Cloud environment failure processing method, device and computer readable storage medium

Abstract

Embodiments of the present application disclose a cloud environment fault processing method and device and a computer readable storage medium. Embodiments of the present application can perform probability analysis on current fault alarm information, determine a target fault category corresponding to the current fault alarm information, and after the target fault category is determined, corresponding service servers can be quickly pulled to perform problem repair. The scheme can improve the classification efficiency and classification accuracy of the fault type corresponding to the alarm information, thereby quickly positioning the fault type and improving the response rate to the alarm information.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a cloud environment fault handling method, apparatus, and computer-readable storage medium. Background Technology

[0002] Cloud Dedicated Cluster (CDC), as a new type of infrastructure-level distributed cloud product, extends the centralized public cloud environment into a near-field service that can be deployed in the user's data center. It combines the advantages of both public cloud and local internet data center (IDC), allowing users to utilize the rich capabilities of the public cloud with localized latency and data security. CDC deploys localized computing power in the user's data center using an integrated rack, with resources connected to nearby public cloud regions for unified management. Users can manage CDC resources through existing public cloud tools (console, application programming interfaces, etc.).

[0003] The Control Center (CDC) can reuse public cloud capabilities. Although deployed in a customer's data center, the CDC can connect to the public cloud's control service via a management channel. The public cloud can then control the CDC to perform various operations by issuing signaling commands, and the CDC can also interact with the public cloud control center by uploading signaling commands. Because the CDC introduces an additional management channel, its instability increases, frequently leading to various alarm issues. For example, when the management channel is disconnected, the CDC may be unable to send and receive control signaling commands, preventing customers from using all of the CDC's capabilities. Therefore, it is crucial to analyze the fault types corresponding to various alarm messages promptly and accurately, and then notify the relevant business teams for remediation.

[0004] However, in traditional fault handling methods, fault alarm information is relatively independent and opaque. When an alarm occurs, the fault type cannot be determined in time, so the business team corresponding to the fault type cannot be notified in time to fix it. Instead, all business teams have to be mobilized to respond together, which results in long problem localization time, many invalid responses, and high manpower consumption, which greatly affects the availability of business services. Summary of the Invention

[0005] This application provides a cloud environment fault handling method, apparatus, and computer-readable storage medium. It can improve the efficiency and accuracy of fault type classification corresponding to the current fault alarm information, thereby quickly locating the fault type and increasing the response rate to the current fault alarm information.

[0006] A cloud environment fault handling method includes:

[0007] Obtain the current fault alarm information of the current cloud environment, wherein the current fault alarm information includes at least one operating status parameter;

[0008] Obtain a pre-stored set of fault alarm information, wherein the set of fault alarm information includes one or more historical fault alarm information prior to the current cloud service;

[0009] Obtain the conditional probability set and prior probability set corresponding to the historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each of the operating state parameters, and the prior probability set includes the prior probability corresponding to each preset fault category.

[0010] Based on the category of the running state parameter, the conditional probability corresponding to the running state parameter is selected from the set of conditional probabilities;

[0011] Based on the conditional probability and the prior probability, determine the posterior probability that the current fault alarm information belongs to each of the preset fault categories;

[0012] Based on the posterior probability, the target fault category corresponding to the current fault alarm information is selected from the preset fault categories, and the current fault alarm information is sent to the service server corresponding to the target fault category.

[0013] Accordingly, embodiments of this application provide a cloud environment fault handling device, including:

[0014] The first acquisition unit is used to acquire the current fault alarm information of the current cloud environment, wherein the current fault alarm information includes at least one operating status parameter;

[0015] The second acquisition unit is used to acquire a pre-stored set of fault alarm information, wherein the set of fault alarm information includes one or more historical fault alarm information prior to the current cloud service;

[0016] The third acquisition unit is used to acquire the conditional probability set and the prior probability set corresponding to the historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each of the operating state parameters, and the prior probability set includes the prior probability corresponding to each preset fault category.

[0017] A filtering unit is used to filter out the conditional probabilities corresponding to the running state parameters from the set of conditional probabilities based on the category of the running state parameters.

[0018] The determining unit is configured to determine the posterior probability that the current fault alarm information belongs to each of the preset fault categories based on the conditional probability and the prior probability.

[0019] The sending unit is configured to, based on the posterior probability, filter out the target fault category corresponding to the current fault alarm information from the preset fault categories, and send the current fault alarm information to the service server corresponding to the target fault category.

[0020] Furthermore, this application also provides an electronic device, including a processor and a memory, wherein the memory stores an application program, and the processor is used to run the application program in the memory to execute the cloud environment fault handling method provided in this application.

[0021] Furthermore, embodiments of this application also provide a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute the steps in the cloud environment fault handling method provided in embodiments of this application.

[0022] Furthermore, this application also provides a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps in the cloud environment fault handling method provided in this application.

[0023] This application embodiment acquires current fault alarm information from the current cloud environment, and obtains a pre-stored set of fault alarm information, as well as a set of conditional probabilities and a set of prior probabilities corresponding to historical fault alarm information. The set of conditional probabilities includes the conditional probability corresponding to the category of each operating state parameter, and the set of prior probabilities includes the prior probability corresponding to each preset fault category. In the above process, the acquired set of conditional probabilities and the set of prior probabilities are used as a foundation so that the conditional probabilities corresponding to the operating state parameters can be filtered from the set of conditional probabilities based on the category of the operating state parameters. After determining the conditional probabilities corresponding to the operating state parameters, probabilistic analysis can be further performed on the current fault alarm information based on the conditional probabilities and prior probabilities. The process involves determining the posterior probability of the current fault alarm information belonging to each preset fault category; then, based on the posterior probability, filtering out the target fault category corresponding to the current fault alarm information from the preset fault categories. This improves the classification efficiency and accuracy of the fault type corresponding to the current fault alarm information, enabling rapid fault type location. After determining the target fault category, the current fault alarm information can be sent to the business server corresponding to the target fault category, allowing the corresponding business server to be quickly and accurately activated for problem repair without having to activate all business servers for joint response and troubleshooting one by one. This reduces invalid responses, avoids excessive resource consumption, improves the response and processing efficiency for the current fault alarm information, and enhances the availability of business services. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram illustrating the deployment of a distributed cloud environment provided in an embodiment of this application;

[0026] Figure 2 This is a schematic diagram of the traditional fault handling process;

[0027] Figure 3 This is a schematic diagram illustrating an application scenario of the cloud environment fault handling method provided in the embodiments of this application;

[0028] Figure 4 This is a flowchart illustrating the cloud environment fault handling method provided in the embodiments of this application;

[0029] Figure 5 This is a schematic diagram of a cloud environment fault handling process that starts with historical fault alarm information, provided in an embodiment of this application.

[0030] Figure 6 This is a schematic diagram of a pre-stored set of fault alarm information provided in an embodiment of this application;

[0031] Figure 7 This is a schematic diagram of the cloud environment fault handling process provided in the embodiments of this application;

[0032] Figure 8 This is another flowchart illustrating the cloud environment fault handling method provided in the embodiments of this application;

[0033] Figure 9 This is a schematic diagram of the cloud environment fault handling device provided in the embodiments of this application;

[0034] Figure 10 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0035] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] This application provides a cloud environment fault handling method, apparatus, and computer-readable storage medium. The cloud environment fault handling apparatus can be integrated into an electronic device, which can be a server corresponding to a cloud resource provider. For ease of distinction, the server corresponding to the cloud resource provider can be referred to as a cloud server.

[0037] A cloud server can be a server that provides basic cloud computing services such as cloud environment, cloud database, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, network acceleration services (Content Delivery Network, CDN), and big data and artificial intelligence platforms. A cloud server can be a standalone physical server or a server cluster or distributed system composed of multiple physical servers.

[0038] Public cloud resources can be divided into multiple regions based on geographical location. A region refers to a physical data center. A region is an independent physical data center, and the internal networks of different regions are not interconnected. Figure 1 A schematic diagram illustrating the deployment of a distributed cloud environment provided in an embodiment of this application is shown. Figure 1 As shown, any region in a public cloud includes the cloud servers corresponding to that region. This region can have multiple data centers, each of which is an available zone (AZ), such as Available Zone 1, Available Zone 2, ..., Available Zone n (n is a natural number greater than 2). An available zone can only belong to one region, and each available zone is independent of others, for example, having its own independent network and power supply system. Although each available zone has its own independent network (at the high availability layer), at the network layer, available zones within each region can communicate with each other.

[0039] Distributed cloud is a cloud computing service in which cloud resource providers distribute public cloud services to different physical locations, and the cloud resource providers are responsible for the operation, governance, updates and evolution of the cloud services. Figure 1In this context, cloud resource providers can deploy public cloud computing, storage, and network resources corresponding to distributed cloud products in a hardware and software integrated manner to the customer's local data center (for ease of distinction, distributed cloud products deployed in the customer's local data center can be called local dedicated clusters, or CDC for short), thereby meeting the customer's business needs for data security, low latency, etc. Since distributed cloud products reuse the capabilities of the public cloud, the management and control of distributed cloud products use the same system as on the public cloud. That is, although distributed cloud products are deployed at the customer's site (i.e., the customer's data center), they are connected to the management and control services of the public cloud center through a management and control channel. The public cloud center can control the distributed cloud products to perform a series of operations by issuing signaling commands, and the distributed cloud products can also interact with the public cloud center's management and control by uploading signaling commands. The local dedicated cluster deployed at the customer's site can directly interconnect with the customer's local Internet data center, allowing users to use the rich capabilities of the public cloud with localized latency and data security.

[0040] Virtual Private Networks (VPNs) serve as a management channel connecting public and distributed cloud products. Based on network tunneling technology, they establish a secure and reliable encrypted channel over the internet, enabling resource connectivity between local data centers, office networks, mobile clients, and cloud products provided by cloud resource providers. In distributed cloud scenarios, the local dedicated cluster deployed at the customer site shares a cloud central control system with the adjacent public cloud region. This cloud central control can be implemented through cloud servers and uses the VPN for control traffic transmission. Control traffic is primarily used for issuing and sending control signaling, forming the basis for the local dedicated cluster to reuse various capabilities on the public cloud. All public cloud capabilities require signaling to be issued through the cloud central control system, while simultaneously, the local dedicated cluster also sends signaling to the cloud central control system.

[0041] Cloud services can provide various sub-services for local dedicated clusters, each with its own dedicated business team. For example, these sub-services may include Virtual Network Connection Service (VPN service), Cloud Disk Service (CBS service), Cloud Virtual Machine Service (CVM service), Cloud Object Storage Service (COS service), and so on. Correspondingly, the business team corresponding to the VPN service can be called the VPN business team, the CBS service the CBS business team, the CVM service the CVM business team, and the COS service the COS business team. These business teams can be collectively referred to as the local dedicated cluster's associated team (CDC associated team for short).

[0042] Virtual network connection service can be understood as a technical service that establishes a secure connection on a public network. It establishes communication between the client's local server and the shared cloud server by using encryption protocols and authentication mechanisms, and performs communication encryption and protection.

[0043] Cloud Block Storage (CBS) service allows users to create, mount, and manage hard drives in the cloud.

[0044] Cloud Virtual Machines (CVMs) are general-purpose cloud servers, suitable for applications of all sizes, including web applications, big data processing, and artificial intelligence. CVM services offer a wide range of configuration options, allowing for customization based on user needs. This includes adjusting the quantity and size of resources such as CPUs, memory, and storage, providing high flexibility and enabling users to configure resources more precisely according to their business requirements.

[0045] Tencent Cloud Object Storage (COS) is a secure, reliable, low-cost, and highly scalable cloud-based storage service. It can store various unstructured data, including images, videos, audio, and text. Supporting standard Hypertext Transfer Protocol / Hypertext Secure Transfer Protocol (HTTP / HTTPS), COS allows data uploading, downloading, and management via Application Programming Interface (API), Software Development Kit (SDK), and a console. COS also provides multiple security measures, including data protection, backup, recovery, and data transmission encryption, ensuring the security and reliability of user data. Furthermore, it supports automated and intelligent data lifecycle management, helping users reduce data storage costs. COS is a crucial component of storage services in the cloud computing field, widely used in cloud storage, backup, archiving, and content distribution scenarios.

