Distributed cloud resource pool cooperative control method for hybrid cryptographic operation service
By extracting tenant requests and target container information from the cloud cryptography service management platform, and establishing detection and anomaly pattern matching models, the problem of low efficiency in resource management and performance optimization in existing technologies is solved, and efficient and stable operation of cryptography services in a distributed cloud environment is achieved.
Patent Information
- Application Number
- CN202510363004.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-26
- Publication Date
- 2026-05-15
- Estimated Expiration
- 2045-03-26
AI Technical Summary
Existing cryptographic service management platforms lack effective resource management and monitoring mechanisms in distributed cloud environments, making it impossible to detect resource usage and performance in real time and accurately. This results in low efficiency in resource management and performance optimization, and existing technologies cannot fully cover the entire distributed cloud environment, making it difficult to identify complex and interconnected anomaly patterns.
By acquiring tenant cryptographic service requests and target container information from the cloud cryptographic service management platform, feature extraction and matching are performed, key performance parameters are obtained in real time, a cryptographic service processing status detection model is established, abnormal target containers are identified, and an abnormal pattern matching model is constructed through performance parameter change curves. Corresponding solutions, such as migration and creating new container instances, are designed to ensure the stability and continuity of services.
It significantly improves data processing efficiency, can quickly identify and optimize abnormal target containers, ensures the stable operation of hybrid cryptographic computing services in the distributed cloud resource pool, and improves the system's stability and performance optimization capabilities.
Smart Images
Figure CN120151055B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of information processing and management technology, and more specifically to a method for collaborative control of distributed cloud resource pools for hybrid cryptographic operations services. Background Technology
[0002] In cloud computing and distributed computing environments, high availability and performance optimization of cryptographic services have become critical challenges. Firstly, with the widespread adoption of cloud computing, numerous enterprises and organizations are migrating their mission-critical businesses to the cloud, necessitating cloud platforms to provide highly secure and high-performance cryptographic services. However, most existing cryptographic service management platforms lack effective distributed resource management and monitoring mechanisms, failing to accurately detect resource usage and cryptographic service performance in the distributed environment in real time. Existing monitoring methods typically rely on single resource monitoring metrics or manual intervention, failing to capture complex anomaly patterns, resulting in low efficiency in resource management and performance optimization.
[0003] Secondly, while some cryptographic service optimization techniques have been developed to address the aforementioned issues, they typically focus on single resource nodes or containers, failing to comprehensively cover the entire distributed cloud environment. For example, some techniques focus on detecting anomalies in individual containers and providing simple resource adjustment solutions, but neglect the interrelationships between containers and the overall system performance. This lack of a global perspective leads to inaccurate analysis of cryptographic service processing status, making it difficult to identify complex and interconnected anomaly patterns. Therefore, there is an urgent need for a method that can dynamically monitor and manage the status of various resources and services in a distributed environment, especially for the comprehensive analysis and optimization of cryptographic computing services. Summary of the Invention
[0004] The purpose of this invention is to solve the problems in the background art by proposing a distributed cloud resource pool collaborative control method for hybrid cryptographic operation services.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: a distributed cloud resource pool collaborative control method for hybrid cryptographic operation services, comprising the following steps:
[0006] Step 1: Obtain the tenant's password service request and target container information from the cloud password service management platform, extract features from the password service request and target container information to obtain the feature information of the password service request, and match the corresponding target container based on the feature information of the password service request;
[0007] Step 2: Based on the matching results, obtain the key performance parameters of each target container in real time, and calculate the comprehensive evaluation coefficient according to each key performance parameter to establish a cryptographic service processing status detection model, analyze the cryptographic service processing status of each target container, and identify abnormal target containers.
[0008] Step 3: Statistically analyze all abnormal target containers and record their performance parameters under abnormal conditions. Based on the statistical performance parameters, obtain the performance parameter change curve of the abnormal target containers. Then, combine the performance parameter change curve to establish a performance parameter and abnormal pattern matching model. Based on the abnormal pattern matching model, obtain the abnormal pattern of each abnormal target container.
[0009] Step 4: Design different solutions based on the anomaly pattern recognition results of each abnormal target container;
[0010] As a further aspect of the present invention, the tenant password service request includes, but is not limited to, key management service request, data encryption service request, educational identity authentication service request, Internet of Things security authentication service request, signature verification service request, timestamp service request, etc.
