Cloud platform monitoring method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a cloud platform monitoring method and device, computer equipment, a readable storage medium and a program product. The method comprises the steps that data processing is conducted on historical data of multiple cloud platforms, a first sample data set is obtained, and the first sample data set comprises multiple pieces of sample data of different data types; obtaining a positive sample pair and a negative sample pair according to the sample data of the text type and the sample data of the numerical value type in the first sample data set, and training an encoder in the to-be-trained model according to the positive sample pair and the negative sample pair to obtain a target encoder; according to the sample data of the numerical value type and the sample data of the alarm type in the first sample data set, training a decoder in the to-be-trained model to obtain a target decoder; obtaining a target monitoring model according to the target encoder and the target decoder; and monitoring the plurality of cloud platforms based on the target monitoring model to obtain a monitoring result of each cloud platform. By adopting the method, monitoring of a plurality of cloud platforms can be realized.

Description

Technical Field

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

[0002] Currently, the use of multi-cloud platforms has become an important strategy for enterprises, because using multiple cloud service providers makes it easier to leverage the advantages of different cloud platforms, optimize costs, and enhance business flexibility. Backing up data and applications on multiple cloud platforms can enhance the redundancy and disaster recovery capabilities of the system.

[0003] However, this multi-cloud environment also brings complex management challenges. Therefore, how to achieve monitoring of multiple cloud platforms has become an urgent problem that needs to be solved. Summary of the invention

[0004] Based on this, it is necessary to provide a cloud platform monitoring method, device, computer equipment, readable storage medium and program product that can realize monitoring of multiple cloud platforms in response to the above technical problems.

[0005] In a first aspect, the present application provides a cloud platform monitoring method, the method comprising: processing historical data of multiple cloud platforms to obtain a first sample data set, the first sample data set including sample data of multiple different data types; obtaining positive sample pairs and negative sample pairs based on the sample data of text type and the sample data of numerical type in the first sample data set, and training an encoder in a training model based on the positive sample pairs and the negative sample pairs to obtain a target encoder; training a decoder in a training model based on the sample data of numerical type and the sample data of alarm type in the first sample data set to obtain a target decoder; obtaining a target monitoring model based on the target encoder and the target decoder; and monitoring multiple cloud platforms based on the target monitoring model to obtain monitoring results for each cloud platform.

[0006] In one embodiment, historical data includes target data, log data and security data, and the target data includes resource usage data and / or application performance data; data processing is performed on the historical data of multiple cloud platforms to obtain a first sample data set, including: superimposing different types of target data corresponding to the same timestamp to obtain sample data of a numerical type corresponding to each timestamp; vectorizing the log data of each timestamp to obtain sample data of a text type corresponding to each timestamp; vectorizing the security data of each timestamp to obtain sample data of an alarm type corresponding to each timestamp; and obtaining a first sample data set based on the sample data of the numerical type, the sample data of the text type and the sample data of the alarm type corresponding to each timestamp.

[0007] In one embodiment, the decoder includes a first decoder, and the decoder in the training model is trained according to sample data of numerical type and sample data of alarm type in a first sample data set to obtain a target decoder, including: inputting the sample data of numerical type into a target encoder to obtain a first output result output by the target encoder, and inputting the first output result into a first decoder to obtain alarm data output by the first decoder; obtaining an alarm loss value according to the alarm data and the sample data of the alarm type, and training the first decoder based on the alarm loss value to obtain a first target decoder; and obtaining a target decoder according to the first target decoder.

[0008] In one embodiment, the decoder also includes a second decoder, and the first sample data set includes sample data of numerical types corresponding to multiple timestamps respectively. According to the first target decoder, a target decoder is obtained, including: splitting the sample data of multiple numerical types into first sample data and second sample data; inputting the first sample data into a target encoder to obtain a second output result output by the target encoder, and inputting the second output result into a second decoder to obtain target prediction data output by the second decoder, the target prediction data including resource demand prediction data and / or application performance prediction data; obtaining a target prediction loss value according to the target prediction data and the second sample data, and training the second decoder based on the target prediction loss value to obtain a second target decoder; obtaining a target decoder according to the first target decoder and the second target decoder.