[0046] As mentioned earlier, the introduction of additional management channels in local dedicated clusters increases their instability, frequently leading to various alarm issues during cloud service operations. However, traditional fault handling methods often result in relatively independent and opaque alarm information. When alarms occur, it's impossible to promptly determine the fault type, requiring a coordinated response from the entire business team. This leads to lengthy problem localization times, numerous invalid responses, and significant manpower consumption, severely impacting business service availability. For example, Figure 2The following is a schematic diagram of the traditional fault handling process: (1) If the CBS business team finds that the task execution failed when performing cloud disk snapshot rollback, the team will trigger an alarm; (2) After the alarm is received by the team's operations and maintenance personnel, if a non-CBS service problem is found, the team can only pull up all the teams under the local dedicated cluster (CDC) product for joint investigation. In the investigation process, the team will start with CBS problems. If it is found that the problem is with calling the COS service, and the COS service is located to find that the problem is with calling the CVM service, the problem will be transferred to the CVM team; (3) The CVM business team finds that the network is not working. They first suspect that the VPN connection is disconnected and get involved in the VPN team to confirm, but the result is that the VPN connection is normal; (4) The CVM business team can only continue to check the logs for investigation and finally find that the corresponding host machine is unreachable by ping; (5) The CVM business team informs the CDC team of the final investigation results; (6) The CDC team receives the conclusion and finally passes it to the CBS business team. The CBS business team waits for the host machine ping unreachable problem to be fixed and blocks the relevant alarms.

[0047] Depend on Figure 2 The information transmission chain shown illustrates that service unavailability was caused by the CVM team's host machine being unreachable via ping. In this situation, the CVM business team should have received the alert and resolved the issue promptly; other related business teams did not need to respond and could have simply waited for the CVM team to fix the problem. However, due to the broken information chain, each business team, upon receiving the alert for its own service, mobilized all related product teams under the CDC and systematically located the problem through the service chain. This resulted in manpower expenditure for both the business and CDC teams. This manpower expenditure could have been avoided, but because the non-problem teams discovered the alert first, the investigation chain was too long, causing the problem teams to get involved too late and consuming excessive manpower, which is extremely detrimental to the live network operation and maintenance.

[0048] Figure 3 This illustration shows an application scenario diagram of the cloud environment fault handling method provided in an embodiment of this application. For example... Figure 3As shown, to address the aforementioned issues, cloud servers in the public cloud region of a cloud resource provider can be connected to local servers in the customer's local dedicated cluster via a Virtual Private Network (VPN). This allows the computing, storage, and network resources of the public cloud to be deployed in an integrated hardware and software manner in the customer's local data center, meeting the customer's business needs for data security and low latency. Furthermore, the public cloud region can also include multiple business servers. These business servers establish communication connections with the cloud servers. A business server can be understood as the server corresponding to each business team providing specific services to the local dedicated cluster during cloud service operations within the cloud environment. For example, these specific services could include Virtual Network Connection Service (VPN), Cloud Disk Service (CBS), Cloud Virtual Machine Service (CVM), Cloud Object Storage Service (COS), and so on.

[0049] In this embodiment, the cloud server can obtain current fault alarm information of the current cloud environment, which includes at least one operating status parameter; obtain a pre-stored set of fault alarm information, which includes one or more historical fault alarm information prior to the current cloud service; obtain a set of conditional probabilities and a set of prior probabilities corresponding to the historical fault alarm information, where the set of conditional probabilities includes the conditional probability corresponding to the category of each operating status parameter, and the set of prior probabilities includes the prior probability corresponding to each preset fault category; based on the category of the operating status parameter, filter out the conditional probabilities corresponding to the operating status parameter from the set of conditional probabilities; determine the posterior probability that the current fault alarm information belongs to each preset fault category based on the conditional probability and the prior probability; based on the posterior probability, filter out the target fault category corresponding to the current fault alarm information from the preset fault categories, and send the current fault alarm information to the business server corresponding to the target fault category.

[0050] The following sections provide detailed descriptions of each example. It should be noted that the order in which the embodiments are described is not intended to limit the preferred order of the embodiments.

[0051] This embodiment will be described from the perspective of a cloud environment fault handling device, which can be integrated into an electronic device, which can be a cloud server.

[0052] A cloud environment fault handling method includes: acquiring current fault alarm information of the current cloud environment, the current fault alarm information including at least one operating status parameter; acquiring a pre-stored set of fault alarm information, the set of fault alarm information including one or more historical fault alarm information prior to the current cloud service; acquiring a set of conditional probabilities and a set of prior probabilities corresponding to the historical fault alarm information, the set of conditional probabilities including the conditional probability corresponding to the category of each operating status parameter, and the set of prior probabilities including the prior probability corresponding to each preset fault category; based on the category of the operating status parameter, filtering out the conditional probabilities corresponding to the operating status parameter from the set of conditional probabilities; determining the posterior probability that the current fault alarm information belongs to each preset fault category based on the conditional probability and the prior probability; based on the posterior probability, filtering out the target fault category corresponding to the current fault alarm information from the preset fault categories, and sending the current fault alarm information to the business server corresponding to the target fault category.

[0053] Figure 4 A flowchart illustrating the cloud environment fault handling method provided in an embodiment of this application is shown. Figure 4 As shown, the specific process of this cloud environment fault handling method is as follows:

[0054] 101. Obtain the current fault alarm information of the current cloud environment. The current fault alarm information includes at least one running status parameter.

[0055] The current cloud environment can be understood as the cloud environment in which the cloud services provided to the customer's site are located at the current moment.

[0056] The current fault alarm information can be understood as alarm information initiated when an anomaly occurs during the operation of the cloud services provided by the current cloud environment. In this embodiment, there are multiple ways to obtain the current fault alarm information of the current cloud environment. For example, the cloud server can detect the current cloud environment during operation, and generate current fault alarm information based on the detected fault. Alternatively, the business servers corresponding to each business team can also detect the current cloud environment during operation, generate current fault alarm information based on the detected fault, and send the current fault alarm information to the cloud server, which can then obtain the current fault alarm information.

[0057] The current fault alarm information may include at least one operational status parameter. Operational status parameters can be understood as the status data corresponding to various evaluation indicators reflecting the current cloud environment status. These evaluation indicators can be understood as categories of operational status parameters. For example, evaluation indicators may include fault scope, switch packet forwarding rate, VPN slave machine traffic rate, host machine traffic rate, etc. In other words, the categories of operational status parameters can be divided into fault scope, switch packet forwarding rate, VPN slave machine traffic rate, host machine traffic rate, etc.

[0058] The fault range can be understood as the percentage of services experiencing failures (i.e., the number of failures) out of the total number of services in the current cloud environment. For ease of distinction, the fault range can be represented by a1. The value of the fault range can be between 0 and 100; the larger the ratio, the larger the fault range. For example, if a1 = 100, meaning the corresponding operating status parameter for the fault range is 100, it indicates that all services have failed.

[0059] Switch packet forwarding rate refers to the number of millions of data packets (Mpps) a switch can forward per second. It represents the switching capacity of a switch, expressed in units of data packets. For ease of understanding, the switch packet forwarding rate can be represented by a2. The value of the switch packet forwarding rate can range from 0 to 100. A higher packet forwarding rate means the switch is less prone to packet loss or latency in high-traffic environments. For example, if a2 = 50, the corresponding operating status parameter for the switch packet forwarding rate is 50, indicating that the switch packet forwarding rate is 50%, and only internal network traffic is being transmitted.

[0060] The VPN sub-machine traffic rate is the percentage of current VPN sub-machine traffic relative to the total VPN bandwidth. For clarity, it can be represented by a3. The VPN sub-machine traffic rate can range from 0 to 100. For example, if a3 = 0, meaning the corresponding operating status parameter is 0, it indicates the VPN channel is completely disconnected and there is no traffic on the VPN sub-machine.

[0061] The host machine traffic rate is the percentage of current host machine traffic relative to the total host machine bandwidth. For ease of distinction, the host machine traffic rate can be represented by a4. The host machine traffic rate can range from 0 to 100. For example, if a4 = 50, meaning the corresponding operating status parameter for the host machine traffic rate is 50, it indicates that 50% of the traffic within the host machine is internal, consisting only of internal network host machine traffic.

[0062] The categories of operational status parameters included in the current fault alarm information may be one or more of the aforementioned evaluation indicators. The categories of operational status parameters included in the current fault alarm information are based on the actual detected data. For example, the current fault alarm information may only include operational status parameters corresponding to the fault range; or, the current fault alarm information may also include operational status parameters corresponding to the fault range and VPN submachine traffic rate; or, the current fault alarm information may also simultaneously include operational status parameters corresponding to the fault range, switch packet forwarding rate, VPN submachine traffic rate, and host machine traffic rate, and so on.

[0063] The operational status parameters are the specific values ​​corresponding to each category of operational status parameters at the current moment, i.e., the values ​​of a1, a2, a3, and a4 at the current moment. For example, a1 = 100, a2 = 50, a3 = 0, and a4 = 50. It should be understood that the operational status parameters can also be other values; the above are merely examples of possible data values ​​and do not constitute a limitation. The operational status parameters reflect the status values ​​of various evaluation indicators in the current cloud environment, and therefore indicate the operational status of the current cloud environment. At different times, the operational status parameters corresponding to the same evaluation indicator may be the same or different.

[0064] 102. Obtain a pre-stored set of fault alarm information, which includes one or more historical fault alarm information prior to the current cloud service.

[0065] The pre-stored fault alarm information set can be understood as a set of sample data corresponding to historical fault alarm information generated before the current moment.

[0066] 103. Obtain the conditional probability set and prior probability set corresponding to historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each operating status parameter, and the prior probability set includes the prior probability corresponding to each preset fault category.

[0067] The conditional probability set and the prior probability set can be understood as sets of Bayesian probabilities calculated from historical fault alarm information in a pre-stored fault alarm information set. The preset fault category can be understood as the possible fault type corresponding to the current fault alarm information when a current fault alarm information for the current cloud environment is detected. In this application embodiment, practical experience has shown that the fault types for the current cloud environment mainly include three categories: "VPN tunnel disconnected"; "Host machine Internet packet explorer unreachable," or simply "Host machine ping unreachable"; and "Support service unavailable." For ease of distinction, the preset fault category can be represented as T, where T can be any one of T1, T2, and T3. T1 represents the preset fault category "VPN tunnel disconnected," T2 represents the preset fault category "Host machine ping unreachable," and T3 represents the preset fault category "Support service unavailable."

[0068] In this embodiment, the cloud server can obtain a prior probability set corresponding to historical fault alarm information. The prior probability set includes the prior probability corresponding to each preset fault category. The prior probability is the probability that a sample belongs to a certain preset fault category without any feature information. The prior probability can be expressed as P(Ti), where i can be 1, 2, or 3. For example, if T1 = 0.11, T2 = 0.53, and T3 = 0.36, it means that without any feature information, the probability of a sample belonging to T1 is 11%, the probability of a sample belonging to T2 is 53%, and the probability of a sample belonging to T3 is 36%.

[0069] As mentioned above, assessment metrics reflecting the current cloud environment status may include fault scope, switch packet forwarding rate, virtual private network (VPN) slave device traffic rate, and host machine traffic rate, etc. For ease of description, in this embodiment, the above assessment metrics are referred to as categories of operating status parameters. Therefore, the categories of operating status parameters include fault scope, switch packet forwarding rate, VPN slave device traffic rate, and host machine traffic rate. In this embodiment, the cloud server can also obtain a set of conditional probabilities corresponding to historical fault alarm information. The set of conditional probabilities may include the conditional probabilities corresponding to each of the fault scope (a1), switch packet forwarding rate (a2), VPN slave device traffic rate (a3), and host machine traffic rate (a4).