[0011] As a further aspect of this invention: the specific process of establishing a cryptographic service processing status detection model based on the comprehensive evaluation coefficient is as follows: the key performance parameters of each target container are standardized to form the key performance parameter sequence of the i-th target container. ;in Let j represent the i-th target container, and j represent the number of key performance parameters;
[0012] Based on the sequence of key performance parameters of the target container, a corresponding weight is assigned to each performance parameter. Then, the weighted key performance parameters are summed to obtain the comprehensive evaluation coefficient of the i-th target container. The formula is as follows: ;in These represent the weights of the corresponding key performance parameters for the target container;
[0013] Based on the comprehensive evaluation coefficient of each target container The constructed cryptographic service processing status detection model is as follows: ;in This is the preset threshold corresponding to the target container.
[0014] As a further aspect of the present invention: the cryptographic service processing status of each target container is analyzed according to the cryptographic service processing status detection model to obtain abnormal target containers.
[0015] As a further aspect of the present invention: the process of obtaining the performance parameter change curve of the abnormal target container is as follows: when the cryptographic service processing status detection model detects that the container is marked as abnormal, all performance parameters within its time window are recorded, and an abnormal time series dataset is formed according to the time order; and multiple performance parameter change curves are formed according to the abnormal time series dataset corresponding to each abnormal target container, and feature extraction is performed on the performance parameter change curves corresponding to each abnormal target container, including trend features, periodic features, fluctuation features, and mutation features; thus forming a performance parameter change feature parameter set for each abnormal target container.
[0016] As a further aspect of the present invention: the process of constructing a matching model based on the abnormal pattern Y is as follows:
[0017] S01: Based on the time-series change characteristic parameters corresponding to the occurrence of abnormal mode Y, establish a matching criterion between abnormal mode Y and the corresponding time-series change characteristic parameter sequence. Abnormal mode Y includes overload abnormality, crash abnormality and mixed abnormality.
[0018] S02: Receive and accumulate the changing feature parameters at different times t according to the time series, extract the changing feature parameters output by the model, obtain M time series sequences of changing feature parameters, fit each time series sequence of changing feature parameters, compare the fitting results with the corresponding matching criteria, and obtain the matching probability between each time series sequence of changing feature parameters and the corresponding abnormal pattern; specifically, calculate the similarity between each feature and the typical features of the abnormal pattern and score it, then calculate the matching probability by combining the scores of each feature, and determine the abnormal pattern recognition result based on the matching probability; that is, when the matching probabilities of overload abnormality and collapse abnormality are P1 and P2 respectively, if P1>P2, it is judged as overload abnormality; otherwise, it is judged as collapse abnormality; if P1=P2, it is judged as mixed abnormality;
[0019] S03: After model training, the changing feature parameters have included the temporal features of multidimensional performance parameters, integrated the key information of performance parameters, and matched abnormal patterns through the temporal change trend of feature parameters.
[0020] As a further aspect of the present invention, different solutions include: when the classification model determines that the abnormal target container is overloaded, it automatically triggers migration, that is, the cryptographic service in the abnormal target container is migrated to a target container with the same service type that is running normally;
[0021] When the classification model determines that the abnormal target container is a crash exception, it automatically triggers the creation process, that is, quickly creates a new container instance to take over the tasks of the crash target container, ensuring service continuity;
[0022] When the classification model determines that the abnormal target container is a hybrid anomaly, it prioritizes creating a new instance to take over the crashed task, while migrating the overloaded task to a target container that is running normally with the same service type.
[0023] Compared with existing technologies, the advantages of this invention are:
[0024] 1. This invention extracts tenant cryptographic service requests and target container information from a cloud cryptographic service management platform, and performs feature extraction and matching. This enables rapid location of target containers, significantly improving data processing efficiency. Furthermore, it obtains key performance parameters of each target container in real time, calculates comprehensive evaluation coefficients, and establishes a cryptographic service processing status detection model. This model comprehensively and accurately analyzes the cryptographic service processing status of each target container, providing reliable data support for anomaly identification. The cryptographic service processing status detection model also enables rapid identification of anomaly-prone target containers, laying the foundation for subsequent anomaly pattern analysis and solution design.