[0009] In one embodiment, a target monitoring model is obtained according to a target encoder and a target decoder, including: obtaining a monitoring model according to the target encoder, the target decoder and the output layer; replacing the output layer in the monitoring model with a linear layer to obtain a target reward model; initializing the monitoring model, and inputting the sample data of the numerical type into the initialized monitoring model to obtain an output result of the initialized monitoring model; inputting the sample data of the numerical type input into the initialized monitoring model and the output result of the initialized monitoring model into the target reward model to obtain a reward value output by the target reward model; adjusting the parameters of the initialized monitoring model based on the reward value output by the target reward model to obtain the target monitoring model.

[0010] In one embodiment, the output layer in the monitoring model is replaced by a linear layer to obtain a target reward model, including: replacing the output layer in the monitoring model with a linear layer to obtain a reward model; inputting numerical sample data into the monitoring model K times to obtain 2K output results output by the monitoring model, wherein the 2K output results include K alarm data and K target prediction data; obtaining a second sample data set based on the numerical sample data input into the monitoring model and the 2K output results; and training the reward model based on the second sample data set to obtain a target reward model.

[0011] In a second aspect, the present application also provides a cloud platform monitoring device, the device comprising:

[0012] A processing module, used for processing historical data of multiple cloud platforms to obtain a first sample data set, wherein the first sample data set includes sample data of multiple different data types;

[0013] A first training module is used to obtain positive sample pairs and negative sample pairs according to the sample data of the text type and the sample data of the numerical type in the first sample data set, and train the encoder in the training model according to the positive sample pairs and the negative sample pairs to obtain a target encoder;

[0014] A second training module is used to train the decoder in the training model according to the sample data of the numerical type and the sample data of the alarm type in the first sample data set to obtain a target decoder;

[0015] A determination module, used for obtaining a target monitoring model according to a target encoder and a target decoder;

[0016] The monitoring module is used to monitor multiple cloud platforms based on the target monitoring model and obtain the monitoring results of each cloud platform.

[0017] In one embodiment, historical data includes target data, log data and security data, and the target data includes resource usage data and / or application performance data; a processing module is specifically used to superimpose different types of target data corresponding to the same timestamp to obtain sample data of a numerical type corresponding to each timestamp; vectorize the log data of each timestamp to obtain sample data of a text type corresponding to each timestamp; vectorize the security data of each timestamp to obtain sample data of an alarm type corresponding to each timestamp; and obtain a first sample data set based on the sample data of the numerical type, the sample data of the text type and the sample data of the alarm type corresponding to each timestamp.

[0018] In one embodiment, the decoder includes a first decoder and a second training module, which are specifically used to input sample data of a numerical type into a target encoder to obtain a first output result output by the target encoder, and input the first output result into a first decoder to obtain alarm data output by the first decoder; obtain an alarm loss value based on the alarm data and sample data of the alarm type, train the first decoder based on the alarm loss value to obtain a first target decoder; and obtain a target decoder based on the first target decoder.

[0019] In one embodiment, the decoder also includes a second decoder, the first sample data set includes sample data of multiple numerical types corresponding to multiple timestamps, and a second training module is specifically used to split the sample data of multiple numerical types into first sample data and second sample data; the first sample data is input into a target encoder to obtain a second output result output by the target encoder, and the second output result is input into a second decoder to obtain target prediction data output by the second decoder, the target prediction data including resource demand prediction data and / or application performance prediction data; according to the target prediction data and the second sample data, a target prediction loss value is obtained, and the second decoder is trained based on the target prediction loss value to obtain a second target decoder; according to the first target decoder and the second target decoder, a target decoder is obtained.

[0020] In one embodiment, the determination module is specifically used to obtain a monitoring model based on the target encoder, the target decoder and the output layer; replace the output layer in the monitoring model with a linear layer to obtain a target reward model; initialize the monitoring model, and input the sample data of the numerical type into the initialized monitoring model to obtain an output result of the initialized monitoring model; input the sample data of the numerical type input into the initialized monitoring model and the output result of the initialized monitoring model into the target reward model to obtain a reward value output by the target reward model; adjust the parameters of the initialized monitoring model based on the reward value output by the target reward model to obtain the target monitoring model.

[0021] In one embodiment, a determination module is specifically used to replace the output layer in the monitoring model with a linear layer to obtain a reward model; input numerical sample data into the monitoring model K times to obtain 2K output results output by the monitoring model, wherein the 2K output results include K alarm data and K target prediction data; obtain a second sample data set based on the numerical sample data input into the monitoring model and the 2K output results; and train the reward model based on the second sample data set to obtain a target reward model.