[0070] In Bayesian theory, conditional probability is the probability distribution of a sample within a certain category. In this embodiment, conditional probability can be understood as the probability that a sample belongs to a certain operating state parameter within a certain evaluation metric when a preset fault category occurs. Conditional probability can be expressed as P(X|Ti), where X = a1, a2, a3, or a4. For example, P(a1|T1) represents the probability that a sample belongs to a certain operating state parameter within the fault range when the preset fault category is "VPN channel disconnected," and P(a2|T1) represents the probability that a sample belongs to a certain operating state parameter within the switch packet forwarding rate when the preset fault category is "VPN channel disconnected."

[0071] In this embodiment of the application, the conditional probability set corresponding to the category of each operating state parameter may include the conditional probabilities corresponding to the category of the operating state parameter under T1, T2, and T3, respectively. For example, the conditional probability set corresponding to the category of the operating state parameter being the switch packet forwarding rate includes P(a2|T1), P(a2|T2), and P(a2|T3).

[0072] The process of obtaining the conditional probability set and prior probability set corresponding to historical fault alarm information may include: obtaining the historical operating status parameters corresponding to the historical fault alarm information and the preset fault categories corresponding to the historical status parameters; and determining the conditional probability set and prior probability set based on the historical operating status parameters and the preset fault categories.

[0073] Figure 5 This illustration shows a cloud environment fault handling process based on historical fault alarm information, as provided in an embodiment of this application. Figure 5 As shown, the cloud environment fault handling method starting from historical fault alarm information mainly includes three stages: preprocessing, training, and application. The preprocessing stage involves acquiring historical fault alarm information and determining the categories of historical operating state parameters and preset fault categories based on the historical operating state parameters in the historical fault alarm information. Then, in the training stage, Bayesian probability distribution calculations can be performed based on the data obtained in the preprocessing stage to obtain a set of prior probabilities P(Ti) and a set of conditional probabilities P(X|Ti). Finally, in the application stage, the posterior probability of the current fault alarm information's operating state parameters for each preset fault category can be further calculated based on the set of prior probabilities P(Ti) and the set of conditional probabilities P(X|Ti). The posterior probabilities are then fused, and the preset fault category corresponding to the highest probability term is selected as the result.

[0074] Figure 6 This illustration shows a schematic diagram of a pre-stored set of fault alarm information provided in an embodiment of this application. For example... Figure 6As shown, the pre-stored fault alarm information set may include multiple historical fault alarm information. For example, the pre-stored fault alarm information set may include historical fault alarm information 1, historical fault alarm information 2, historical fault alarm information 3, ..., historical fault alarm information m, where m is a natural number greater than 3.

[0075] As mentioned earlier, the pre-stored fault alarm information set can be understood as a collection of sample data corresponding to historical fault alarm information generated before the current moment. Since the historical fault alarm information occurred before the current moment, it means that the cloud server has already performed risk positioning on the historical fault alarm information and determined the fault category corresponding to the historical fault alarm information. Therefore, the historical fault alarm information may also include the preset fault category corresponding to the historical fault alarm information. Accordingly, each historical fault alarm information in this application embodiment may include two parts: the first part is at least one historical operating status parameter included in the historical fault alarm information; the second part is the preset fault category corresponding to the historical fault alarm information, that is, the preset fault category corresponding to the historical operating status parameter in the historical fault alarm information.

[0076] For any historical fault alarm message, it can be represented as {a1, a2, a3, a4, T}, where a1 is the historical operating status parameter corresponding to the fault range, a2 is the historical operating status parameter corresponding to the switch packet forwarding rate, a3 is the historical operating status parameter corresponding to the VPN slave machine traffic rate, and a4 is the historical operating status parameter corresponding to the master machine traffic rate. T is a preset fault category, which can be any one of T1, T2, and T3. T1 represents the fault category "VPN tunnel disconnected", T2 represents the fault category "master machine Internet packet explorer unreachable", abbreviated as "master machine ping unreachable", and T3 represents the fault category "support services unavailable".

[0077] If the sample data of the pre-stored fault alarm information set is as follows:

[0078] Historical fault alarm information 1: {100, 50, 0, 50, T1};

[0079] Historical fault alarm information 2: {50, 80, 95, 90, T3};

[0080] Historical fault alarm information 3: {10, 20, 80, 0, T2};

[0081] Historical fault alarm information 4: {12, 23, 88, 0, T2};

[0082] Historical fault alarm information 5: {49, 79, 95, 90, T3};

[0083] Historical fault alarm information 6: {100, 52, 0, 46, T1};

[0084] Historical fault alarm information 7: {100, 54, 0, 49, T1};

[0085] Historical fault alarm information 8: {11, 28, 100, 0, T2};

[0086] Historical fault alarm information 9: {53, 78, 93, 89, T3};

[0087] Historical fault alarm information 10: {52, 81, 94, 91, T3};

[0088] Historical fault alarm information 11: {10, 25, 89, 0, T2};

[0089] ...

[0090] Taking historical fault alarm information 2 as an example, it can be represented as {50, 80, 95, 90, T3}. Here, 50, 80, 95, and 90 belong to the historical operating status parameters of the first part, and T3 belongs to the preset fault category of the second part. That is, the preset fault category corresponding to historical fault alarm information 2 is "Support service unavailable". In historical fault alarm information 2, 50 represents the historical operating status parameter corresponding to the fault range (a1), 80 represents the historical operating status parameter corresponding to the switch packet forwarding rate (a2), 95 represents the historical operating status parameter corresponding to the virtual private network submachine traffic rate (a3), and 90 represents the historical operating status parameter corresponding to the master machine traffic rate (a4).

[0091] The conditional probability set and the prior probability set can be calculated based on the pre-stored fault alarm information set. Specifically, the conditional probability set and the prior probability set can be determined based on historical operating status parameters and preset fault categories.

[0092] The process of determining the conditional probability set and prior probability set based on historical operating status parameters and preset fault categories may include: counting the number of historical fault alarm messages corresponding to each preset fault category in historical fault alarm information to obtain the category sample number; calculating the prior probability of each preset fault category based on the category sample number to obtain the prior probability set; extracting the mapping relationship between historical operating status parameters and preset fault categories from historical fault alarm information; and calculating at least one conditional probability of the category of historical operating status parameters under each preset fault category based on the mapping relationship to obtain the conditional probability set.

[0093] The pre-stored fault alarm information set includes multiple historical fault alarm messages. Each historical fault alarm message includes its corresponding preset fault category. Therefore, the number of historical fault alarm messages corresponding to each preset fault category can be calculated from the pre-stored fault alarm information set to obtain the category sample size corresponding to that preset fault category. For example, if the pre-stored fault alarm information set includes 100,000 historical fault alarm messages, of which 55,000 historical fault alarm messages correspond to the preset fault category of VPN tunnel disconnection (T1), 27,000 historical fault alarm messages correspond to the preset fault category of host machine ping unreachable (T2), and the remaining 18,000 historical fault alarm messages correspond to the preset fault category of support service unavailable (T3), then the category sample size for VPN tunnel disconnection (T1) is 55,000, the category sample size for host machine ping unreachable (T2) is 27,000, and the category sample size for support service unavailable (T3) is 18,000.

[0094] For any preset fault category, after determining the number of samples corresponding to that category, the prior probability of that preset fault category can be calculated based on the number of samples, thus obtaining a prior probability set. The prior probability set includes the prior probability corresponding to each preset fault category. Specifically, calculating the prior probability of each preset fault category based on the number of samples can include: obtaining the total number of historical fault alarm messages in a pre-stored fault alarm information set to obtain the total number of samples; and using the proportion of the category sample count in the total sample count as the prior probability to obtain the prior probability set.

[0095] The total number of samples is the number of historical fault alarm messages contained in the pre-stored fault alarm information set. For example, in the example above, if the pre-stored fault alarm information set includes 100,000 historical fault alarm messages, then the total number of samples is 100,000.

[0096] After determining the total number of samples and the number of samples for each preset fault category, the proportion of each category sample in the total number of samples can be calculated to obtain the prior probability corresponding to the preset fault category. For example, if the number of samples for the VPN tunnel disconnection (T1) category is 55,000, then the prior probability P(T1) for VPN tunnel disconnection (T1) is 55%; if the number of samples for the host machine ping unreachable (T2) category is 27,000, then the prior probability P(T2) for host machine ping unreachable (T2) is 27%; and if the number of samples for the support service unavailable (T3) category is 18,000, then the prior probability P(T3) for support service unavailable (T3) is 18%.

[0097] As mentioned above, the conditional probability in this embodiment can be understood as the probability that a sample belongs to a certain operating state parameter in a certain evaluation index when a certain preset fault category occurs, denoted as P(X|Ti), where X = a1, a2, a3 or a4, i = 1, 2 or 3. For example, P(a1|T1) represents the probability that historical fault alarm information belongs to a certain historical operating state parameter in evaluation index a1 when the preset fault category T1 occurs. Since the value range of the historical operating state parameter of evaluation index a1 is 0-100, that is, the historical operating state parameter is a continuous value, P(a1|T1) can represent the probability that historical fault alarm information belongs to any historical operating state parameter in evaluation index a1 when the preset fault category T1 occurs, and the historical operating state parameter is between 0 and 100. For example, P(a1|T1) can represent the probability that a historical fault alarm message belongs to the historical operating state parameter of 100 in a1 when a preset fault category T1 occurs; P(a1|T1) can also represent the probability that a historical fault alarm message belongs to the historical operating state parameter of 20 in a1 when a preset fault category T1 occurs, and so on.

[0098] For different historical fault alarm messages, the historical operating status parameters and preset fault categories contained therein may be the same or different. This means that the mapping relationship between the historical operating status parameters and preset fault categories may be the same or different in different historical fault alarm messages. For example, in the aforementioned pre-stored fault alarm message set, taking the category of historical operating status parameters as a1 and the preset fault category as T1 as an example, in historical fault alarm messages 1, 6, and 7, the historical operating status parameter 100 of a1 has a mapping relationship with the preset fault category T1; in historical fault alarm message 2, the historical operating status parameter 50 of a1 has a mapping relationship with the preset fault category T1; and in historical fault alarm message 2, the historical operating status parameter 10 of a1 has a mapping relationship with the preset fault category T1.

[0099] Therefore, to determine the conditional probabilities, it is necessary to first determine the mapping relationship between the historical operating state parameters in each historical fault alarm information in the pre-stored fault alarm information set and the preset fault categories; then, based on the mapping relationship, calculate at least one conditional probability of the category of the historical operating state parameter under each preset fault category to obtain a set of conditional probabilities. Specifically, calculating at least one conditional probability of the category of the historical operating state parameter under each preset fault category based on the mapping relationship to obtain a set of conditional probabilities may include: obtaining at least two preset discrete intervals corresponding to the category of each historical operating state parameter, and a preset range corresponding to each preset discrete interval, wherein the preset discrete intervals indicate the data distribution intervals corresponding to the historical operating state parameters; determining the historical operating state parameters contained in each preset discrete interval based on the preset range to obtain interval historical state parameters; calculating the interval conditional probability of each preset discrete interval corresponding to the category of the historical operating state parameter under the preset fault category based on the mapping relationship and the interval historical state parameters; and using the interval conditional probability as the conditional probability to obtain the set of conditional probabilities corresponding to the category of the historical operating state parameter.

[0100] Since the historical operating state parameter ranges from 0 to 100, and can be any value between 0 and 100 (i.e., a continuous variable), it is not suitable to calculate the probability based on a specific value within the aforementioned range. In this embodiment, the historical operating state parameter can be divided into at least two discrete intervals. For ease of description, each discrete interval is referred to as a preset discrete interval. For example, the historical operating state parameter can be divided into two preset discrete intervals, three preset discrete intervals, four, or other preset discrete intervals.

[0101] The preset range corresponding to each preset discrete interval can be set according to the actual situation. When there are two or more preset discrete intervals corresponding to each category of historical operating status parameter, the preset range span of each preset discrete interval can be the same or different. Taking the historical operating status parameter corresponding to category a1 of historical operating status parameter as an example, the preset ranges corresponding to these three preset discrete intervals can be set to [0,30), [30,60), and [60,100] respectively.