[0025] 2. This invention statistically records the performance parameters of all abnormal target containers, plots performance parameter change curves, and establishes a performance parameter-abnormal pattern matching model. This enables precise identification of specific abnormal patterns of abnormal target containers, providing a scientific basis for solution design. Based on the abnormal pattern identification results, different solutions can be designed and implemented to specifically optimize and restore system performance, significantly improving system stability. Furthermore, by dynamically adjusting and optimizing solutions, the performance of the cloud resource pool can be continuously improved, ensuring the stable operation of hybrid cryptographic computing services in the distributed cloud resource pool. Attached Figure Description
[0026] Figure 1 The overall flowchart of the distributed cloud resource pool collaborative control method for hybrid cryptographic operation service proposed in this invention;
[0027] Figure 2 This is a flowchart illustrating the application of the distributed cloud resource pool collaborative control method for hybrid cryptographic operations proposed in this invention in a cloud cryptographic service management platform. Detailed Implementation
[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0029] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.
[0030] Reference Figures 1-2 A method for collaborative control of a distributed cloud resource pool for hybrid cryptographic operations services includes the following steps:
[0031] Step 1: Obtain the tenant's password service request and target container information from the cloud password service management platform, extract features from the password service request and target container information to obtain the feature information of the password service request, and match the corresponding target container based on the feature information of the password service request;
[0032] In this example, tenant cryptographic service requests include, but are not limited to, key management service requests, data encryption service requests, educational identity authentication service requests, IoT security authentication service requests, signature verification service requests, timestamp service requests, etc.
[0033] The characteristic information of the cryptographic service request includes, but is not limited to, cryptographic service type (encryption, decryption, signature, signature verification, etc.), data type, data volume, security level requirements, tenant-related information, etc.
[0034] The target container is a containerized deployment of various cryptographic services in the cloud cryptographic resource pool of the cloud cryptographic service management platform. Specifically, the target container information includes, but is not limited to, detailed information such as the functional characteristics, service scope, resource configuration, and security level of each target container.
[0035] In this embodiment, the cloud password resource pool is defined as V={ ...., }, i={ ...., } represents multiple target containers. = ( ...., This indicates that a target container contains multiple cryptographic service types.
[0036] Matching the corresponding target container based on the feature information of the cryptographic service request specifically involves: based on the extracted cryptographic service request features and target container information, the platform uses a specific matching algorithm or strategy to find the most suitable target container; a common strategy is to prioritize matching based on the type of cryptographic service, and then consider other factors, such as security level and resource availability; for example, for an encryption service request, the platform will first filter out target containers with encryption functions, and then select containers from these containers that meet the security level requirements and have low current resource load to process the request;
[0037] It should be noted that the above feature extraction and matching algorithms are existing mature technologies, and will not be elaborated on in this embodiment.
[0038] Step 2: Based on the matching results, obtain the key performance parameters of each target container in real time, and calculate the comprehensive evaluation coefficient according to each key performance parameter to establish a cryptographic service processing status detection model, analyze the cryptographic service processing status of each target container, and identify abnormal target containers.
[0039] Based on the matching results, the performance parameters of each target container are obtained in real time, specifically as follows:
[0040] Obtain key performance parameters for each target container during the cryptographic service request processing process. These key performance parameters include, but are not limited to, CPU utilization. Cryptographic operation response time Storage capacity utilization wait;
[0041] The logic for obtaining the cryptographic operation response time is to obtain the first timestamp of receiving the cryptographic service request, and to monitor and statistically analyze the entire process of processing the cryptographic service request to obtain the second timestamp corresponding to the return result of the cryptographic service request, and to use the difference between the second timestamp and the first timestamp as the cryptographic operation response time.
[0042] The CPU utilization rate reflects the computational resource usage of tasks such as cryptographic operations within the target container; the storage capacity utilization rate reflects the degree of utilization of storage resources within the container; both of these performance parameters can be directly obtained through the system.
[0043] The key performance parameters of each target container are standardized to form the key performance parameter sequence of the i-th target container. ;in Let j represent the i-th target container, and j represent the number of key performance parameters;
[0044] Based on the sequence of key performance parameters of the target container, a corresponding weight is assigned to each performance parameter. Then, the weighted key performance parameters are summed to obtain the comprehensive evaluation coefficient of the i-th target container. The formula is as follows: ;in These represent the weights of the corresponding key performance parameters of the target container, and their specific values are dynamically set by researchers based on the importance of each performance parameter to the processing status of the target container.
