traffic of the original container to the new container, wherein the original container is a container that meets the container migration conditions.

[0006] Secondly, embodiments of this application provide a container migration bootstrapping method based on a large model, wherein the method is applied to the container migration bootstrapping system based on a large model provided in the first aspect, and the method includes: Acquire container environment data, including: container environment monitoring metrics, cloud service provider quotes, and container topology; and generate dynamic migration strategies based on the container environment data. Based on the dynamic migration strategy, a pre-inspection list corresponding to the dynamic migration strategy is generated, and resource availability verification, dependent service topology verification, and compliance verification are performed based on the pre-inspection list. If the resource availability verification, the dependent service topology verification, and the compliance verification all pass, the dynamic migration strategy is executed to migrate the request traffic of the original container to the new container, wherein the original container is a container that meets the container migration conditions.

[0007] Thirdly, embodiments of this application provide an electronic device, wherein the electronic device includes: a processor; and a memory storing a program; wherein the program includes instructions, which, when executed by the processor, cause the processor to perform the container migration bootstrapping method based on a large model as described in the second aspect.

[0008] Fourthly, embodiments of this application provide a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the large-model-based container migration bootstrapping method described in the second aspect.

[0009] The beneficial effects of this application are: This application provides a container migration bootstrapping system, method, and electronic device based on a large model. The system utilizes a smart decision engine module and a migration execution engine module to acquire container environment data. A pre-defined large model unit generates a dynamic migration strategy based on this data, and a corresponding pre-check list is generated. This list is used to verify resource availability, dependent service topology, and compliance. Once all three checks pass, the dynamic migration strategy is executed, migrating request traffic from the original container that meets the migration criteria to the new container. This allows for dynamic analysis of container runtime status and cloud environment data, adaptively generating corresponding dynamic container migration strategies to migrate containers, thus adapting to dynamically changing cloud environments and improving container migration efficiency. Attached Figure Description

[0010] Further details, features, and advantages of this application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1A schematic diagram of a system architecture for the container migration and bootstrapping system based on a large model provided in this application is shown. Figure 2 This paper illustrates another system architecture diagram of the container migration and bootstrapping system based on a large model provided in this application; Figure 3 This paper presents a schematic diagram of the logical architecture of the intelligent decision engine module provided in this application. Figure 4 A schematic diagram of the logical architecture of the cognitive interaction interface provided in this application is shown. Figure 5 This paper illustrates a logical architecture diagram of the evolutionary learning unit provided in this application. Figure 6 This paper illustrates a logical architecture diagram of the migration execution engine module provided in this application. Figure 7 This paper illustrates a cross-module collaboration process among modules within the container migration bootstrapping system based on a large model provided in this application. Figure 8 A schematic diagram of a container migration bootstrapping method based on a large model provided in this application is shown. Figure 9 A structural block diagram of an exemplary electronic device that can be used to implement embodiments of this application is shown. Detailed Implementation

[0011] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0012] It should be understood that the steps described in the method embodiments of this application may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this application is not limited in this respect.

[0013] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this application are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0014] It should be noted that the terms "a" and "a plurality of" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0015] To adapt to the dynamically changing cloud environment, this application provides a large-model-based container migration bootstrapping system, method, and electronic device. Firstly, this application provides a large-model-based container migration bootstrapping system; secondly, it provides a large-model-based container migration bootstrapping method, which can be applied to the large-model-based container migration bootstrapping system provided in the first aspect. The large-model-based container migration bootstrapping system is a software system that can be deployed on any electronic device with container migration capabilities, including but not limited to: personal mobile terminals, computers, servers, or server clusters, etc.

[0016] Firstly, this application provides a container migration and bootstrapping system based on a large model, which can, as Figure 1 As shown, system 1 includes: The intelligent decision engine module 11 is pre-configured with a large model unit. The intelligent decision engine module is used to acquire container environment data, including container environment monitoring indicators, cloud service operator quotations, and container topology; and generates dynamic migration strategies based on the container environment data through the large model unit. The migration execution engine module 12 is used to obtain the dynamic migration strategy, generate a pre-inspection list corresponding to the dynamic migration strategy, and perform resource availability verification, dependent service topology verification, and compliance verification based on the pre-inspection list. If the resource availability verification, dependent service topology verification, and compliance verification all pass, the dynamic migration strategy is executed to migrate the request traffic of the original container to the new container, wherein the original container is a container that meets the container migration conditions.

[0017] The intelligent decision-making engine module and migration execution engine module in the large-model-based container migration bootstrapping system provided in this application work together to acquire container environment data. Based on this data, a dynamic migration strategy is generated using a pre-defined large-model unit, along with a corresponding pre-check list. This list is used to verify resource availability, dependent service topology, and compliance. Once all three checks pass, the dynamic migration strategy is executed, migrating request traffic from the original container that meets the migration criteria to the new container. This allows for dynamic analysis of container runtime status and cloud environment data, adaptively generating corresponding dynamic container migration strategies to migrate containers, adapting to dynamically changing cloud environments and improving container migration efficiency.