[0022] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.

[0023] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any one of the methods described in the first aspect are implemented.

[0024] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect.

[0025] The cloud platform monitoring method, device, computer equipment, readable storage medium and program product obtain sample data including multiple different data types by processing the historical data of multiple cloud platforms. Then, according to the sample data of text type and the sample data of numerical type in the first sample data set, positive sample pairs and negative sample pairs are obtained, and according to the positive sample pairs and negative sample pairs, the encoder in the training model is trained to obtain the target encoder; according to the sample data of numerical type and the sample data of alarm type in the first sample data set, the decoder in the training model is trained to obtain the target decoder; according to the target encoder and the target decoder, the target monitoring model is obtained. Then, based on the target monitoring model, multiple cloud platforms are monitored to obtain the monitoring results of each cloud platform. In this way, the target monitoring model obtained by training the historical data of multiple cloud platforms monitors multiple cloud platforms, which not only realizes the monitoring of multiple cloud platforms, but also improves the management efficiency and reliability of the multi-cloud environment compared to the method of realizing cloud platform monitoring by manually analyzing cloud platform data. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 An application environment diagram of a cloud platform monitoring method in an embodiment;

[0028] Figure 2 A schematic diagram of a cloud platform monitoring method in one embodiment;

[0029] Figure 3 is a schematic diagram of an encoder training process in one embodiment;

[0030] Figure 4 is a schematic diagram of a decoder training process in one embodiment;

[0031] Figure 5 A schematic diagram of a monitoring model reinforcement training process in one embodiment;

[0032] Figure 6 It is a structural block diagram of a cloud platform monitoring device in one embodiment;

[0033] Figure 7 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0035] Before introducing the cloud platform monitoring method provided by this application, some technical terms involved in this application are introduced:

[0036] PPO (Proximal Policy Optimization): is an optimization algorithm for reinforcement learning. It is mainly used to train intelligent agents to make decisions in the environment. The goal is to maximize long-term returns. The core idea is to update the strategy by optimizing the objective function while controlling the update amplitude to ensure the stability of training.

[0037] RLHF (Reinforcement Learning with Human Feedback): is a machine learning method that combines reinforcement learning and human feedback. With this approach, the model not only relies on reward signals from the environment, but is also able to use human judgment to adjust its strategy. This is particularly useful in some tasks, such as natural language processing and complex decision-making systems, where human intuition and values ​​can help the model better understand the task and optimize its performance.

[0038] Transformer: It is a deep learning model architecture based on the attention mechanism. It uses the self-attention mechanism to perform correlation calculations on each position in the input sequence. It is widely used in natural language processing and sequence tasks. Its core innovation is the introduction of the self-attention mechanism, which enables the model to focus on information at different positions when processing the input sequence, thereby capturing contextual relationships. Transformer usually consists of two parts: an encoder and a decoder. The encoder converts the input sequence into a feature representation, and the decoder uses this representation to generate the target output. The multi-head attention mechanism allows the model to process multiple self-attention mechanisms in parallel, enhancing the ability to extract information. In order to solve the problem of missing position information in sequence data, Transformer introduces position encoding to ensure that the model can understand the order of elements in the sequence. In addition, the design of layer normalization and residual connection makes the training of deep networks more stable. Compared with traditional recurrent neural networks, Transformer has stronger parallel processing capabilities and can efficiently train and process long-distance dependencies. Due to these advantages, the Transformer architecture has achieved remarkable results in various natural language processing tasks, and has also been extended to other fields such as image processing and time series data prediction, becoming an important milestone in the field of deep learning.

[0039] Currently, the use of multi-cloud platforms has become an important strategy for enterprises, because using multiple cloud service providers makes it easier to take advantage of the advantages of different cloud platforms, optimize costs, and enhance business flexibility. However, this multi-cloud environment also brings complex management challenges, especially in terms of resource monitoring, security management, and performance optimization. Among them, multi-cloud platforms refer to enterprises or organizations using multiple different cloud service providers to deploy and manage their computing resources and services at the same time. This strategy is usually used to: reduce dependence on a single supplier by distributing resources among different cloud platforms; select the most cost-effective solution based on the price and service characteristics of different cloud service providers; back up data and applications in multiple cloud platforms to enhance the redundancy and disaster recovery capabilities of the system.