[0102] It should be understood that for different categories of historical operating state parameters, the number of their respective preset discrete intervals and the preset range of each preset discrete interval can be the same or different. For example, for category a2 of historical operating state parameters, there can also be three preset discrete intervals, and the preset ranges corresponding to these three preset discrete intervals can be set to [0,30), [30,60), and [60,100].

[0103] After determining the preset discrete interval corresponding to the category of each historical operating status parameter, and the preset range corresponding to each preset discrete interval, the historical operating status parameters in each historical fault alarm information can be divided into the corresponding preset discrete intervals according to the preset ranges, thereby determining the historical operating status parameters contained in each preset discrete interval. At this time, the preset discrete intervals can indicate the data distribution intervals corresponding to each historical operating status parameter.

[0104] For example, if the category a1 of the historical operating status parameter corresponds to three preset discrete intervals, and the preset ranges of each preset discrete interval are [0,30), [30,60), and [60,100] respectively, then in the aforementioned pre-stored fault alarm information set, the historical operating status parameter a1=100 in historical fault alarm information 1 falls within the interval [60,100], the historical operating status parameter a1=50 in historical fault alarm information 2 falls within the interval [30,60), and so on. Therefore, the historical operating status parameters contained in each preset discrete interval can be statistically analyzed to obtain the interval's historical status parameters. Based on this, for the preset discrete interval [0, 30) corresponding to category a1 of historical operating status parameters, the historical operating status parameters included in the preset discrete interval [0, 30) are: a1 (a1 = 10) in historical fault alarm information 3, a1 (a1 = 12) in historical fault alarm information 4, a1 (a1 = 11) in historical fault alarm information 8, a1 (a1 = 10) in historical fault alarm information 11, and so on. Similarly, for the preset discrete interval [30, 60) corresponding to category a2 of historical operating status parameters, the historical operating status parameters included in the preset discrete interval [30, 60) are: a1 (a1 = 50) in historical fault alarm information 2, a1 (a1 = 49) in historical fault alarm information 5, a1 (a1 = 53) in historical fault alarm information 9, a1 (a1 = 52) in historical fault alarm information 10, and so on.

[0105] After determining the interval historical state parameters, the interval conditional probability of each preset discrete interval corresponding to the category of the historical operating state parameter under the preset fault category can be calculated based on the mapping relationship and the interval historical state parameters. This interval conditional probability is then used as the conditional probability to obtain the set of conditional probabilities corresponding to the category of the historical operating state parameter. Specifically, calculating the interval conditional probability of each preset discrete interval corresponding to the category of the historical operating state parameter under the preset fault category based on the mapping relationship and the interval historical state parameters can include: taking each preset fault category as the target fault category and the number of category samples corresponding to the target fault category as the target category sample quantity; taking each preset discrete interval corresponding to the category of the historical operating state parameter as the target first discrete interval; finding the number of target historical state parameters that have a mapping relationship with the target fault category from the interval historical state parameters of the target first discrete interval to obtain the target data volume; and calculating the proportion of the target data volume in the target category sample quantity to obtain the interval conditional probability of the target first discrete interval under the target fault category.

[0106] For any category of historical operating state parameter, its conditional probability set includes the interval conditional probabilities of each preset discrete interval under each preset fault category. For example, for category a1 of historical operating state parameter, its conditional probability set includes three interval conditional probabilities under the preset discrete interval [0,30), namely P(a1|T1), P(a1|T2), and P(a1|T3); three interval conditional probabilities under the preset discrete interval [30,60), namely P(a1|T1), P(a1|T2), and P(a1|T3); and three interval conditional probabilities under the preset discrete interval [60,100], namely P(a1|T1), P(a1|T2), and P(a1|T3).

[0107] Similarly, for category a2 of historical operating state parameters, its conditional probability set includes three interval conditional probabilities under the preset discrete interval [0,30), namely P(a2|T1), P(a2|T2), and P(a2|T3), three interval conditional probabilities under the preset discrete interval [30,60), namely P(a2|T1), P(a2|T2), and P(a2|T3), and three interval conditional probabilities under the preset discrete interval [60,100], namely P(a2|T1), P(a2|T2), and P(a2|T3).

[0108] Based on the above, for any interval conditional probability, the calculation method is as follows: determine the corresponding target fault category and take the number of category samples corresponding to the target fault category as the target category sample number; determine the corresponding target discrete interval and select the number of target historical state parameters that have a mapping relationship with the target fault category in the target discrete interval to obtain the target data volume; then divide the target data volume by the target category sample number, and the ratio obtained is the interval conditional probability.

[0109] Taking the calculation of the interval conditional probability P(a1|T1) under the preset discrete interval [0,30) as an example, T1 is the target fault category. As mentioned earlier, the number of category samples of T1 in the pre-stored fault alarm information set is 55,000, so the number of target category samples is 55,000. At this time, the preset discrete interval [0,30) is taken as the first target discrete interval. If the number of interval historical state parameters of category a1 containing historical operating state parameters in the preset discrete interval [0,30) is 20,000, and the number of target historical state parameters that have a mapping relationship with T1 are selected from the interval historical state parameters of the preset discrete interval [0,30) is 3,850 (target data volume). After determining the number of target category samples and the target data volume, the proportion of the target data volume in the number of target category samples can be calculated. The calculation yields 3850 / 55000 = 0.07. Therefore, the interval conditional probability of the preset discrete interval [0,30) of category a1 of the historical operating state parameter under T1 is 0.07, which means that when T1 occurs, the probability that the historical operating state parameter belongs to the preset discrete interval [0,30) of category a1 is 0.07.

[0110] Taking the calculation of the interval conditional probability P(a1|T1) under the preset discrete interval [30,60) as an example, T1 is the target fault category. As mentioned earlier, the number of category samples of T1 in the pre-stored fault alarm information set is 55,000, so the number of target category samples is 55,000. At this time, the preset discrete interval [30,60) is taken as the first target discrete interval. If the number of interval historical state parameters of category a1 containing historical operating state parameters in the preset discrete interval [30,60) is 20,000, and the number of target historical state parameters that have a mapping relationship with T1 are selected from the interval historical state parameters of the preset discrete interval [30,60) is 4,400 (target data volume). After determining the number of target category samples and the target data volume, the proportion of the target data volume in the number of target category samples can be calculated. The calculation yields 4400 / 55000 = 0.08. Therefore, the interval conditional probability of the preset discrete interval [30, 60) of category a1 of the historical operating state parameter under T1 is 0.08, which means that when T1 occurs, the probability that the historical operating state parameter belongs to the preset discrete interval [30, 60) of category a1 is 0.08.

[0111] Taking the calculation of the interval conditional probability P(a1|T1) under the preset discrete interval [60,100] as an example, T1 is the target fault category. As mentioned earlier, the number of category samples of T1 in the pre-stored fault alarm information set is 55,000, so the number of target category samples is 55,000. At this time, the preset discrete interval [60,100] is taken as the first target discrete interval. If the number of interval historical state parameters of category a1 containing historical operating state parameters in the preset discrete interval [60,100] is 60,000, and the number of target historical state parameters that have a mapping relationship with T1 are selected from the interval historical state parameters of the preset discrete interval [60,100] is 46,750 (target data volume). After determining the number of target category samples and the target data volume, the proportion of the target data volume in the number of target category samples can be calculated. The calculation yields 46750 / 55000 = 0.85. Therefore, the interval conditional probability of the preset discrete interval [60,100] of category a1 of the historical operating state parameter under T1 is 0.85, which means that when T1 occurs, the probability that the historical operating state parameter belongs to the preset discrete interval [60,100] of category a1 is 0.85.

[0112] For ease of explanation, based on the aforementioned pre-stored fault alarm information set, the conditional probabilities corresponding to the categories of each historical operating state parameter are calculated according to the method described above, resulting in a set of conditional probabilities, as shown in Table 1 below. It should be understood that the set of conditional probabilities listed in Table 1 is merely for illustrative purposes and does not impose any limitations on the actual set of conditional probabilities.

[0113] Table 1. Conditional probability set of the pre-stored fault alarm information set

[0114]

[0115]

[0116] Table 1 contains four conditional probability sets: the conditional probability set corresponding to a1, a2, a3, and a4. Each conditional probability set further includes multiple preset discrete intervals with interval conditional probabilities for each preset fault category. For example, the set of conditional probabilities corresponding to a1 includes three interval conditional probabilities in the interval [0,30), three interval conditional probabilities in the interval [30,60], and three interval conditional probabilities in the interval [60,100]. The three interval conditional probabilities in the interval [0,30) are P(a1|T1) = 0.07, P(a1|T2) = 0.72, and P(a1|T3) = 0.13, respectively; the three interval conditional probabilities in the interval [30,60] are P(a1|T1) = 0.08, P(a1|T2) = 0.17, and P(a1|T3) = 0.73, respectively; and the three interval conditional probabilities in the interval [60,100] are P(a1|T1) = 0.85, P(a1|T2) = 0.11, and P(a1|T3) = 0.14, respectively.

[0117] 104. Based on the category of the running state parameter, select the conditional probability corresponding to the running state parameter from the conditional probability set.

[0118] After obtaining the conditional probability set and prior probability set corresponding to the pre-stored fault alarm information set, the conditional probability corresponding to the category of the running status parameter can be filtered out from the conditional probability set based on the current fault alarm information detected in the current cloud environment.

[0119] In this embodiment, the category of the operating status parameters is the evaluation indicators involved in the pre-stored fault alarm information set, which may include fault range (a1), switch packet forwarding rate (a2), virtual private network (VPN) slave device traffic rate (a3), and host machine traffic rate (a4). The current fault alarm information may include all evaluation indicators or only some evaluation indicators. If the current fault alarm information includes operating status parameters of all evaluation indicators, the category of the operating status parameters includes all evaluation indicators; if the current fault alarm information only includes operating status parameters of some evaluation indicators, the category of the operating status parameters includes some evaluation indicators. That is, the category of the operating status parameters in the current fault alarm information may include at least one of fault range (a1), switch packet forwarding rate (a2), VPN slave device traffic rate (a3), and host machine traffic rate (a4).

[0120] Based on this, the conditional probabilities corresponding to the operating state parameters are selected from the conditional probability set according to the category of the operating state parameters. This may include: selecting a target conditional probability set corresponding to the category of each operating state parameter from the conditional probability set, wherein the target conditional probability set includes conditional probabilities of at least two preset discrete intervals, and the preset discrete intervals correspond to a preset range; and extracting the conditional probability of the operating state parameters under each preset fault category from the target conditional probability set based on the preset range.

[0121] For example, if the categories of the current fault alarm information's operating status parameters simultaneously include four types: fault range (a1), switch packet forwarding rate (a2), virtual private network submachine traffic rate (a3), and host traffic rate (a4), then the conditional probability sets corresponding to a1, a2, a3, and a4 are all target conditional probability sets. However, if the categories of the current fault alarm information's operating status parameters only include fault range (a1) and switch packet forwarding rate (a2), then only the conditional probability sets corresponding to a1 and a2 are target conditional probability sets.

[0122] For each selected target conditional probability set, there are at least two preset discrete intervals and a preset range corresponding to each preset discrete interval. Since the preset range indicates the upper and lower limits of the operating state parameter of its corresponding preset discrete interval, the preset discrete interval corresponding to the operating state parameter can be selected from the target conditional probability set according to the preset range, and the interval conditional probability corresponding to the selected preset discrete interval can be used as the conditional probability of the operating state parameter under each preset fault category.

[0123] Furthermore, as mentioned earlier, the target conditional probability set also includes the interval conditional probability of each preset discrete interval under each preset fault category. Therefore, based on this, extracting the conditional probability of the operating state parameter under each preset fault category from the target conditional probability set according to a preset range can include: based on a preset range, filtering the target discrete interval corresponding to the operating state parameter from the preset discrete intervals under the category of the operating state parameter; filtering the conditional probability corresponding to the target discrete interval from the target conditional probability set to obtain the target interval conditional probability; and using the target interval conditional probability as the conditional probability of the operating state parameter under the preset fault category.