[0018] The following section will provide a detailed explanation of the container migration and bootstrapping system and method based on a large model provided in this application, using specific examples: In some possible embodiments, the large-model-based container migration bootstrapping system provided in this application may also include, for example... Figure 2 As shown, it also includes: The cognitive interaction interface 13 is used to obtain natural language commands input by the user and identify the command type corresponding to the natural language commands. The command type includes: strategy adjustment and status query.

[0019] The intelligent decision engine module 11 is also used to respond to the instruction type corresponding to the natural language instruction, generate a user container migration strategy, and display the migration process corresponding to the user container migration strategy in natural language on the cognitive interaction interface.

[0020] As one implementation method, the execution logic of this cognitive interaction interface can be as follows: Figure 3 As shown: This cognitive interactive interface constructs intelligent operation logic driven by natural language, enabling "conversational" human-computer collaboration in container migration management. By parsing the container migration instructions in natural language form input by the user, it can identify the instruction type corresponding to the natural language instruction, thereby identifying the user's actual intention, whether it is a strategy adjustment or a container status query.

[0021] The cognitive interaction interface includes an intelligent parsing layer, a dynamic negotiation layer, and an execution feedback layer. The intelligent parsing layer deploys a multimodal intent recognition unit, which identifies various types of user input, such as voice, text, and images. Using a domain-adaptive large language model, it accurately converts user natural language commands (such as migrating a database container to a more secure area) into structured technical parameters and automatically supplements implicit constraints from the business context, such as simultaneously meeting GDPR (a compliance regulation) requirements.

[0022] Furthermore, through the interactive decision tree logic mechanism pre-built in the dynamic negotiation layer, when conflicting requirements exist, such as the trade-off between low cost and high performance, optimization options are generated in real time, and corresponding comparison indicators are quantified. For example, option A can save 85 yuan / day, but the latency increases by 12ms. Then, the output is presented in multiple modalities such as voice, visual images, and text. A visualized 3D evaluation radar chart can be used, with cost, security, and performance as the three evaluation indicators to intuitively display the corresponding content to the user.

[0023] In this embodiment of the application, the execution feedback layer is pre-configured with an immersive monitoring dashboard. During the migration process of the dynamic migration strategy, the real-time progress can be displayed in the form of animation, and executable suggestions and compliance audit traces can be automatically pushed when an anomaly occurs.

[0024] In real-world applications, users can input natural language commands such as "Help me migrate high-load containers to a more cost-effective availability zone" through a cognitive interactive interface. The system can then automatically parse the intent, generate a migration strategy, and provide feedback on the execution process and expected results in natural language. This interaction method greatly simplifies operations and maintenance, allowing users to complete complex migration tasks without needing in-depth knowledge of underlying resource details. Simultaneously, the cognitive interactive interface translates complex resource scheduling logic into easily understandable natural language feedback, helping users clearly grasp the migration progress and resource optimization status, thereby improving overall operational efficiency and transparency.

[0025] In this application, the intelligent decision engine module is a software module and serves as the core decision-making module of the entire large-model-based container migration guidance system. Its architecture is a multimodal data processing and hybrid intelligent decision-making architecture. Specifically, the intelligent decision engine module includes an environment perception layer, a decision reasoning layer, and a verification feedback layer. Container environment data can be acquired through the environment perception layer. As one implementation method, the environment perception layer deploys a lightweight data probe cluster to achieve multi-dimensional data collection at the second level (e.g., 5 seconds / time) to obtain container runtime indicators and cloud environment metadata. Container environment data refers to the specific parameters of the runtime environment of each container, which may include: container runtime indicators and cloud environment metadata. Container runtime status metrics may include: CPU core count, GPU utilization, memory usage, I / O read / write count, etc. A container cloud environment is provided by cloud providers via the internet, offering computing resources (including servers, storage, networks, software, etc.). Users are required to use resources on demand, scale elastically, and pay according to usage. Cloud environment data may include: real-time pricing from cloud providers, availability zone capacity, SLA (Service Level Agreement) breach records, etc.

[0026] As one implementation method, the intelligent decision engine module is also used for: The system acquires the running status data of each container and the cloud environment status data according to a preset cycle. Based on the running status data of each container and the cloud environment status data, a resource call interface is generated based on the dynamic migration strategy; The request traffic from the original container is migrated to the new container based on the resource call interface.

[0027] The preset period can be flexibly set according to actual needs. When the preset period is 0, it is to obtain the running status data of each container and the cloud environment status data in real time.

[0028] In this application, standardized coding can be applied to the resource configurations of heterogeneous cloud platforms. For example, resource configuration information provided by different cloud vendors can be uniformly converted into a data structure with a preset format to facilitate subsequent processing and analysis. Standardized coding can include abstract descriptions of computing resources, storage resources, network resources, etc., enabling heterogeneous data from different cloud platforms to be compared, scheduled, and optimized within a unified framework, thereby improving the flexibility and efficiency of resource management. Through this standardized coding mechanism, the system can quickly identify resource differences in heterogeneous cloud environments and intelligently recommend the optimal migration path based on real-time load conditions. For example, AWS EC2 (Amazon Elastic Compute Cloud) instance types can be mapped to common performance parameters, which can then be used for unified scheduling.