[0045] Based on the comprehensive evaluation coefficient of each target container The constructed cryptographic service processing status detection model is as follows: ;in The specific value of the preset threshold corresponding to the target container is determined by researchers based on experience, and different thresholds are set for each container.
[0046] when The time indicates that the i-th target container is in a normal processing state and no further analysis is needed;
[0047] when The state of the i-th target container is abnormal, and it is marked as an abnormal target container. In order to ensure that the tenant's business system can continue to obtain password services, the abnormal target container needs to be further processed.
[0048] Step 3: Collect statistics on all abnormal target containers and record their performance parameters under abnormal states. Based on the collected performance parameters, obtain the performance parameter change curves for the abnormal target containers. Then, combine these performance parameter change curves to establish a performance parameter and abnormal pattern matching model. Based on the abnormal pattern matching model, the abnormal patterns of each abnormal target container are obtained, including:
[0049] In this embodiment, before establishing the correlation between performance parameters and failure modes, a large number of historical abnormal target container anomaly cases are collected, and the anomaly modes are defined as specifically including overload anomalies, crash anomalies, and mixed anomalies. For each anomaly mode, the corresponding performance parameter change characteristics are summarized and marked as typical anomaly mode characteristics. In this embodiment, the definition principle of each anomaly mode can be based on the change range of each performance parameter. For example, the typical characteristics of overload anomalies show a continuous upward trend in CPU utilization, and remain at a relatively high level for a long period, such as consistently in the 70%-90% range. The typical characteristics of crash anomalies show a sudden spike in CPU utilization to nearly 100%, followed by a rapid decline. Mixed anomalies fall between the two.
[0050] The logic for obtaining the performance parameter change trend of abnormal target containers is as follows: When the cryptographic service processing status detection model detects that a container is marked as abnormal, all performance parameters within its time window are recorded and an abnormal time series dataset is formed according to the time order; and based on the abnormal time series dataset corresponding to each abnormal target container, multiple performance parameter change curves are formed, and feature extraction is performed on the performance parameter change curves corresponding to each abnormal target container, including trend features, periodic features, fluctuation features, and mutation features; thus forming a performance parameter change feature parameter set for each abnormal target container.
[0051] It should be noted that: trend characteristics indicate the overall trend of the analysis curve, such as rising, falling, or remaining stable; periodic characteristics indicate whether there are periodic changes in the curve; fluctuation characteristics indicate the amplitude and frequency of fluctuations in the curve; and statistical measures such as standard deviation and variance are calculated to measure the degree of fluctuation; abrupt change characteristics indicate the search for abrupt change points on the curve, that is, the locations where performance parameters suddenly change significantly.
[0052] The process of constructing a matching model based on the abnormal pattern Y is as follows:
[0053] S01: Based on the time-series change characteristic parameters corresponding to the occurrence of abnormal mode Y, establish a matching criterion between abnormal mode Y and the corresponding time-series change characteristic parameter sequence. Abnormal mode Y includes overload abnormality, crash abnormality and mixed abnormality.
[0054] S02: Receive and accumulate the changing feature parameters at different times t according to the time series, extract the changing feature parameters output by the model, obtain M time series sequences of changing feature parameters, fit each time series sequence of changing feature parameters, compare the fitting results with the corresponding matching criteria, obtain the matching probability between each time series sequence of changing feature parameters and the corresponding abnormal pattern, and determine the abnormal pattern recognition result based on the matching probability; specifically, calculate the similarity between each feature and the typical features of the abnormal pattern and score it, then calculate the matching probability by combining the scores of each feature; for example, the matching probabilities of overload abnormality and collapse abnormality are P1 and P2 respectively. If P1>P2, it is judged as overload abnormality; otherwise, it is judged as collapse abnormality; if P1=P2, it is judged as mixed abnormality.
[0055] S03: After model training, the changing feature parameters have included the temporal features of multidimensional performance parameters, and have integrated the key information of performance parameters. Abnormal patterns can be matched by the temporal change trend of the feature parameters.
[0056] The matching model learns the relationship between performance parameter changes and abnormal patterns, and can automatically learn the abnormal patterns of abnormal target containers and their corresponding performance parameter change characteristics. It takes the characteristics of the performance parameter change curve as input and the abnormal pattern as output, thus realizing the judgment of the abnormal pattern of the target container.