[0124] For example, if the current fault alarm information for the current cloud environment is {90, 45, 10, 69}, where 90 is the operating status parameter of the fault range (a1), 45 is the operating status parameter of the switch packet forwarding rate (a2), 10 is the operating status parameter of the virtual private network submachine traffic rate (a3), and 69 is the operating status parameter of the host machine traffic rate (a4), then it can be determined that a1 = 90 falls within the interval [60, 100]. That is, in the conditional probability set corresponding to a1, the target discrete interval is [60, 100]. The interval conditional probability of the target discrete interval is the target interval conditional probability. That is, the target interval conditional probability includes three parts, namely P(a1|T1) = 0.85, P(a1|T2) = 0.11, and P(a1|T3) = 0.14. The above three target interval conditional probabilities can be used as the conditional probability of the operating status parameter a1 = 90 under the preset fault category.

[0125] For the operating state parameter a2 = 45, it falls within the interval [30, 60). That is, in the conditional probability set corresponding to a2, the target discrete interval is [30, 60). The interval conditional probability of the target discrete interval is the target interval conditional probability. That is, the target interval conditional probability includes three parts, namely P(a2|T1) = 0.69, P(a2|T2) = 0.09, and P(a2|T3) = 0.15. The above three target interval conditional probabilities can be used as the conditional probability of the operating state parameter a2 = 45 under the preset fault category.

[0126] Similarly, for the operating state parameter a3 = 10, it falls within the interval [0, 20). The conditional probabilities of the target interval within this interval are P(a3|T1) = 0.85, P(a3|T2) = 0.03, and P(a3|T3) = 0.10. These three conditional probabilities can be used as the conditional probabilities of the operating state parameter a3 = 10 under the preset fault category. For the operating state parameter a4 = 69, it falls within the interval [60, 100]. The conditional probabilities of the target interval within this interval are P(a4|T1) = 0.15, P(a4|T2) = 0.04, and P(a4|T3) = 0.80. These three conditional probabilities can be used as the conditional probabilities of the operating state parameter a4 = 69 under the preset fault category.

[0127] 105. Based on conditional probability and prior probability, determine the posterior probability that the current fault alarm information belongs to each preset fault category.

[0128] In the embodiments of this application, the posterior probability can be understood as the probability that a sample belongs to a certain preset fault category under given operating state parameters. The posterior probability can be expressed as P(Ti|X), where i = 1, 2 or 3, and X = a1, a2, a3 or a4.

[0129] Since the Naive Bayes algorithm assumes that each feature attribute an is independent, according to Bayes' theorem, we can deduce: P(Ti|X)=P(X|Ti)P(Ti) / P(X). Since the denominator is a constant for all categories, we only need to maximize the numerator. Furthermore, because the features are mutually independent, we ultimately derive: Where P(Ci) is the prior probability that X is classified into a certain category, and it is a constant in the pre-stored fault alarm information set.

[0130] Similarly, it can be deduced that:

[0131] The process of determining the posterior probability of each preset fault category based on conditional probability and prior probability may include: taking each preset fault category as a candidate fault category and determining at least two conditional probabilities of the current fault alarm information under the candidate fault category; fusing the conditional probabilities to obtain the target conditional probability corresponding to the candidate fault category; and fusing the prior probability of the candidate fault category with the target conditional probability to obtain the posterior probability of the candidate fault category.

[0132] Taking the current fault alarm information in the current cloud environment containing the operating status parameters {90, 45, 10, 69} as an example, when the candidate fault category is T1 (VPN channel disconnected), according to Table 1, the conditional probability of operating status parameter 90 under T1 is P(a1|T1) = 0.85, the conditional probability of operating status parameter 45 under T1 is P(a2|T1) = 0.69, the conditional probability of operating status parameter 10 under T1 is P(a3|T1) = 0.85, and the conditional probability of operating status parameter 69 under T1 is P(a4|T1) = 0.15. At this time, the target conditional probability corresponding to the candidate fault category (T1) is equal to the product of the above conditional probabilities, that is, the target conditional probability = 0.85 × 0.69 × 0.85 × 0.15 = 0.07477875. After determining the target conditional probability corresponding to the candidate fault category (T1), the target conditional probability can be multiplied by the prior probability of the candidate fault category to obtain the posterior probability of the candidate fault category. As mentioned earlier, in the pre-stored fault alarm information set described in Table 1, T1 = 55%. Therefore, the posterior probability of the candidate fault category (T1) = 0.07477875 × 0.55 = 0.0411283125, indicating that the probability of the current fault alarm information being a "VPN channel disconnected" type fault is 0.0411283125.

[0133] When the candidate fault category is T2 (master machine ping unreachable), referring to Table 1, the conditional probability of operating state parameter 90 under T2 is P(a1|T2) = 0.11, the conditional probability of operating state parameter 45 under T2 is P(a2|T2) = 0.09, the conditional probability of operating state parameter 10 under T2 is P(a3|T2) = 0.03, and the conditional probability of operating state parameter 69 under T2 is P(a4|T2) = 0.04. At this time, the target conditional probability corresponding to the candidate fault category (T2) is equal to the product of the above conditional probabilities, i.e., target conditional probability = 0.11 × 0.09 × 0.03 × 0.04 = 0.00001188. After determining the target conditional probability corresponding to the candidate fault category (T2), this target conditional probability can be multiplied by the prior probability of the candidate fault category to obtain the posterior probability of the candidate fault category. As mentioned earlier, in the pre-stored fault alarm information set described in Table 1, T2 = 27%. Therefore, the posterior probability of the candidate fault category (T2) is 0.00001188 × 0.27 = 0.0000032076, which means that the probability of the current fault alarm information being a "master machine ping unreachable" type fault is 0.0000032076.

[0134] When the candidate fault category is T3 (support services unavailable), referring to Table 1, the conditional probability of operating state parameter 90 under T3 is P(a1|T3) = 0.14, the conditional probability of operating state parameter 45 under T3 is P(a2|T3) = 0.15, the conditional probability of operating state parameter 10 under T3 is P(a3|T3) = 0.10, and the conditional probability of operating state parameter 69 under T3 is P(a4|T3) = 0.80. At this point, the target conditional probability corresponding to the candidate fault category (T3) is equal to the product of the above conditional probabilities, i.e., target conditional probability = 0.14 × 0.15 × 0.10 × 0.80 = 0.00168. After determining the target conditional probability corresponding to the candidate fault category (T3), this target conditional probability can be multiplied by the prior probability of the candidate fault category to obtain the posterior probability of the candidate fault category. As mentioned earlier, in the pre-stored fault alarm information set described in Table 1, T3 = 18%, and the posterior probability of the candidate fault category (T3) is 0.00168 × 0.18 = 0.0003024, which means that the probability of the current fault alarm information being a "support service unavailable" type fault is 0.0003024.

[0135] 106. Based on the posterior probability, select the target fault category corresponding to the current fault alarm information from the preset fault categories, and send the current fault alarm information to the business server corresponding to the target fault category.

[0136] After determining the posterior probability of the current fault alarm information for each preset fault category in the current cloud environment, a preset fault category can be selected as the target fault category based on the specific probability value of each posterior probability. The current fault alarm information is then sent to the business server corresponding to the target fault category so that the business server can repair the current fault alarm information. Specifically, selecting the target fault category corresponding to the current fault alarm information from the preset fault categories based on the posterior probability can include: sorting the posterior probabilities of each preset fault category to obtain a posterior probability sorting result; and selecting the target fault category corresponding to the current fault alarm information from the preset fault categories based on the posterior probability sorting result.

[0137] In this embodiment, the preset fault category with the largest posterior probability value can be selected as the target fault category. Therefore, the posterior probabilities of each preset fault category calculated above can be sorted in descending order, and the preset fault category ranked first can be selected as the target fault category; alternatively, the posterior probabilities of each preset fault category calculated above can be sorted in ascending order, and the preset fault category ranked last can be selected as the target fault category.

[0138] Regardless of whether the order is sorted from largest to smallest or smallest to largest, 0.0411283125 is determined to be the first value. Therefore, the target fault category corresponding to the current fault alarm information can be determined to be T1, i.e., "VPN tunnel disconnected".

[0139] Sending the current fault alarm information to the business server corresponding to the target fault category may include: selecting the business server corresponding to the target fault category from the set of business servers corresponding to the current cloud environment to obtain the target business server; and sending the current fault alarm information and the target fault category to the target business server so that the target business server can perform fault troubleshooting based on the current fault alarm information and the target fault category.

[0140] The set of business servers includes multiple business servers, each corresponding to a different business team, and each business team is responsible for resolving specific faults. In this embodiment, a correspondence between business servers and preset fault categories can be pre-established so that, after determining the target fault category, the corresponding business server can be determined based on the correspondence. The correspondence between business servers and preset fault categories can be shown in Table 2.

[0141] Table 2. Correspondence between business servers and preset fault categories

[0142] Preset fault categories Business Team Business Server VPN tunnel disconnected (T1) VPN Business Team VPN Service Server The host machine is unreachable (T2). CVM Business Team CVM Service Server Support services unavailable (T3) CBS Business Team CBS Service Server

[0143] As shown in Table 1, VPN tunnel disconnection (T1) corresponds to the VPN service team. This means that when the target fault category is VPN tunnel disconnection (T1), the cloud server will route the current fault alarm information to the VPN service server corresponding to the VPN service team. In this case, the VPN service server is the target service server, which will then investigate and repair the current fault alarm information. Similarly, host machine ping unreachable (T2) corresponds to the CVM service team. This means that when the target fault category is host machine ping unreachable (T2), the cloud server will route the current fault alarm information to the CVM service server corresponding to the CVM service team. In this case, the CVM service server is the target service server, which will then investigate and repair the current fault alarm information. Support Service Unavailable (T3) corresponds to the CBS business team. It means that when the target fault category is Support Service Unavailable (T3), the cloud server will route the current fault alarm information to the CBS business server corresponding to the CBS business team. At this time, the CBS business server is the target business server, and the CBS business server will investigate and repair the current fault alarm information.

[0144] Figure 7The diagram illustrates a cloud environment fault handling process provided in some embodiments of this application. For example... Figure 7 As shown, after a current fault alarm is input into the cloud server, if the cloud server identifies the target fault category of the current fault alarm as T1, it will route the current fault alarm to the VPN service server corresponding to the VPN service team. That is, the VPN service server will be used as the target service server to investigate and repair the current fault alarm. Furthermore, during the process of routing the current fault alarm to the VPN service server, the cloud server can also simultaneously send the current fault alarm and its corresponding target fault category to other service servers (such as CVM service servers and CBS service servers) to inform them that there is a fault in the current cloud environment. However, other servers that receive the current fault alarm and its corresponding target fault category do not need to respond to it, i.e., they do not need to participate in the investigation and repair of the current fault alarm.

[0145] As mentioned earlier, after receiving the current fault alarm information and target fault category from the cloud server, the target service server can perform fault troubleshooting based on these information. Since the target fault category is predicted by the cloud server based on data from a pre-stored set of fault alarm information, the prediction result is merely a probability-maximizing choice and still carries some uncertainty. Therefore, during the fault troubleshooting process of the target service server based on the current fault alarm information, the troubleshooting result may have two possibilities: First, the detection result after the target service server's troubleshooting of the current fault alarm information indicates that the target fault category matches the fault handling category corresponding to the target service server, meaning the target fault category prediction is correct; second, the detection result after the target service server's troubleshooting of the current fault alarm information indicates that the target fault category differs from the fault handling category corresponding to the target service server, meaning the target fault category prediction is incorrect.

[0146] In the first possible scenario, it means that the target fault category calculated by the cloud server is correct. In this case, the selection of the target business server is also correct, and the target business server can directly repair the historical fault alarm information.