[0029] At the decision-making and reasoning layer, the system utilizes a large model to perform in-depth analysis of the collected multi-dimensional data, predicting the performance of containers in the target environment after migration. Simultaneously, the verification and feedback layer simulates migration scenarios, verifies the feasibility of the strategy, and dynamically adjusts the recommended solution based on the feedback results. This closed-loop architecture not only improves the success rate of container migration but also effectively reduces operational costs.

[0030] Specifically, the core decision-making process of the decision-making reasoning layer includes the following three stages: 1) Short-term forecasting. As one implementation method, it can be as follows: Figure 4 As shown, by acquiring container environment data as input, and then preprocessing the data to filter and categorize it, the monitoring metrics, cloud service provider quotes, and network topology are determined. Then, the temporal prediction model (TCN) in the pre-built large model unit extracts the temporal features of the monitoring metrics, obtains their temporal characteristics, and predicts the load trend over a future period (e.g., the next 30 minutes), thus identifying potential hotspots in advance.

[0031] 2) Multi-objective optimization. The cost optimizer obtains the operating costs of each container and optimizes these costs. The latency calculation engine calculates the specific latency of each container based on the network topology. In the multi-objective optimization phase, a pre-configured multi-objective optimizer optimizes based on multiple dimensions such as time-series characteristics, cost, and latency. It integrates load prediction results, cost optimization schemes, and latency calculation results, and a large model unit generates a large model LLM strategy to obtain a dynamic migration strategy. The large model LLM can then evaluate and weigh multiple candidate target cloud platforms to ultimately determine the most economical and efficient migration target while meeting performance requirements. Throughout this process, the system incorporates Service Level Agreement (SLA) requirements to provide early warnings and avoidance of potential service quality violations, ensuring the continuity and stability of business operations.

[0032] 3) Strategy Generation. During the strategy generation phase, the system generates specific container migration strategies based on short-term predictions and multi-objective optimization results, combined with predefined strategy templates and real-time environment conditions. These strategies include key elements such as migration priority, target cloud platform selection, resource allocation schemes, and migration execution time windows. The generated strategies are then passed to the execution layer for implementation and simultaneously fed back to the verification feedback layer for simulation verification, ensuring the strategies' feasibility and effectiveness in real-world environments. Through this process, the system achieves end-to-end automation from data collection to strategy implementation, significantly improving the intelligence and execution efficiency of container migration in heterogeneous cloud environments.

[0033] Furthermore, the dynamic migration strategy is evaluated to determine its feasibility. If feasible, the evaluation passes, the corresponding dynamic migration strategy is output, and the result is sent to the migration execution engine module. If infeasible, the strategy is rejected, and the result is fed back to the environment-aware layer (also known as the data layer).

[0034] The verification feedback layer introduces a digital twin simulation environment to pre-execute and verify the generated migration strategy: resource conflict detection (checking target node port occupancy, storage quota, etc.), performance impact assessment (predicting latency changes after migration through traffic mirroring), secondary cost calibration (comparing the calculation results with the latest price API update from the cloud provider), and strategies that pass verification enter the execution queue. Failed strategies trigger an adaptive learning mechanism, and the failure features are recorded in the decision knowledge base for model iteration.

[0035] In this embodiment, the intelligent decision engine module can automatically adjust and optimize the weight coefficients of containers according to the actual business scenario requirements. For example, during e-commerce promotional periods, the container performance weight can be increased by 40%. This intelligent decision engine module can also simultaneously process structured monitoring data and unstructured operation and maintenance documents (such as cloud service provider SLA text clause parsing), extract key constraints through natural language processing technology, and incorporate them into the optimization objective function, thereby achieving intelligent adaptation and optimization decisions for complex business rules. Furthermore, the intelligent decision engine module can also be used to generate a technical basis chain for each generated dynamic migration strategy to explain the generated dynamic migration strategy. For example, for generated dynamic migration strategy B, it can be explained that the effect of choosing dynamic migration strategy B is: compared to generated dynamic migration strategy A, dynamic migration strategy B can reduce carbon dioxide emissions by 23 kg.

[0036] As one implementation method, in a simulated 10,000-node cluster environment, the decision latency of the intelligent decision engine can be controlled within 800ms, the strategy effectiveness rate reaches 98.7%, the decision dimension is increased by 12 times compared with the traditional rule engine solution, and the anomaly prediction accuracy is increased by 65%.