[0040] Existing traditional monitoring technologies usually monitor resources on a single cloud platform. However, with the advancement of enterprise multi-cloud strategies, these traditional monitoring tools and methods face challenges in multi-cloud environments because they usually lack cross-platform integration capabilities and are difficult to uniformly manage data and indicators on different cloud platforms. Therefore, how to achieve monitoring of multiple cloud platforms has become an urgent problem to be solved.

[0041] Based on this, it is necessary to propose effective technical means to solve the above problems. The following is a detailed description of the technical solution of the present application and how the technical solution of the present application solves the above technical problems with specific embodiments.

[0042] The cloud platform monitoring method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 12 communicates with the server 14 through the network. The data storage system can store the data that the server 14 needs to process. The data storage system can be integrated on the server 14, or it can be placed on the cloud or other network servers. Among them, the terminal 12 can be, but is not limited to, various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 14 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0043] In an exemplary embodiment, Figure 2 As shown, a cloud platform monitoring method is provided, which is applied to Figure 1 The server in the example is used to illustrate, including the following steps 101 to 105. Among them:

[0044] Step 101 : Process historical data of multiple cloud platforms to obtain a first sample data set, where the first sample data set includes sample data of multiple different data types.

[0045] The historical data includes target data, log data and security data, and the target data includes resource usage data and / or application performance data.

[0046] Resource usage data includes at least one of CPU (Central Processing Unit) utilization, memory usage data, storage usage data, and network traffic. CPU utilization refers to the CPU usage of different instances or virtual machines; memory usage data refers to memory usage and memory idle status; storage usage data refers to disk usage, including read and write operations and space occupancy; network traffic refers to inbound network traffic, outbound network traffic, network delay data, and packet loss rate.

[0047] Application performance data includes at least one of response time, throughput and error rate. Response time refers to the request time and response time of the application; throughput refers to the number of requests or transactions processed per second; error rate refers to the error rate and exception rate in the application or service.

[0048] Log data includes at least one of transaction logs, operating system logs, application logs, and security logs. Transaction logs refer to detailed transaction and operation logs recorded by applications; operating system logs refer to system event and error logs generated by the operating system; application logs refer to runtime logs generated by applications, including debugging information and error reports; and security logs refer to logs that record security events, user access records, and abnormal activities.

[0049] Security data includes at least one of security events and intrusion detection data. Security events refer to records including attack attempts, unauthorized access and other security threats; intrusion detection data includes alarms and reports from intrusion detection systems.

[0050] The historical data of multiple cloud platforms can be obtained regularly through the API (Application Programming Interface) provided by different cloud service providers.

[0051] Resource usage data and application performance data in the target data are both numerical data; log data is text data; and security data is alarm data. Data processing is performed on the historical data of multiple cloud platforms, including at least one of normalization, data smoothing, and encoding processing of numerical data; vectorization processing is performed on text data; and vectorization processing is performed on alarm data.

[0052] Step 102, obtaining positive sample pairs and negative sample pairs according to the sample data of text type and the sample data of numerical type in the first sample data set, and training the encoder in the to-be-trained model according to the positive sample pairs and the negative sample pairs to obtain a target encoder.

[0053] Among them, the sample data of multiple different data types in the first sample data set correspond to each other according to timestamps. Therefore, based on the sample data of text type and the sample data of numerical type in the first sample data set, positive sample pairs and negative sample pairs are obtained, including: taking the sample data of text type and the sample data of numerical type corresponding to the same timestamp as positive sample pairs; taking the sample data of text type and the sample data of numerical type corresponding to different timestamps as negative sample pairs.

[0054] It can be understood that historical data is obtained from multiple cloud platforms. Therefore, in the process of obtaining positive sample pairs and negative sample pairs, the consistency of the cloud platforms must also be considered, that is, the text type sample data and the numerical type sample data in the positive sample pairs belong to the same cloud platform; the text type sample data and the numerical type sample data in the negative sample pairs may belong to the same cloud platform or may not belong to the same cloud platform.