[0147] The second possible scenario implies a discrepancy between the target fault category calculated by the cloud server and the actual fault category. Consequently, the identified target service server also becomes inaccurate, and the target service server is unable to resolve the fault alarm. In response, the cloud server can designate service servers other than the target service server from the service server set as candidate service servers and send the current fault alarm information to these candidate service servers, requesting each candidate service server to troubleshoot based on the current fault alarm information.

[0148] In the second possible scenario, the cloud server in this application embodiment can adopt any one of the following two methods to solve the problem. The first method, after sending the current fault alarm information and the target fault category to the target service server, may include: when receiving the fault feedback information returned by the target service server in response to the current fault alarm information, determining the detection result of the target fault category based on the fault feedback information; when the detection result indicates that the target fault category is incorrect, taking the service servers other than the target service server in the service server set as candidate service servers; sending the current fault alarm information to the candidate service servers so that the candidate service servers can perform fault troubleshooting based on the current fault alarm information.

[0149] For example, taking the aforementioned target service server as a VPN service server, the cloud server can send current fault alarm information to all candidate service servers (VPN service server and CVM service server) other than the target service server, requesting the VPN service server and CVM service server to simultaneously perform fault investigation on the current fault alarm information, thereby determining the actual fault category corresponding to the current fault alarm information.

[0150] The second solution, after determining the detection result of the target fault category based on the fault feedback information, may further include: when the detection result indicates that the target fault category is incorrect, selecting a substitute fault category corresponding to the current fault alarm information from the preset fault categories based on the posterior probability, wherein the posterior probability of the substitute fault category is only less than the posterior probability corresponding to the target fault category; and sending the current fault alarm information to the service server corresponding to the substitute fault category so that the service server corresponding to the substitute fault category can perform fault troubleshooting based on the current fault alarm information and the substitute fault category.

[0151] For example, the cloud server can reorder the preset fault categories based on the posterior probabilities calculated above, and select the preset fault category with the second largest posterior probability value as a substitute fault category. It then sends the current fault alarm information to the business server corresponding to this substitute fault category, requesting the business server to troubleshoot based on the current fault alarm information and the preset fault category. For instance, in the data example above, the probability of the current fault alarm information being of the "Support Service Unavailable" type is 0.0003024, ranking second in probability value from largest to smallest; therefore, "Support Service Unavailable" can be used as the substitute fault category. Since the business server corresponding to "Support Service Unavailable" is the CBS business server, the cloud server can route the current fault alarm information to the CBS business server so that the CBS business server can re-troubleshoot the current fault alarm information.

[0152] If the CBS service server determines after re-troubleshooting that the current fault alarm information is still outside its service processing scope, the cloud server can select the preset fault category corresponding to the third largest posterior probability value and send the current fault alarm information to the service server corresponding to the preset fault category, requesting the service server to troubleshoot based on the current fault alarm information and the preset fault category until the service server corresponding to the current fault alarm information is located.

[0153] In this embodiment of the application, after selecting the business servers other than the target business server in the business server set as candidate business servers, the method may further include: updating the preset range corresponding to the preset discrete interval based on the detection result to obtain the updated first discrete interval and the updated first range corresponding to the updated first discrete interval; updating the conditional probability set according to the updated first range to obtain the updated conditional probability set; and using the updated conditional probability set as the conditional probability set corresponding to the category of the historical running state parameter.

[0154] When the target service server's detection results after investigating the current fault alarm information indicate that the target fault category differs from the fault handling category corresponding to the target service server, it indicates that the cloud server's prediction of the target fault category has a certain deviation. This is because the target fault category is obtained by the cloud server sorting according to the posterior probabilities of each preset fault category. The calculation of the posterior probability depends on the conditional probability of the historical operating state parameter category under each preset fault category. As mentioned earlier, when calculating the conditional probability, the historical operating state parameters need to be divided into corresponding preset discrete intervals. The preset range of these discrete intervals determines the accuracy of the historical operating state parameter division. If the preset range is scientifically divided, the accuracy of the conditional probability calculated based on the historical fault alarm information in the pre-stored fault alarm information set will be high, and correspondingly, the accuracy of predicting the target fault category for each current fault alarm information will also be high.

[0155] In this embodiment, when the target service server's detection result after investigating the current fault alarm information indicates that the target fault category is different from the fault handling category corresponding to the target service server, the preset discrete interval of each historical operating status parameter category can be updated. For example, the number of preset discrete intervals can be adjusted, and the preset range of the adjusted preset discrete intervals can be adjusted to obtain the updated first discrete interval and the updated first range corresponding to the updated first discrete interval. Alternatively, the preset range of the preset discrete intervals can be directly adjusted to obtain the updated first discrete interval and the updated first range corresponding to the updated first discrete interval.

[0156] After obtaining the updated first discrete interval and the corresponding updated first range, the conditional probability of each historical operating state parameter category in the pre-stored fault alarm information set under each preset fault category can be recalculated based on the updated first discrete interval and the updated first range, according to the aforementioned conditional probability calculation method, to obtain the updated conditional probability set. This updated conditional probability set is then used as the conditional probability set corresponding to the categories of the historical operating state parameters, and the target fault category corresponding to each current fault alarm information is calculated based on this set.

[0157] As can be seen from the above, the embodiments of this application can obtain current fault alarm information of the current cloud environment, the current fault alarm information including at least one operating status parameter; obtain a pre-stored set of fault alarm information, the set of fault alarm information including one or more historical fault alarm information prior to the current cloud service; obtain a set of conditional probabilities and a set of prior probabilities corresponding to the historical fault alarm information, the set of conditional probabilities including the conditional probability corresponding to the category of each operating status parameter, the set of prior probabilities including the prior probability corresponding to each preset fault category; based on the category of the operating status parameter, filter out the conditional probability corresponding to the operating status parameter from the set of conditional probabilities; determine the posterior probability that the current fault alarm information belongs to each preset fault category according to the conditional probability and prior probability; based on the posterior probability, filter out the target fault category corresponding to the current fault alarm information in the preset fault categories, and send the current fault alarm information to the business server corresponding to the target fault category. This solution employs Bayesian algorithms to perform probabilistic analysis on detected current fault alarm information, determining the target fault category corresponding to the current fault alarm information. This improves the classification efficiency and accuracy of fault types corresponding to alarm information, enabling rapid fault type location. Once the target fault category is determined, the corresponding business server can be accurately and quickly launched for problem repair without having to launch all business servers for joint response and troubleshooting one by one. This reduces invalid responses, avoids excessive resource consumption, improves the response and processing efficiency for current fault alarm information, and enhances the availability of business services.

[0158] Based on the method described in the above embodiments, the following examples will provide further detailed explanations.

[0159] In this embodiment, the cloud environment fault handling device will be specifically integrated into an electronic device, and the electronic device will be a cloud server corresponding to the cloud resource provider, as an example for explanation.

[0160] Figure 8 This paper illustrates another flowchart of the cloud environment fault handling method provided in an embodiment of this application.

[0161] like Figure 8 As shown, a cloud environment fault handling method has the following specific process:

[0162] 201. The cloud server obtains the current fault alarm information of the current cloud environment. The current fault alarm information includes at least one running status parameter.

[0163] For example, a cloud server can monitor the current cloud environment during operation, and upon detecting a fault, generate a fault alarm. Similarly, business servers belonging to various business teams can also monitor the current cloud environment during operation, generate fault alarms upon detecting faults, and send these alarms to the cloud server, which can then receive the alarm information.

[0164] For example, the categories of operating status parameters can include fault range, switch packet forwarding rate, virtual private network slave machine traffic rate, master machine traffic rate, and so on.

[0165] 202. The cloud server obtains a pre-stored set of fault alarm information, which includes one or more historical fault alarm information prior to the current cloud service.

[0166] The pre-stored fault alarm information set can be understood as a set of sample data corresponding to historical fault alarm information generated before the current moment.

[0167] 203. The cloud server obtains the historical operating status parameters corresponding to the historical fault alarm information, as well as the preset fault categories corresponding to the historical operating status parameters.

[0168] 204. The cloud server determines the conditional probability set and the prior probability set based on historical operating status parameters and preset fault categories.

[0169] For example, a cloud server can count the number of historical fault alarm messages corresponding to each preset fault category in historical fault alarm information to obtain the category sample number; based on the category sample number, calculate the prior probability of each preset fault category to obtain the prior probability set; extract the mapping relationship between historical operating status parameters and preset fault categories from historical fault alarm information; and based on the mapping relationship, calculate at least one conditional probability of the category of historical operating status parameters under each preset fault category to obtain the conditional probability set.

[0170] For example, a cloud server can obtain the total number of historical fault alarm messages in a pre-stored fault alarm message set to get the total number of samples; and use the proportion of the number of category samples in the total number of samples as the prior probability to obtain the prior probability set.

[0171] For example, a cloud server can obtain at least two preset discrete intervals corresponding to the category of each historical operating status parameter, as well as a preset range corresponding to each preset discrete interval. The preset discrete intervals indicate the data distribution range corresponding to the historical operating status parameters. Based on the preset range, the historical operating status parameters contained in each preset discrete interval are determined to obtain the interval historical status parameters. Based on the mapping relationship and the interval historical status parameters, the interval conditional probability of each preset discrete interval corresponding to the category of the historical operating status parameter under the preset fault category is calculated. The interval conditional probability is used as the conditional probability to obtain the set of conditional probabilities corresponding to the category of the historical operating status parameter.

[0172] For example, a cloud server can take each preset fault category as the target fault category and the number of category samples corresponding to the target fault category as the target category sample quantity; take each preset discrete interval corresponding to the category of historical operating status parameters as the target first discrete interval, find the number of target historical status parameters that have a mapping relationship with the target fault category from the interval historical status parameters of the target first discrete interval, and obtain the target data volume; calculate the proportion of the target data volume in the target category sample quantity, and obtain the interval conditional probability of the target first discrete interval under the target fault category.

[0173] 205. Based on the category of the running status parameters, the cloud server selects the conditional probabilities corresponding to the running status parameters from the set of conditional probabilities.

[0174] For example, a cloud server can filter out the target conditional probability set corresponding to the category of each running state parameter from the conditional probability set. The target conditional probability set includes the conditional probabilities of at least two preset discrete intervals, and the preset discrete intervals correspond to a preset range. Based on the preset range, the conditional probability of the running state parameter under each preset fault category is extracted from the target conditional probability set.

[0175] For example, a cloud server can filter the target discrete interval corresponding to the running status parameter in the preset discrete interval under the category of the running status parameter based on a preset range; filter the conditional probability corresponding to the target discrete interval from the target conditional probability set to obtain the target interval conditional probability; and use the target interval conditional probability as the conditional probability of the running status parameter under the preset fault category.

[0176] 206. The cloud server determines the posterior probability of the current fault alarm information belonging to each preset fault category based on conditional probability and prior probability.

[0177] For example, a cloud server can use each preset fault category as a candidate fault category and determine at least two conditional probabilities of the current fault alarm information under the candidate fault category; fuse the conditional probabilities to obtain the target conditional probability corresponding to the candidate fault category; and fuse the prior probability of the candidate fault category with the target conditional probability to obtain the posterior probability of the candidate fault category.

[0178] 207. The cloud server uses posterior probability to filter out the target fault category corresponding to the current fault alarm information from the preset fault categories.

[0179] For example, the cloud server can sort the posterior probabilities of each preset fault category to obtain the posterior probability sorting result; based on the posterior probability sorting result, the target fault category corresponding to the current fault alarm information can be selected from the preset fault categories.

[0180] 208. The cloud server sends the current fault alarm information to the business server corresponding to the target fault category.

[0181] For example, a cloud server can filter out the business server corresponding to the target fault category from the set of business servers in the current cloud environment to obtain the target business server; and send the current fault alarm information and the target fault category to the target business server so that the target business server can perform fault troubleshooting based on the current fault alarm information and the target fault category.