[0037] In some possible embodiments, the intelligent decision engine module also includes a pre-built reinforcement learning unit, and the intelligent decision engine module is further used for: By collaborating with the large model unit, the reinforcement learning unit quantizes each container environment data into different weight parameters. Based on the weight parameters corresponding to the container environment data, a Pareto optimal dynamic migration strategy set is generated, and natural language explanation information corresponding to each dynamic migration strategy in the Pareto optimal dynamic migration strategy set is output.

[0038] As one implementation method, the intelligent decision engine module is also used for: The reinforcement learning unit, combined with the large model unit, is used to perform time-series analysis on the container environment data; Based on the time series analysis results, a predictive migration strategy is generated, and the reserved containers are migrated to the reserved instances corresponding to the predictive migration strategy based on the predictive migration strategy.

[0039] In this application, the reinforcement learning unit continuously interacts and learns from the environment to optimize the weight parameter configuration, thereby improving the decision-making quality of dynamic migration strategies. During this process, the intelligent decision engine module utilizes the powerful semantic understanding capabilities of the large model unit to automatically transform business rules, operational documents, and user requirements written in natural language into executable constraints and incorporate them into the strategy generation process. Through collaboration between the reinforcement learning unit and LLM, various indicators can be quantified into computable weight parameters. For example, based on the user's input of natural language instructions for container migration: reducing costs by 15% or ensuring latency <200ms, the system can automatically weigh the conflict between two constraints, provide a Pareto optimal solution set, and then offer corresponding natural language explanations. For instance, "Choosing solution A increases costs by $50, but avoids the loss of 2000 orders." This allows operations personnel to intuitively choose strategies, breaking the limitations of traditional technologies that only see the strategy but not the underlying reasons—a "seeing the trees but not the forest" approach to decision-making.

[0040] This achieves a seamless transition from business rules to technical decisions, further enhancing the intelligence and adaptability of the decision engine. The reinforcement learning unit optimizes strategy selection through continuous trial and error and feedback, improving the system's self-adaptability and ensuring the intelligent decision engine maintains high efficiency and stability in complex and ever-changing container environments.

[0041] It can be understood that this reinforcement learning unit is an evolutionary learning unit, pre-configured with a closed-loop learning logic mechanism of "perception-evaluation-evolution," which enables continuous autonomous optimization of the container migration strategy. In some possible embodiments, it can be as follows: Figure 5 As shown, the intelligent decision engine module is also used for: After executing the dynamic migration strategy to migrate the request traffic of the original container to the new container, the migration results are obtained and an effect evaluation matrix is ​​generated. Based on the effect evaluation matrix, it is determined whether the dynamic migration strategy is abnormal. If there is no abnormality, the container environment data corresponding to the dynamic migration strategy and the dynamic migration strategy are used as training sample data and stored in the model incremental training sample database to realize feature entry into the database, so that the model can be incrementally trained based on the data in the model incremental training sample database in the future. If the dynamic migration strategy is abnormal, then a fault analysis is performed on the dynamic migration strategy to analyze the root cause of the abnormality, and a patch strategy is generated and executed, or a strategy patch is generated to correct the abnormality.

[0042] In this embodiment, the reinforcement learning unit (or evolutionary learning unit) has the following four functions: Function 1: Multi-dimensional effect evaluation. Specifically, this reinforcement learning unit establishes 12 core indicators strongly correlated with containers, and a quantitative evaluation system based on these core indicators. The core indicators include: basic resource indicators (CPU / memory utilization changes), business impact indicators (service success rate, latency fluctuations), economic indicators (cost savings rate, ROI calculation), and stability indicators (anomaly recovery time, rollback frequency).

[0043] Function 2: Feature Engineering. A migration feature knowledge graph is constructed to convert unstructured logs into trainable vector features. For example, based on container environment data, the following migration features are extracted: environment features, load features, policy features, and outcome features. Environment features include: cloud platform type and Kubernetes version. Load features include: time-series fluctuation patterns and probability of traffic surges. Policy features include: migration path and traffic switching method. Outcome features include: actual cost deviation and SLA achievement rate, etc.

[0044] Function 3: Incremental Model Optimization. Based on the vector features constructed in Function 2, the model parameters of the pre-built large model can be fine-tuned using a dual-channel learning architecture (online channel and offline channel). For example, the online learning channel can be used to fine-tune the prediction model parameters in real time. The offline channel, also known as the batch training channel, can deeply optimize the core algorithm daily, retrain on the full dataset, and verify the effect of the new model through a model A / B testing framework. Furthermore, the winning model obtained from the verification can be hot-updated and deployed to facilitate hot switching deployment using the optimized model.

[0045] Function 4: Knowledge Accumulation and Application. By building a container migration case library, successful dynamic migration strategies are archived as best practice templates, failed migration cases are generated as pitfall avoidance guides, and specialized solutions are developed for specific scenarios (such as government cloud migration). These are then compiled into a comprehensive container migration case library. This library not only supports keyword search and scenario matching but also provides strategy optimization suggestions through an association recommendation mechanism, further improving container migration efficiency and success rate. Furthermore, the library supports continuous updates and iterations, constantly enriching its knowledge system with the latest migration practices to ensure the forward-looking and practical nature of migration strategies. This enables a virtuous cycle of evolution, where migration models become increasingly intelligent. In addition, this container migration case library can be shared to achieve cross-enterprise knowledge sharing.