[0055] The encoder in the model to be trained includes a first encoder and a second encoder, and the first encoder and the second encoder can both be transformer encoders; according to the positive sample pair and the negative sample pair, the encoder in the model to be trained is trained to obtain a target encoder, including: inputting the sample data of the numerical type in the positive sample pair into the first encoder to obtain a first numerical feature, inputting the sample data of the text type in the positive sample pair into the second encoder to obtain a first text feature; performing cosine similarity processing on the first numerical feature and the first text feature to obtain the cosine similarity of the positive sample pair; inputting the sample data of the numerical type in the negative sample pair into the first encoder to obtain a second numerical feature, and inputting the sample data of the text type in the negative sample pair into the second encoder to obtain a second text feature; performing cosine similarity processing on the second numerical feature and the second text feature to obtain the cosine similarity of the negative sample pair; based on the cosine similarity of the positive sample pair, the cosine similarity of the negative sample pair and the constraint condition, the first encoder and the second encoder are trained to obtain the target encoder. The constraint condition is that the cosine similarity of the positive sample is greater than or equal to the first threshold, and the cosine similarity of the negative sample pair is less than or equal to the second threshold; the first threshold is close to 1, and the second threshold is close to 0.

[0056] The first encoder and the second encoder are trained based on the cosine similarity of the positive sample pair, the cosine similarity of the negative sample pair and the constraint conditions, including obtaining the cosine similarity loss value of the positive sample pair based on the cosine similarity of the positive sample pair and the cosine similarity loss function; obtaining the cosine similarity loss value of the negative sample pair based on the cosine similarity of the negative sample pair and the cosine similarity loss function; based on the cosine similarity loss value of the positive sample pair, the cosine similarity loss value of the negative sample pair and the constraint conditions, the first encoder and the second encoder are trained to obtain a first target encoder and a second target encoder.

[0057] After obtaining the first target encoder and the second target encoder, the first target encoder may be used as the target encoder.

[0058] The cosine similarity loss function is as follows:

[0059]

[0060]

[0061] in, represents model parameters; E represents expectation; i represents axis, i=0 represents row axis, i=1 represents column axis, L 0 and L 1They represent the losses on different axes, namely, the loss values ​​of text features and the loss values ​​of numerical features; j represents a sample pair, j=0 represents a positive sample pair, and j=1 represents a negative sample pair; n, w are the numerical features and text features after L2 normalization, respectively; Sim is the cosine similarity; both the first encoder and the second encoder include multiple projection layers, and D is the same projection layer of the first encoder and the second encoder.

[0062] In order to facilitate the understanding of the encoder training process in the above-mentioned training model, the following Figure 3 Schematic diagram of the encoder training process shown.

[0063] It can be understood that, given that the text part already has mature Transformer feature extraction parameters, in order to reduce the complexity of parameter updating, the model parameters in the second encoder can be kept fixed during the training process, or after several rounds of training, the model parameters of the two encoder parts can no longer be updated, and only the projection layer with fewer parameters can be updated to speed up the training process.

[0064] Step 103: Train the decoder in the to-be-trained model according to the sample data of the numerical type and the sample data of the alarm type in the first sample data set to obtain a target decoder.

[0065] The decoder in the model to be trained may include a first decoder, and the corresponding target decoder includes a first target decoder. The decoder in the model to be trained may also include a first decoder and a second decoder, and the corresponding target decoder includes a first target decoder and a second target decoder.

[0066] The training of the first decoder may be to use the sample data of the numerical type as input and the sample data of the alarm type as a label to obtain the first target decoder. It can be understood that since the first target decoder is trained based on the sample data of the alarm type, the first target decoder can output alarm information.

[0067] The training of the second decoder may be to use a portion of the sample data of the numerical type in the first sample data set as input and another portion of the sample data of the numerical type as labels to obtain a second target decoder. It can be understood that if the second target decoder is trained based on the resource usage data, the second target decoder can output resource demand prediction information, and if the second target decoder is trained based on the application performance data, the second target decoder can output application performance prediction information.

[0068] Step 104, obtaining a target monitoring model according to the target encoder and the target decoder.

[0069] A monitoring model is obtained according to the target encoder, the target decoder and the output layer. After the monitoring model is obtained, the monitoring model can be directly used as the target monitoring model, or the monitoring model can be further optimized to obtain the target monitoring model.

[0070] Step 105: monitor multiple cloud platforms based on the target monitoring model to obtain monitoring results of each cloud platform.

[0071] The monitoring result includes at least one of alarm information, resource demand prediction information and application performance prediction information.

[0072] In one example, after obtaining the target monitoring model, resource usage data and / or application performance data are collected from each cloud platform, and the resource usage data and / or application performance data are input into the target monitoring model to obtain the monitoring results of the cloud platform output by the target monitoring model.