[0057] It should be noted that the above-mentioned model building process uses existing mature technologies, and will not be elaborated on in detail here.
[0058] Step 4: Based on the anomaly pattern recognition results of each abnormal target container, design different solutions. This includes: obtaining the anomaly pattern recognition results corresponding to each abnormal target container, and designing different solutions based on different anomaly pattern sets. Specifically, when the classification model determines that an abnormal target container is overloaded, automatically trigger migration, i.e., migrate the cryptographic services in the abnormal target container to a normally operating target container with the same service type. Specifically, when the classification model determines that an abnormal target container is overloaded, it means that the amount of cryptographic service tasks currently carried by the container exceeds its reasonable processing capacity, causing various performance parameters to show corresponding overload characteristics, but the container itself is still running and has not yet crashed. Intelligently select target containers with the same service type that are currently operating normally, and then migrate the cryptographic services in the overloaded abnormal target containers to these normally operating target containers in an orderly manner according to preset rules and priority order. During the migration process, fully consider the current load, resource availability, and network connectivity of each normal container to ensure that the migrated cryptographic services can be reasonably allocated and efficiently processed in the new container environment, avoiding new overload situations.
[0059] When the classification model determines that an abnormal target container has crashed, it automatically triggers a creation process to quickly create a new container instance to take over the tasks of the crashed container, ensuring service continuity. For example, this indicates that the container has completely crashed due to some serious problems, such as hardware failure, software vulnerability, or resource exhaustion, and is no longer able to maintain the normal operation of the password service, thus losing all ability to respond to external requests. Once a container crash is detected, the system will immediately start the mechanism to quickly create a new container instance. Using a pre-configured container template and related parameters, a brand new container instance with the corresponding functions will be created in a very short time. This newly created container instance will quickly take over the password service tasks originally performed by the crashed container, ensuring that service continuity is not affected.
[0060] When the classification model determines that the abnormal target container is a hybrid anomaly, it prioritizes creating a new instance to take over the crashed tasks, while migrating the overloaded tasks to a neighboring container. This means that within the same container, there is both an overload problem caused by excessive load and a potential crash that causes some functions of the container to malfunction. This situation poses a more severe challenge to the stability and continuity of the cryptographic service. The priority measure is to quickly create a new container instance and let the newly created container instance take over the tasks involved in the crash first, ensuring that the services interrupted by the crash can be restored to normal operation as soon as possible.
[0061] This embodiment can monitor and diagnose the cryptographic service processing status of each target container in the cloud resource pool in real time, thereby identifying and classifying abnormal target containers, and finally designing corresponding solutions. Its working principle is as follows: First, it obtains the tenant's cryptographic service requests and related information from the cloud cryptographic service platform. Through feature extraction technology, it accurately matches the corresponding target containers, thus beginning in-depth analysis of the cryptographic services of these target containers. The system monitors the key performance parameters of these containers in real time, calculates a comprehensive evaluation coefficient to assess the cryptographic service processing status, and thus quickly identifies containers with abnormalities. Furthermore, by statistically analyzing the changes in the performance parameters of these abnormal containers, it can construct a matching model between performance parameters and abnormal patterns to determine the abnormal patterns of the containers. Based on these abnormal pattern identification results, effective solutions can be tailored for each abnormal container, thereby ensuring that the cryptographic computing service in the entire cloud resource pool can operate efficiently and stably, providing tenants with high-quality cryptographic service support.
[0062] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A distributed cloud resource pool collaborative control method for hybrid cryptographic operations services, characterized in that, include: Obtain tenant cryptographic service requests and target container information from the cloud cryptographic service management platform, extract features from the cryptographic service requests and target container information to obtain feature information of the cryptographic service requests, and match the corresponding target container based on the feature information of the cryptographic service requests; Based on the matching results, the key performance parameters of each target container are obtained in real time, and a comprehensive evaluation coefficient is calculated based on each key performance parameter. In this way, a cryptographic service processing status detection model is established to analyze the cryptographic service processing status of each target container and identify abnormal target containers. All abnormal target containers are statistically analyzed and their performance parameters under abnormal states are recorded. Based on the statistical performance parameters, the performance parameter change curves of the abnormal target containers are obtained. Then, a performance parameter and abnormal pattern matching model is established in combination with the performance parameter change curves. Based on the abnormal pattern matching model, the abnormal pattern of each abnormal target container is obtained. Different solutions are designed based on the abnormal pattern recognition results of each abnormal target container.