[0046] After the dynamic migration strategy is executed and user requests are smoothly switched from the original container to the new container, the intelligent decision engine module can collect migration results, such as migration time, request loss rate, new container response time, resource utilization fluctuations, and other parameters. The collected migration results are then summarized and organized into a matrix, which is the effect evaluation matrix. This effect evaluation matrix contains multi-dimensional quantitative indicators of the migration effect of the dynamic migration strategy.

[0047] Furthermore, the migration effect can be analyzed based on the effect evaluation matrix, such as whether service terminals appeared, changes in response time before and after migration, etc. If the effect evaluation matrix shows that the migration was successful, the successful dynamic migration strategy can be used as training sample data and stored in the model incremental training sample database. Then, in some possible embodiments, the intelligent decision engine module can use the training sample data in the model incremental training sample database to iteratively update the algorithm weight parameters of the large model unit to generate a new intelligent decision engine module.

[0048] If the performance evaluation matrix shows problems with the migration, such as too many requests, overload of the new container, or excessive migration time, it indicates that the dynamic migration strategy has defects. In this case, the intelligent decision engine module can locate and analyze the cause of the error and generate corresponding execution patch strategies to correct the original dynamic migration strategy, such as temporarily expanding the capacity of the new container, adjusting the traffic switching rhythm, and supplementing the caching mechanism.

[0049] Actual engineering verification has shown that using the reinforcement learning unit provided in this application to provide a container migration case library improves the decision accuracy of the intelligent decision engine module by 5% per month, and after dynamic migration, the adaptation time to the new environment is shortened from 72 hours to 4 hours.

[0050] In actual migration execution, this dynamic migration strategy can be scheduled and implemented through the migration execution engine module, initiating migration tasks in priority order. During the migration process, the migration execution engine module monitors resource usage and network status in real time, automatically adjusting the migration rate to avoid impacting business operations due to sudden load fluctuations. Simultaneously, the migration execution engine maintains linkage with the verification feedback layer, feeding back execution logs and actual results to the intelligent decision engine module to continuously optimize the accuracy and adaptability of subsequent strategies. The entire process operates in a closed loop, forming a collaborative mechanism of dynamic perception, intelligent decision-making, rapid execution, and continuous optimization, effectively supporting the efficient scheduling and elastic scaling of large-scale containerized applications in multi-cloud environments.

[0051] As one implementation method, the built-in logic of this migration execution engine module can be as follows: Figure 6 As shown, it consists of four stages: Step 1: Pre-inspection and Verification. This involves acquiring the dynamic migration strategy (also known as the migration plan) and generating a corresponding pre-inspection checklist. This checklist includes multiple dimensions, specifically: resource availability verification, dependent service topology verification, and compliance review. Resource availability verification checks the available resources of each container node, such as CPU and memory reserves. Dependent service topology verification checks the communication status of each container node and the normality of network connections between containers. Compliance verification reviews whether users have the necessary permissions and whether there are any restrictions on data sovereignty areas. Specifically, this is achieved by real-time calls to the cloud platform's API and cluster status checking tools to verify each pre-inspection indicator within a set time. If any item is abnormal, LLM-assisted repair suggestion generation is automatically triggered, generating and outputting repair suggestions, and triggering an alarm process. If the checks pass, Step 2 is then executed.

[0052] Step Two: Transaction Execution. For dynamic migration strategies that pass verification, the pre-defined Saga (a design pattern for solving cross-service transaction consistency issues in distributed systems) distributed transaction logic breaks down the dynamic migration strategy into multiple reversible atomic operation sequences. These sequences include checkpoint creation, resource pre-configuration, and tiered traffic switching. Checkpoint creation utilizes CRIU (Checkpoint / Restore In Userspace) technology to achieve sub-second container state snapshots, freezing and preserving all processes within the container to efficiently obtain the complete state of processes. Furthermore, resources are pre-allocated in the new container, such as storage volumes and network policies. Finally, the new container is connected to the communication network to achieve network topology reconstruction.

[0053] Furthermore, the migration execution engine module can adopt a progressive migration mode, gradually transferring container instances from the source node to the target node through batch migration and canary release mechanisms. After each batch migration is completed, health checks and performance comparisons are automatically performed. Only after confirming that there are no abnormalities in business indicators can the next batch migration be promoted, thereby effectively reducing the overall migration risk and improving system stability.

[0054] In this embodiment of the application, each operation node is pre-configured with a timeout circuit breaker mechanism. If the migration time exceeds the preset migration time specified in the timeout circuit breaker mechanism logic, the timeout compensation transaction is automatically triggered, such as automatic rollback.