[0182] As described above, this embodiment obtains the current fault alarm information of the current cloud environment through a cloud server. The current fault alarm information includes at least one operating status parameter. It also obtains a pre-stored set of fault alarm information, which includes one or more historical fault alarm information prior to the current cloud service. Furthermore, it obtains the historical operating status parameters corresponding to the historical fault alarm information, and the preset fault categories corresponding to the historical operating status parameters. Then, based on the historical operating status parameters and the preset fault categories, the cloud server determines a set of conditional probabilities and a set of prior probabilities. Next, based on the category of the operating status parameter, the cloud server filters out the conditional probabilities corresponding to the operating status parameter from the set of conditional probabilities. Then, based on the conditional probabilities and prior probabilities, the cloud server determines the posterior probability that the current fault alarm information belongs to each preset fault category. Next, based on the posterior probability, the cloud server filters out the target fault category corresponding to the current fault alarm information from the preset fault categories. Finally, the cloud server sends the current fault alarm information to the business server corresponding to the target fault category. Because the cloud server in this solution can use Bayesian algorithm to perform probability analysis on the detected current fault alarm information and determine the target fault category corresponding to the current fault alarm information, after determining the target fault category, the corresponding business server can be accurately and quickly launched to repair the problem, without having to launch all business servers for joint response and troubleshooting one by one. This can reduce the fault category location time, reduce invalid responses, and improve the efficiency of alarm information processing and the availability of business services.

[0183] To better implement the above methods, this application also provides a cloud environment fault handling device, which can be integrated into network devices, such as servers or terminals. The server can be a cloud server, and the terminal can include tablet computers, laptops, and / or personal computers.

[0184] Figure 9 A schematic diagram of the cloud environment fault handling device provided in an embodiment of this application is shown. Figure 9 As shown, the cloud environment fault handling device may include a first acquisition unit 301, a second acquisition unit 302, a third acquisition unit 303, a filtering unit 304, a determining unit 305, and a sending unit 306, as follows:

[0185] (1) First acquisition unit 301;

[0186] The first acquisition unit 301 is used to acquire the current fault alarm information of the current cloud environment, which includes at least one operating status parameter.

[0187] (2) Second acquisition unit 302;

[0188] The second acquisition unit 302 is used to acquire a pre-stored set of fault alarm information, which includes one or more historical fault alarm information from the current cloud environment.

[0189] (3) Third acquisition unit 303;

[0190] The third acquisition unit 303 is used to acquire the conditional probability set and prior probability set corresponding to historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each operating state parameter, and the prior probability set includes the prior probability corresponding to each preset fault category.

[0191] For example, the third acquisition unit 303 can be used to acquire the historical operating status parameters corresponding to the historical fault alarm information, and the preset fault category corresponding to the historical operating status parameters; and to determine the conditional probability set and the prior probability set based on the historical operating status parameters and the preset fault category.

[0192] For example, the third acquisition unit 303 can be used to count the number of historical fault alarm messages corresponding to each preset fault category in the historical fault alarm information to obtain the category sample number; based on the category sample number, calculate the prior probability of each preset fault category to obtain the prior probability set; extract the mapping relationship between historical operating status parameters and preset fault categories from the historical fault alarm information; and based on the mapping relationship, calculate at least one conditional probability of the category of the historical operating status parameter under each preset fault category to obtain the conditional probability set.

[0193] For example, the third acquisition unit 303 can be used to acquire the total number of historical fault alarm information in the pre-stored fault alarm information set to obtain the total number of samples; and use the proportion of the number of category samples in the total number of samples as the prior probability to obtain the prior probability set.

[0194] For example, the third acquisition unit 303 can be specifically used to acquire at least two preset discrete intervals corresponding to the category of each historical operating state parameter, and a preset range corresponding to each preset discrete interval. The preset discrete intervals indicate the data distribution intervals corresponding to the historical operating state parameters. Based on the preset range, the historical operating state parameters contained in each preset discrete interval are determined to obtain the interval historical state parameters. Based on the mapping relationship and the interval historical state parameters, the interval conditional probability of each preset discrete interval corresponding to the category of the historical operating state parameter under the preset fault category is calculated. The interval conditional probability is used as the conditional probability to obtain the set of conditional probabilities corresponding to the category of the historical operating state parameter.

[0195] For example, the third acquisition unit 303 can be specifically used to take each preset fault category as the target fault category and take the number of category samples corresponding to the target fault category as the number of target category samples; take each preset discrete interval corresponding to the category of historical operating state parameters as the target first discrete interval, find the number of target historical state parameters that have a mapping relationship with the target fault category from the interval historical state parameters of the target first discrete interval, and obtain the target data volume; calculate the proportion of the target data volume in the number of target category samples, and obtain the interval conditional probability of the target first discrete interval under the target fault category.

[0196] (4) Filtering unit 304;

[0197] The filtering unit 304 is used to filter out the conditional probabilities corresponding to the running state parameters from the set of conditional probabilities based on the category of the running state parameters.

[0198] For example, the filtering unit 304 can be used to filter out the target conditional probability set corresponding to the category of each operating state parameter from the conditional probability set. The target conditional probability set includes the conditional probabilities of at least two preset discrete intervals, and the preset discrete intervals correspond to a preset range. Based on the preset range, the conditional probability of the operating state parameter under each preset fault category is extracted from the target conditional probability set.

[0199] For example, the filtering unit 304 can be used to filter the target discrete interval corresponding to the operating state parameter in the preset discrete interval under the category of the operating state parameter based on the preset range; filter the conditional probability corresponding to the target discrete interval from the target conditional probability set to obtain the conditional probability of the target interval; and use the conditional probability of the target interval as the conditional probability of the operating state parameter under the preset fault category.

[0200] (5) Determine unit 305;

[0201] The determining unit 305 is used to determine the posterior probability that the current fault alarm information belongs to each preset fault category based on the conditional probability and the prior probability.

[0202] For example, the determining unit 305 can be used to take each preset fault category as a candidate fault category and determine at least two conditional probabilities of the current fault alarm information under the candidate fault category; fuse the conditional probabilities to obtain the target conditional probability corresponding to the candidate fault category; and fuse the prior probability of the candidate fault category with the target conditional probability to obtain the posterior probability of the candidate fault category.

[0203] (6) Transmitting unit 306;

[0204] The sending unit 306 is used to filter out the target fault category corresponding to the current fault alarm information from the preset fault categories based on the posterior probability, and send the current fault alarm information to the service server corresponding to the target fault category.

[0205] For example, the sending unit 306 can be used to sort the posterior probabilities of each preset fault category to obtain the posterior probability sorting result; based on the posterior probability sorting result, the target fault category corresponding to the current fault alarm information is selected from the preset fault categories.

[0206] For example, the sending unit 306 can be used to filter out the business server corresponding to the target fault category from the set of business servers corresponding to the current cloud environment, and obtain the target business server; send the current fault alarm information and the target fault category to the target business server so that the target business server can perform fault troubleshooting based on the current fault alarm information and the target fault category.

[0207] For example, the sending unit 306 can be specifically used to determine the detection result of the target fault category based on the fault feedback information when it receives the fault feedback information returned by the target service server in response to the current fault alarm information; when the detection result indicates that the target fault category is incorrect, it can take the service servers other than the target service server in the service server set as candidate service servers; and send the current fault alarm information to the candidate service servers so that the candidate service servers can perform fault troubleshooting based on the current fault alarm information.

[0208] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0209] As described above, this embodiment acquires current fault alarm information of the current cloud environment through the first acquisition unit 301, which includes at least one operating status parameter; and acquires a pre-stored fault alarm information set through the second acquisition unit 302, which includes one or more historical fault alarm information from the current cloud environment; acquires a conditional probability set and a prior probability set corresponding to the historical fault alarm information through the third acquisition unit 303, where the conditional probability set includes the conditional probability corresponding to the category of each operating status parameter, and the prior probability set includes the prior probability corresponding to each preset fault category; then, the filtering unit 304 filters out the conditional probability corresponding to the operating status parameter from the conditional probability set based on the category of the operating status parameter; then, the determining unit 305 determines the posterior probability that the current fault alarm information belongs to each preset fault category based on the conditional probability and the prior probability; finally, the sending unit 306 filters out the target fault category corresponding to the current fault alarm information from the preset fault categories based on the posterior probability, and sends the current fault alarm information to the business server corresponding to the target fault category. This solution employs Bayesian algorithms to perform probabilistic analysis on detected current fault alarm information, determining the target fault category corresponding to the current fault alarm information. This improves the classification efficiency and accuracy of fault types corresponding to alarm information, enabling rapid fault type location. Once the target fault category is determined, the corresponding business server can be accurately and quickly launched for problem repair without having to launch all business servers for joint response and troubleshooting one by one. This reduces invalid responses, avoids excessive resource consumption, improves the response and processing efficiency for current fault alarm information, and enhances the availability of business services.

[0210] This application also provides an electronic device, such as... Figure 10 As shown, it illustrates a structural schematic diagram of the electronic device involved in the embodiments of this application, specifically:

[0211] The electronic device may include components such as a processor 401 with one or more processing cores, a memory 402 with one or more computer-readable storage media, a power supply 403, and an input unit 404. Those skilled in the art will understand that... Figure 10 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein:

[0212] The processor 401 is the control center of the electronic device, connecting various parts of the device via various interfaces and lines. It executes software programs and / or modules stored in the memory 402, and calls data stored in the memory 402, to perform various functions and process data. Optionally, the processor 401 may include one or more processing cores; preferably, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 401.

[0213] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 402 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0214] The electronic device also includes a power supply 403 that supplies power to the various components. Preferably, the power supply 403 can be logically connected to the processor 401 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 403 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0215] The electronic device may also include an input unit 404, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0216] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 401 in the electronic device loads the executable files corresponding to the processes of one or more applications into the memory 402 according to the following instructions, and the processor 401 runs the applications stored in the memory 402 to realize various functions, as follows:

[0217] The system retrieves current fault alarm information from the current cloud environment, including at least one operational status parameter; retrieves a pre-stored set of fault alarm information, including one or more historical fault alarms prior to the current cloud service; retrieves a set of conditional probabilities and a set of prior probabilities corresponding to the historical fault alarm information, where the conditional probability set includes the conditional probability corresponding to the category of each operational status parameter, and the prior probability set includes the prior probability corresponding to each preset fault category; based on the category of the operational status parameter, it filters out the conditional probabilities corresponding to the operational status parameter from the set of conditional probabilities; based on the conditional probabilities and prior probabilities, it determines the posterior probability that the current fault alarm information belongs to each preset fault category; based on the posterior probability, it filters out the target fault category corresponding to the current fault alarm information from the preset fault categories and sends the current fault alarm information to the business server corresponding to the target fault category.

[0218] For example, an electronic device can obtain current fault alarm information of the current cloud environment, which includes at least one operating status parameter; and obtain a pre-stored set of fault alarm information, which includes one or more historical fault alarms prior to the current cloud service; then obtain the historical operating status parameters corresponding to the historical fault alarm information, and the preset fault categories corresponding to the historical operating status parameters; then, based on the historical operating status parameters and the preset fault categories, determine a set of conditional probabilities and a set of prior probabilities; then, based on the category of the operating status parameter, filter out the conditional probabilities corresponding to the operating status parameter from the set of conditional probabilities; then, based on the conditional probabilities and prior probabilities, determine the posterior probability that the current fault alarm information belongs to each preset fault category; then, based on the posterior probabilities, filter out the target fault category corresponding to the current fault alarm information from the preset fault categories; finally, send the current fault alarm information to the business server corresponding to the target fault category.