2. The distributed cloud resource pool collaborative control method for hybrid cryptographic operation services according to claim 1, characterized in that; The tenant cryptographic service request includes at least one of the following: key management service request, data encryption service request, educational identity authentication service request, Internet of Things security authentication service request, signature verification service request, and timestamp service request.
3. The distributed cloud resource pool collaborative control method for hybrid cryptographic operation services according to claim 1, characterized in that: Based on the comprehensive evaluation coefficient, the specific process of establishing the cryptographic service processing status detection model is as follows: The key performance parameters of each target container are standardized to form the key performance parameter sequence of the i-th target container. ;in Let j represent the i-th target container, and j represent the number of key performance parameters; Based on the sequence of key performance parameters of the target container, a corresponding weight is assigned to each performance parameter. Then, the weighted key performance parameters are summed to obtain the comprehensive evaluation coefficient of the i-th target container. The formula is as follows: ;in These represent the weights of the corresponding key performance parameters for the target container; Based on the comprehensive evaluation coefficient of each target container The constructed cryptographic service processing status detection model is as follows: ;in This is the preset threshold corresponding to the target container.
4. The distributed cloud resource pool collaborative control method for hybrid cryptographic operation services according to claim 3, characterized in that: The cryptographic service processing status of each target container is analyzed based on the cryptographic service processing status detection model to identify abnormal target containers.
5. The distributed cloud resource pool collaborative control method for hybrid cryptographic operation services according to claim 1, characterized in that: The process of obtaining the performance parameter change curves of abnormal target containers is as follows: when the cryptographic service processing status detection model detects that a container is marked as abnormal, all performance parameters within its time window are recorded and an abnormal time series dataset is formed according to the time order; and multiple performance parameter change curves are formed according to the abnormal time series datasets corresponding to each abnormal target container, and feature extraction is performed on the performance parameter change curves corresponding to each abnormal target container, including trend features, periodic features, fluctuation features, and mutation features. A set of characteristic parameters for the performance parameter changes of each abnormal target container is formed.
6. The distributed cloud resource pool collaborative control method for hybrid cryptographic operation services according to claim 1, characterized in that: The process of constructing a matching model based on the abnormal pattern Y is as follows: S01: Based on the time-series change characteristic parameters corresponding to the occurrence of abnormal mode Y, establish a matching criterion between abnormal mode Y and the corresponding time-series change characteristic parameter sequence. Abnormal mode Y includes overload abnormality, crash abnormality and mixed abnormality. S02: Receive and accumulate the changing feature parameters at different times t according to the time sequence, extract the changing feature parameters output by the model, obtain M time series of changing feature parameters, fit each time series of changing feature parameters, compare the fitting results with the corresponding matching criteria, obtain the matching probability between each time series of changing feature parameters and the corresponding abnormal pattern, and determine the abnormal pattern recognition result based on the matching probability; specifically, calculate the similarity between each feature and the typical features of the abnormal pattern and score it, then calculate the matching probability by combining the scores of each feature; that is, when the matching probabilities of overload abnormality and collapse abnormality are P1 and P2 respectively, if P1>P2, it is judged as overload abnormality; otherwise, it is judged as collapse abnormality; if P1=P2, it is judged as mixed abnormality; S03: After model training, the changing feature parameters have included the temporal features of multidimensional performance parameters, integrated the key information of performance parameters, and matched abnormal patterns through the temporal change trend of feature parameters.
7. The distributed cloud resource pool collaborative control method for hybrid cryptographic operation services according to claim 1, characterized in that: Different solutions include: automatically triggering migration when the classification model determines that the abnormal target container is overloaded, that is, migrating the cryptographic service in the abnormal target container to a target container with the same service type that is running normally; When the classification model determines that the abnormal target container is a crash exception, it automatically triggers the creation process, that is, quickly creates a new container instance to take over the tasks of the crash target container, ensuring service continuity; When the classification model determines that the abnormal target container is a hybrid anomaly, it prioritizes creating a new instance to take over the crashed task, while migrating the overloaded task to a target container that is running normally with the same service type.