[0055] Step 3: Health Check. After the dynamic migration strategy takes effect and the container migration is successfully completed, a health monitoring and check process for the new container can be initiated. This process monitors the health status of three layers: the infrastructure layer, the service layer, and the business layer. Specifically, the infrastructure layer monitors the resource utilization of monitoring nodes, the service layer monitors the HTTP success rate of monitoring nodes, and the business layer monitors the latency of critical transactions. If the health check succeeds, the old resources occupied by the original container can be released. If the health check fails, the system can automatically roll back to the state before the strategy migration.

[0056] Step 4: Closed-loop cleanup. After the dynamic migration strategy takes effect and the container migration is successfully completed, a closed-loop cleanup process for the original container can be initiated. Specifically, the original container can be retained for 24 hours as a disaster recovery rollback container. The storage resource access status is automatically analyzed, and the backup retention strategy is optimized. At the same time, the log records of the actual migration process are fed back to the reinforcement learning unit to optimize the model.

[0057] This migration execution engine, through its three design principles of operation atomicity, state observability, and failure recovery, increases the success rate of complex migration tasks from the industry average of 76% to 99.2%, and reduces the average execution time to one-fifth of traditional tools. In particular, its unique "progressive-rollback" hybrid execution model effectively solves the industry problem of the "avalanche effect" in large-scale service migration.

[0058] In this embodiment, the intelligent decision engine module may further include a cross-module collaboration unit, which is used to seamlessly coordinate the intelligent decision engine module, the transfer execution module, the cognitive interaction interface, and the reinforcement learning unit. The built-in logic of the cross-module collaboration unit can be as follows: Figure 7 As shown, an event-driven distributed architecture is used to build an end-to-end autonomous container migration system. As one implementation method, this cross-module collaboration unit specifically performs the following steps: Step 1: Data Acquisition. This can be understood as event triggering and data flow. Data is collected through the environment perception layer in the intelligent decision engine module. Then, the LLM model in the intelligent decision engine module acts as the decision center, performing unified event bus processing and generating structured data packets. These data packets include: environmental state data, pre-generated migration strategies (i.e., multiple feasible solutions generated by the multi-objective optimizer), risk assessment results (including migration success probability, potential impact service list), and other information.

[0059] Step Two: The intelligent decision engine module executes human-machine collaborative decision-making. After receiving the natural language command submitted by the user, the cognitive interaction interface confirms the user's request and determines the corresponding processing scenario, such as an automated scenario or an interactive scenario. In an automated scenario, a low-risk strategy (such as a conventional load balancing strategy) is output as a dynamic migration strategy to the migration execution engine module, which then executes the corresponding dynamic migration strategy. In an interactive scenario, for high-risk migration strategies (such as migrating container traffic across clouds), a confirmation process is initiated with the user to obtain their confirmation command. During this process, natural language can be used to ask the customer why they chose a high-risk migration strategy, such as why this strategy was chosen and the potential for exceeding carbon emission standards. A comparison of multiple feasible solutions from multiple dimensions is provided to allow the user to confirm or adjust the migration strategy. Finally, the user's confirmation or adjustment command is synchronously distributed to the migration execution engine module or the intelligent decision engine module.

[0060] Step 3: The migration execution engine module performs transactional migrations. The migration execution engine operates according to an atomic transaction process: Pre-check phase: Verifies target environment resources; if failure occurs, triggers the decision center to regenerate the strategy. Execution phase: Implements resource allocation, data synchronization, and traffic switching in batches, with real-time status reporting to the interactive interface. Monitoring phase: Continuously collects performance metrics; automatically rolls back and triggers the learning system to analyze the root cause when anomalies occur.

[0061] Step 4: The learning unit performs closed-loop feedback optimization. After receiving the migration end-to-end data, the evolutionary learning unit performs the following: Short-term optimization: generating hotfix patches within 24 hours (e.g., adjusting the API call frequency of a specific cloud provider). Long-term evolution: weekly iterative updates of the algorithm weights in the decision-making center. The optimization results are back-synchronized to each module via the event bus, forming a continuous improvement closed loop.

[0062] The container migration bootstrapping system based on a large model provided in this application generates an optimal migration strategy that balances cost, performance, and compliance by dynamically analyzing container runtime status, cloud environment data, and historical migration records, based on a hybrid decision architecture of Large Language Model (LLM) and reinforcement learning. Furthermore, it employs the Saga distributed transaction model and progressive traffic switching technology to break down the migration process into atomic, reversible operations (checkpoint creation, resource pre-provisioning, tiered traffic switching, etc.). It pioneers a "digital twin pre-verification" mechanism, simulating the runtime environment before actual migration to predict potential conflicts, increasing the success rate of complex migration tasks from the industry average of 76% to 99.2%. Moreover, it builds a domain-specific natural language understanding engine, supporting fully automated migration processes triggered by voice / text commands (such as "migrate AI services to a carbon-neutral region"). It innovatively introduces a "decision tracing" function, generating a technical basis chain (such as carbon emission calculation logic) for each operation, enabling even non-expert users to efficiently manage cross-cloud containers. Furthermore, the decision-making logic is continuously optimized through a transfer effect quantification model (including 12-dimensional evaluation indicators), forming a closed-loop learning system of "execution-evaluation-training". Its core innovation lies in transforming failure cases into reusable pitfall-avoidance rules, reducing the adaptation time in the new environment from 72 hours to 4 hours, and model iteration does not affect online services at all.