[0073] In the above-mentioned cloud platform monitoring method, by processing the historical data of multiple cloud platforms, sample data including multiple different data types are obtained. Then, according to the sample data of text type and the sample data of numerical type in the first sample data set, positive sample pairs and negative sample pairs are obtained, and according to the positive sample pairs and negative sample pairs, the encoder in the training model is trained to obtain the target encoder; according to the sample data of numerical type and the sample data of alarm type in the first sample data set, the decoder in the training model is trained to obtain the target decoder; according to the target encoder and the target decoder, the target monitoring model is obtained. Then, based on the target monitoring model, multiple cloud platforms are monitored to obtain the monitoring results of each cloud platform. In this way, the target monitoring model obtained by training the historical data of multiple cloud platforms monitors multiple cloud platforms, which not only realizes the monitoring of multiple cloud platforms, but also improves the management efficiency and reliability of the multi-cloud environment compared to the method of realizing cloud platform monitoring by manually analyzing cloud platform data.

[0074] In an exemplary embodiment, the historical data includes target data, log data, and security data, and the target data includes resource usage data and / or application performance data. The historical data of multiple cloud platforms are processed to obtain a first sample data set, including the following steps 1011 to 1014, wherein:

[0075] Step 1011 , performing superposition processing on target data of different types corresponding to the same timestamp to obtain sample data of numerical types corresponding to each timestamp.

[0076] Optionally, multiple target data in the historical data are arranged according to timestamps, and the data range is the timestamp range of log data of two adjacent timestamps. Each target data includes at least two of CPU utilization, memory usage data, storage usage data, network traffic, response time, throughput and error rate.

[0077] Each type of data in the multiple target data is subjected to normalization processing, data smoothing processing and encoding processing respectively.

[0078] Normalize each type of data in multiple target data to obtain each type of data after normalization. The formula for normalization is as follows:

[0079]

[0080] Take CPU utilization as an example. Indicates the normalized CPU utilization corresponding to time point t; Indicates the CPU utilization corresponding to time point t; represents the average value; Represents standard deviation.

[0081] Perform data smoothing on each type of data after normalization to obtain each type of data after data smoothing. The formula for data smoothing is as follows:

[0082]

[0083] Take CPU utilization as an example. Indicates the CPU utilization after smoothing of the data corresponding to time point t; It means standard deviation, which is understandable. is the standard deviation of the Gaussian function, which determines the degree of smoothing; N is the latest time point.

[0084] After data smoothing, each type of data after data smoothing is one-hot encoded to obtain each type of data after one-hot encoding; each type of data after one-hot encoding corresponding to the same timestamp is superimposed to obtain the sample data of the numerical type corresponding to each timestamp. The sample data of the numerical type corresponding to multiple timestamps constitute a time series vector, which can be recorded as .

[0085] For example, the target data includes CPU utilization and throughput. The CPU utilization after one-hot encoding corresponding to timestamp t is 1000000, and the throughput after one-hot encoding corresponding to timestamp t is 0100000. The various types of data after one-hot encoding corresponding to timestamp t are superimposed, and the sample data corresponding to time t is 1100000.

[0086] Step 1012, perform vectorization processing on the log data of each timestamp to obtain sample data of the text type corresponding to each timestamp.

[0087] The log data includes at least one log type of transaction log, operating system log, application log, security log, etc.

[0088] For the log data of each log type, the log data of each timestamp of the log type is segmented and vectorized to obtain the sample data of the text type corresponding to each timestamp of the log type. The sample data of the text type corresponding to multiple timestamps of the log type can constitute the word vector sequence of the log data of the log type, which can be recorded as Among them, word segmentation and vectorization processing can use NLP (Natural Language Processing).

[0089] Step 1013, vectorize the security data of each timestamp to obtain sample data of the alarm type corresponding to each timestamp.

[0090] Among them, security data is similar to log data, both of which belong to text type. Therefore, the security data of each timestamp is also processed by word segmentation and vectorization to obtain the sample data of the alarm type corresponding to each timestamp. The sample data of the alarm type is used as the label of the training model. Therefore, the sample data of the alarm type corresponding to multiple timestamps constitute the word vector sequence of the label, which can be recorded as .

[0091] Step 1014: Obtain a first sample data set according to the sample data of the numerical type, the sample data of the text type, and the sample data of the alarm type corresponding to each time stamp.