[0219] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0220] As described above, this application embodiment obtains current fault alarm information of the current cloud environment, which includes at least one operating status parameter, and obtains a pre-stored set of fault alarm information, which includes one or more historical fault alarm information prior to the current cloud service. It also obtains a conditional probability set and a priori probability set corresponding to the historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each operating status parameter, and the priori probability set includes the priori probability corresponding to each preset fault category. In the above process, the obtained conditional probability set and priori probability set are used as a basis so that the conditional probability corresponding to the operating status parameter can be filtered from the conditional probability set based on the category of the operating status parameter. After determining the conditional probability corresponding to the operating status parameter, the process can continue... Based on conditional and prior probabilities, a probabilistic analysis is performed on the current fault alarm information to determine the posterior probability of the current fault alarm information belonging to each preset fault category. Then, based on the posterior probability, the target fault category corresponding to the current fault alarm information is selected from the preset fault categories. This can improve the classification efficiency and accuracy of the fault type corresponding to the current fault alarm information, and quickly locate the fault type. After determining the target fault category, the current fault alarm information can be sent to the business server corresponding to the target fault category, so that the corresponding business server can be accurately and quickly launched to fix the problem, without having to launch all business servers for joint response and troubleshooting one by one. This reduces invalid responses, avoids excessive resource consumption, improves the response and processing efficiency of the current fault alarm information, and improves the availability of business services.

[0221] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0222] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute steps in any of the cloud environment fault handling methods provided in embodiments of this application. For example, the instructions can execute the following steps:

[0223] The system retrieves current fault alarm information from the current cloud environment, including at least one operational status parameter; retrieves a pre-stored set of fault alarm information, including one or more historical fault alarms prior to the current cloud service; retrieves a set of conditional probabilities and a set of prior probabilities corresponding to the historical fault alarm information, where the conditional probability set includes the conditional probability corresponding to the category of each operational status parameter, and the prior probability set includes the prior probability corresponding to each preset fault category; based on the category of the operational status parameter, it filters out the conditional probabilities corresponding to the operational status parameter from the set of conditional probabilities; based on the conditional probabilities and prior probabilities, it determines the posterior probability that the current fault alarm information belongs to each preset fault category; based on the posterior probability, it filters out the target fault category corresponding to the current fault alarm information from the preset fault categories and sends the current fault alarm information to the business server corresponding to the target fault category.

[0224] For example, the system acquires current fault alarm information of the current cloud environment, which includes at least one operating status parameter; it also acquires a pre-stored set of fault alarm information, which includes one or more historical fault alarms prior to the current cloud service; it then acquires the historical operating status parameters corresponding to the historical fault alarm information, and the preset fault categories corresponding to the historical operating status parameters; based on the historical operating status parameters and the preset fault categories, it determines a set of conditional probabilities and a set of prior probabilities; then, based on the category of the operating status parameter, it filters out the conditional probabilities corresponding to the operating status parameter from the set of conditional probabilities; then, based on the conditional probabilities and prior probabilities, it determines the posterior probability that the current fault alarm information belongs to each preset fault category; then, based on the posterior probabilities, it filters out the target fault category corresponding to the current fault alarm information from the preset fault categories; finally, it sends the current fault alarm information to the business server corresponding to the target fault category.

[0225] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0226] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0227] Since the instructions stored in the computer-readable storage medium can execute the steps in any of the cloud environment fault handling methods provided in the embodiments of this application, the beneficial effects that any of the cloud environment fault handling methods provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0228] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the methods provided in the various alternative implementations of the above-described alarm information processing.

[0229] The above provides a detailed description of a cloud environment fault handling method, apparatus, and computer-readable storage medium provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A cloud environment fault handling method, characterized in that, include: Obtain the current fault alarm information of the current cloud environment, wherein the current fault alarm information includes at least one operating status parameter; Obtain a pre-stored set of fault alarm information, wherein the set of fault alarm information includes one or more historical fault alarm information prior to the current cloud environment; Obtain the conditional probability set and prior probability set corresponding to the historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each of the operating state parameters, and the prior probability set includes the prior probability corresponding to each preset fault category. Based on the category of the running state parameter, the conditional probability corresponding to the running state parameter is selected from the set of conditional probabilities; Based on the conditional probability and the prior probability, determine the posterior probability that the current fault alarm information belongs to each of the preset fault categories; Based on the posterior probability, the target fault category corresponding to the current fault alarm information is selected from the preset fault categories, and the current fault alarm information is sent to the service server corresponding to the target fault category.

2. The cloud environment fault handling method as described in claim 1, characterized in that, The step of obtaining the conditional probability set and prior probability set corresponding to the historical fault alarm information includes: Obtain the historical operating status parameters corresponding to the historical fault alarm information, and the preset fault category corresponding to the historical operating status parameters; Based on the historical operating status parameters and the preset fault category, the conditional probability set and the prior probability set are determined.

3. The cloud environment fault handling method as described in claim 2, characterized in that, The step of determining the conditional probability set and the prior probability set based on the historical operating state parameters and the preset fault category includes: The number of historical fault alarm messages corresponding to each preset fault category is counted in the historical fault alarm information to obtain the category sample number; Based on the number of samples in the aforementioned categories, the prior probability of each of the preset fault categories is calculated to obtain the prior probability set. Extract the mapping relationship between the historical operating status parameters and the preset fault categories from the historical fault alarm information; Based on the mapping relationship, calculate at least one conditional probability of the category of the historical operating state parameter under each preset fault category to obtain the conditional probability set.

4. The cloud environment fault handling method as described in claim 3, characterized in that, The prior probability set is obtained by calculating the prior probability set for each preset fault category based on the number of samples in the category, including: Obtain the total number of historical fault alarm messages in the pre-stored fault alarm information set to get the total number of samples; The proportion of the number of samples in the category to the total number of samples is used as the prior probability to obtain the prior probability set.

5. The cloud environment fault handling method as described in claim 3, characterized in that, Based on the mapping relationship, the system calculates at least one conditional probability of the category of the historical operating state parameter under each preset fault category, thereby obtaining the set of conditional probabilities, including: Obtain at least two preset discrete intervals corresponding to the category of each historical operating state parameter, and a preset range corresponding to each preset discrete interval, wherein the preset discrete intervals indicate the data distribution intervals corresponding to the historical operating state parameters; Based on the preset range, the historical operating state parameters contained in each preset discrete interval are determined to obtain the interval historical state parameters; Based on the mapping relationship and the interval historical state parameters, calculate the interval conditional probability of each preset discrete interval corresponding to the category of the historical operating state parameters under the preset fault category; The interval conditional probability is used as the conditional probability to obtain the set of conditional probabilities corresponding to the categories of the historical operating state parameters.

6. The cloud environment fault handling method as described in claim 5, characterized in that, The step of calculating the interval conditional probability of each preset discrete interval under the preset fault category corresponding to the category of the historical operating state parameter, based on the mapping relationship and the interval historical state parameters, includes: Each of the preset fault categories is taken as the target fault category, and the number of category samples corresponding to the target fault category is taken as the target category sample number; Each preset discrete interval corresponding to the category of the historical operating state parameter is taken as the target first discrete interval. The number of target historical state parameters that have the mapping relationship with the target fault category is found from the interval historical state parameters of the target first discrete interval to obtain the target data volume. Calculate the proportion of the target data volume in the number of samples of the target category to obtain the interval conditional probability of the first discrete interval of the target under the target fault category.

7. The cloud environment fault handling method as described in claim 1, characterized in that, The step of filtering the conditional probabilities corresponding to the operating state parameters from the set of conditional probabilities based on the category of the operating state parameters includes: Filter out the target conditional probability set corresponding to the category of each of the operating state parameters from the conditional probability set. The target conditional probability set includes conditional probabilities of at least two preset discrete intervals, and the preset discrete intervals correspond to a preset range. Based on the preset range, the conditional probability of the operating state parameter under each preset fault category is extracted from the target conditional probability set.

8. The cloud environment fault handling method as described in claim 7, characterized in that, The target conditional probability set also includes the conditional probability of each preset discrete interval under each preset fault category. The step of extracting the conditional probability of the operating state parameter under each preset fault category from the target conditional probability set based on the preset range includes: Based on the preset range, the target discrete interval corresponding to the operating state parameter is selected from the preset discrete interval under the category of the operating state parameter; The conditional probabilities corresponding to the target discrete intervals are selected from the target conditional probability set to obtain the target interval conditional probabilities; The conditional probability of the target interval is used as the conditional probability of the operating state parameter under the preset fault category.

9. The cloud environment fault handling method as described in claim 1, characterized in that, Determining the posterior probability that the current fault alarm information belongs to each of the preset fault categories based on the conditional probability and the prior probability includes: Each of the preset fault categories is used as a candidate fault category, and at least two conditional probabilities of the current fault alarm information under the candidate fault category are determined. The conditional probabilities are fused to obtain the target conditional probability corresponding to the candidate fault category; The prior probability of the candidate fault category is fused with the target conditional probability to obtain the posterior probability of the candidate fault category.

10. The cloud environment fault handling method as described in claim 1, characterized in that, The step of filtering the target fault category corresponding to the current fault alarm information from the preset fault categories based on the posterior probability includes: The posterior probabilities of each preset fault category are sorted to obtain the posterior probability sorting result; Based on the posterior probability ranking result, the target fault category corresponding to the current fault alarm information is selected from the preset fault categories.

11. The cloud environment fault handling method as described in claim 1, characterized in that, Sending the current fault alarm information to the service server corresponding to the target fault category includes: The target business server is obtained by filtering out the business server corresponding to the target fault category from the set of business servers corresponding to the current cloud environment; The current fault alarm information and the target fault category are sent to the target service server so that the target service server can perform fault troubleshooting based on the current fault alarm information and the target fault category.

12. The cloud environment fault handling method as described in claim 11, characterized in that, After sending the current fault alarm information and the target fault category to the target service server, the method further includes: When the target service server returns fault feedback information in response to the current fault alarm information, the detection result of the target fault category is determined based on the fault feedback information. When the detection result indicates that the target fault category is incorrect, the business servers in the set of business servers other than the target business server are selected as candidate business servers. The current fault alarm information is sent to the candidate service server so that the candidate service server can perform fault troubleshooting based on the current fault alarm information.

13. The cloud environment fault handling method as described in claim 12, characterized in that, After determining the detection result of the target fault category based on the fault feedback information, the method further includes: When the detection result indicates that the target fault category is incorrect, based on the posterior probability, a substitute fault category corresponding to the current fault alarm information is selected from the preset fault categories. The posterior probability of the substitute fault category is less than the posterior probability corresponding to the target fault category. The current fault alarm information is sent to the service server corresponding to the alternative fault category, so that the service server corresponding to the alternative fault category can perform fault troubleshooting based on the current fault alarm information and the alternative fault category.

14. The cloud environment fault handling method as described in claim 12 or 13, characterized in that, After selecting all service servers in the set of service servers other than the target service server as candidate service servers, the method further includes: Based on the detection results, the preset range corresponding to the preset discrete interval is updated to obtain the updated first discrete interval and the updated first range corresponding to the updated first discrete interval. Based on the updated first range, the conditional probability set is updated to obtain the updated conditional probability set; The updated conditional probability set is used as the conditional probability set corresponding to the category of the historical operating state parameter.

15. A cloud environment fault handling device, characterized in that, include: The first acquisition unit is used to acquire the current fault alarm information of the current cloud environment, wherein the current fault alarm information includes at least one operating status parameter; The second acquisition unit is used to acquire a pre-stored set of fault alarm information, which includes one or more historical fault alarm information prior to the current cloud environment. The third acquisition unit is used to acquire the conditional probability set and the prior probability set corresponding to the historical fault alarm information. The conditional probability set includes the conditional probability corresponding to the category of each of the operating state parameters, and the prior probability set includes the prior probability corresponding to each preset fault category. A filtering unit is used to filter out the conditional probabilities corresponding to the running state parameters from the set of conditional probabilities based on the category of the running state parameters. The determining unit is configured to determine the posterior probability that the current fault alarm information belongs to each of the preset fault categories based on the conditional probability and the prior probability. The sending unit is configured to, based on the posterior probability, filter out the target fault category corresponding to the current fault alarm information from the preset fault categories, and send the current fault alarm information to the service server corresponding to the target fault category.

16. An electronic device, characterized in that, It includes a processor and a memory, the memory storing an application program, and the processor running the application program within the memory to perform the steps in the cloud environment fault handling method for the terminal application according to any one of claims 1 to 14.

17. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps in the cloud environment fault handling method according to any one of claims 1 to 14.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to perform the steps of the cloud environment fault handling method according to any one of claims 1 to 14.

Images