[0063] Furthermore, this application designs a unified event bus to connect all modules, achieving sub-second-level state synchronization and anomaly response. The key breakthrough lies in the "cross-module circuit breaker mechanism": when the execution engine detects an anomaly, it can simultaneously trigger the recalculation of the decision-making center strategy, the push of alarms to the interactive interface, and the root cause analysis of the learning system, enabling the system-level self-healing rate to reach 92%, far exceeding the average level of 35% of similar solutions.

[0064] Based on the system provided in the first aspect, in the second aspect, this application provides a container migration bootstrapping method based on a large model, wherein, as... Figure 8 As shown, the method includes: S81. Obtain container environment data, wherein the container environment data includes: container environment monitoring metrics, cloud service provider quotes, and container topology; and generate a dynamic migration strategy based on the container environment data; S82. Based on the dynamic migration strategy, generate a pre-inspection list corresponding to the dynamic migration strategy, and perform resource availability verification, dependent service topology verification, and compliance verification based on the pre-inspection list. S83. If the resource availability verification, the dependent service topology verification, and the compliance verification all pass, then the dynamic migration strategy is executed to migrate the request traffic of the original container to the new container, wherein the original container is a container that meets the container migration conditions.

[0065] In some possible embodiments, the method further includes: By combining reinforcement learning units with large model units, the data of each container environment are quantized into different weight parameters. Based on the weight parameters corresponding to the container environment data, a Pareto optimal dynamic migration strategy set is generated, and natural language explanation information corresponding to each dynamic migration strategy in the Pareto optimal dynamic migration strategy set is output.

[0066] In some possible embodiments, the method further includes: The system acquires the running status data of each container and the cloud environment status data according to a preset cycle. Based on the running status data of each container and the cloud environment status data, a resource call interface is generated based on the dynamic migration strategy; The request traffic from the original container is migrated to the new container based on the resource call interface.

[0067] In some possible embodiments, the method further includes: The reinforcement learning unit, combined with the large model unit, is used to perform time-series analysis on the container environment data; Based on the time series analysis results, a predictive migration strategy is generated, and the reserved containers are migrated to the reserved instances corresponding to the predictive migration strategy based on the predictive migration strategy.

[0068] In some possible embodiments, the method further includes: Obtain natural language commands input by the user, identify the command type corresponding to the natural language commands, and the command type includes: strategy adjustment and status query; In response to the instruction type corresponding to the natural language instruction, a user container migration strategy is generated, and the migration process corresponding to the user container migration strategy is displayed in natural language on the cognitive interaction interface.

[0069] In some possible embodiments, the method further includes: Based on the cloud service provider's quote, calculate the operating cost of the new container after implementing the dynamic migration strategy to migrate the request traffic of the original container to the new container, and output the operating cost.

[0070] In some possible embodiments, the method further includes: After executing the dynamic migration strategy to migrate the request traffic of the original container to the new container, the migration results are obtained and an effect evaluation matrix is ​​generated. Based on the effect evaluation matrix, determine whether the dynamic migration strategy is abnormal. If there is no abnormality, store the container environment data corresponding to the dynamic migration strategy and the dynamic migration strategy as training sample data in the model incremental training sample database. If the dynamic migration strategy is abnormal, an error analysis is performed on the dynamic migration strategy, and a patch strategy is generated and executed.

[0071] In some possible embodiments, the method further includes: The algorithm weight parameters of the large model unit are iteratively updated using the training sample data in the incremental training sample database of the model.

[0072] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in this application comply with relevant laws and regulations and do not violate public order and good morals.

[0073] The names of the messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0074] Thirdly, exemplary embodiments of this application also provide an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the electronic device to perform a method according to an embodiment of this application.

[0075] An exemplary embodiment of this application also provides a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to an embodiment of this application.

[0076] An exemplary embodiment of this application also provides a computer program product, including a computer program, wherein, when executed by a computer's processor, the computer program is used to cause the computer to perform a method according to an embodiment of this application.

[0077] refer to Figure 9 The present invention describes a structural block diagram of an electronic device 900 that can serve as a server or client of this application, which is an example of a hardware device that can be applied to various aspects of this application. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the application described and / or claimed herein.

[0078] like Figure 9 As shown, the electronic device 900 includes a computing unit 901, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM 902) or a computer program loaded from a storage unit 908 into a random access memory (RAM 903). The RAM 903 may also store various programs and data required for the operation of the electronic device 900. The computing unit 901, ROM 902, and RAM 903 are interconnected via a bus 904. An input / output interface (I / O interface 905) is also connected to the bus 904.

[0079] Multiple components in electronic device 900 are connected to I / O interface 905, including: input unit 906, output unit 907, storage unit 908, and communication unit 909. Input unit 906 can be any type of device capable of inputting information to electronic device 900. Input unit 906 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of electronic device. Output unit 907 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 908 may include, but is not limited to, disk and optical disk. Communication unit 909 allows electronic device 900 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers and / or chipsets, such as Bluetooth™ devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0080] The computing unit 901 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above. For example, in some embodiments, the aforementioned large-model-based container migration bootstrapping method can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 908. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 900 via ROM 902 and / or communication unit 909. In some embodiments, the computing unit 901 can be configured to perform the aforementioned large-model-based container migration bootstrapping method by any other suitable means (e.g., by means of firmware).

[0081] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0082] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0083] As used in this application, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0084] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0085] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0086] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

Claims

1. A large model-based container migration guide system, characterized by, The system comprises: An intelligent decision engine module, which is pre-installed with a large model unit, is used to acquire container environment data and generate a dynamic migration strategy based on the container environment data through the large model unit; wherein the container environment data comprises container environment monitoring indicators, cloud service operator quotes, and container topology; A migration execution engine module is used to acquire the dynamic migration strategy, generate a pre-inspection list corresponding to the dynamic migration strategy, and perform resource availability verification, dependent service topology verification, and compliance verification based on the pre-inspection list. If the resource availability verification, the dependent service topology verification, and the compliance verification all pass, the dynamic migration strategy is executed to migrate the request traffic of an original container to a new container, wherein the original container is a container that meets the container migration conditions.

2. The system of claim 1, wherein, The intelligent decision engine module is also pre-installed with a reinforcement learning unit, and the intelligent decision engine module is also used to: Quantify each of the container environment data into different weight parameters through the reinforcement learning unit in cooperation with the large model unit; Generate a Pareto optimal dynamic migration strategy set based on the weight parameters corresponding to each of the container environment data, and output natural language explanation information corresponding to each dynamic migration strategy in the Pareto optimal dynamic migration strategy set.

3. The system of claim 1, wherein, The migration execution engine module is also used to: Acquire running state data of each container and cloud environment state data according to a preset period; Generate a resource calling interface based on the dynamic migration strategy according to the running state data of each container and the cloud environment state data; Migrate the request traffic of the original container to the new container based on the resource calling interface.

4. The system of claim 2, wherein, The intelligent decision engine module is also used to: Perform time series analysis on the container environment data by using the reinforcement learning unit in combination with the large model unit; Generate a predicted migration strategy based on the time series analysis result, and migrate a reserved container to a reserved instance corresponding to the predicted migration strategy based on the predicted migration strategy.

5. The system of claim 1, wherein, The system further comprises: A cognitive interaction interface is used to acquire a natural language instruction input by a user, identify an instruction type corresponding to the natural language instruction, and the instruction type comprises a strategy adjustment and a state query; The intelligent decision engine module is also used to generate a user container migration strategy in response to the instruction type corresponding to the natural language instruction, and display a migration process corresponding to the user container migration strategy in the cognitive interaction interface in natural language.

6. The system of claim 1, wherein, The intelligent decision engine module is also used to: Calculate the running cost of the new container after executing the dynamic migration strategy to migrate the request traffic of the original container to the new container based on the cloud service operator quotes, and output the running cost.

7. The system of claim 1, wherein, The intelligent decision engine module is also used to: After executing the dynamic migration strategy to migrate the request traffic of the original container to the new container, acquire a migration result and generate an effect evaluation matrix. determining whether the dynamic migration strategy is abnormal based on the effect evaluation matrix, if not, storing container environment data corresponding to the dynamic migration strategy and the dynamic migration strategy as training sample data into a model incremental training sample database; if the dynamic migration strategy is abnormal, performing error cause analysis on the dynamic migration strategy, generating and executing a patch strategy.

8. The system of claim 7, wherein, The intelligent decision engine module is further configured to: update algorithm weight parameters of the large model unit iteratively by using training sample data in the model incremental training sample database. 9.A method for guiding container migration based on a large model, characterized in that, The method is applied to the large model-based container migration guidance system according to any one of claims 1-8, and the method comprises: obtaining container environment data and generating a dynamic migration strategy based on the container environment data by using the large model unit; wherein the container environment data comprises container environment monitoring indicators, cloud service operator quotes, and container topology; generating a pre-inspection list corresponding to the dynamic migration strategy based on the dynamic migration strategy, and performing resource availability verification, dependent service topology verification, and compliance verification based on the pre-inspection list; if the resource availability verification, the dependent service topology verification, and the compliance verification are all passed, executing the dynamic migration strategy to migrate request traffic of an original container to a new container, wherein the original container is a container that meets the container migration condition.

10. An electronic device, comprising: The electronic device comprises a processor and a memory storing a program; wherein the program comprises instructions which, when executed by the processor, cause the processor to perform the method according to claim 9.

Images