[0092] Optionally, sample data of the numerical type, sample data of the text type, and sample data of the alarm type are matched according to timestamps to construct a first sample data set.

[0093] In an exemplary embodiment, the decoder includes a first decoder, and according to the sample data of the numerical type and the sample data of the alarm type in the first sample data set, the decoder in the training model is trained to obtain a target decoder, including steps 1031 to 1033:

[0094] Step 1031: input sample data of numerical type into a target encoder to obtain a first output result output by the target encoder, and input the first output result into a first decoder to obtain alarm data output by the first decoder.

[0095] It can be understood that, when the target data includes resource usage data, the sample data of the numerical type obtained after data processing on the target data is the resource usage sample data, and thus the first output result is the resource numerical feature.

[0096] In the case where the target data includes application performance data, sample data of a numerical type obtained after data processing of the target data is application performance sample data, and thus the first output result is a performance numerical feature.

[0097] When the target data includes application performance data and resource usage data, the sample data of numerical type obtained after data processing of the target data is fused sample data with resource usage and application performance, so that the first output result is the fused numerical feature.

[0098] Step 1032, obtaining an alarm loss value according to the alarm data and the sample data of the alarm type, and training the first decoder based on the alarm loss value to obtain a first target decoder;

[0099] The sample data of the alarm type is equivalent to the real label. According to the alarm data and the sample data of the alarm type, the alarm loss value is obtained, including: calculating the cross entropy loss of the alarm data and the sample data of the alarm type to obtain the alarm loss value.

[0100] The calculation formula of the alarm loss value is as follows:

[0101]

[0102] in, represents the alarm loss value; E represents the expectation; c represents the alarm data, and w represents the sample data of the alarm type.

[0103] It can be understood that, in the process of training the first decoder, the sample data of the numerical type input to the target encoder each time has the same timestamp as the sample data of the alarm type used as a label.

[0104] Step 1033, obtaining a target decoder according to the first target decoder.

[0105] After obtaining the first target decoder, the first target decoder may be directly used as the target decoder; or the target decoder may be obtained based on the first target decoder and the second target decoder.

[0106] In one example, the decoder further includes a second decoder, and the first sample data set includes sample data of numerical types corresponding to a plurality of timestamps; step 1033 includes the following steps 10331 to 10334, wherein:

[0107] Step 10331: split the sample data of multiple numerical types into first sample data and second sample data.

[0108] The sample data of multiple numerical types within a preset time period are split into first sample data and second sample data; the timestamp corresponding to the first sample data is smaller than the timestamp corresponding to the second sample data.

[0109] It can be understood that when the target data includes resource usage data, the sample data of numerical type obtained after data processing of the target data is the resource usage sample data, so that the first sample data and the second sample data are the resource usage sample data corresponding to different timestamps.

[0110] When the target data includes application performance data, sample data of numerical type obtained after data processing of the target data is application performance sample data, and thus the first sample data and the second sample data are application performance sample data corresponding to different timestamps.

[0111] Step 10332, input the first sample data into the target encoder to obtain a second output result output by the target encoder, and input the second output result into the second decoder to obtain target prediction data output by the second decoder, wherein the target prediction data includes resource demand prediction data and / or application performance prediction data.

[0112] It can be understood that, when the target data includes resource usage data, the second output result is the resource numerical feature. When the target data includes application performance data, the second output result is the performance numerical feature. When the target data includes application performance data and resource usage data, the second output result is the fusion numerical feature.

[0113] Step 10333, obtain a target prediction loss value according to the target prediction data and the second sample data, train the second decoder based on the target prediction loss value, and obtain a second target decoder.

[0114] The second sample data is equivalent to the true label. According to the target prediction data and the second sample data, a target prediction loss value is obtained, including: calculating the cross entropy loss of the target prediction data and the second sample data to obtain the target prediction loss value.

[0115] The target prediction loss value is calculated as follows:

[0116]

[0117] in, represents the alarm loss value; E represents the expectation; c represents the target prediction data, and w represents the second sample data.

[0118] Step 10334, obtain a target decoder according to the first target decoder and the second target decoder.

[0119] In order to facilitate the understanding of the training process of the decoder in the above-mentioned training model, the following is provided: Figure 4 Figure 1 shows the schematic diagram of the decoder training process.

[0120] In this embodiment, since the first target decoder is trained based on the sample data of the alarm type, the first target decoder can output alarm information; if the second target decoder is trained based on the resource usage data, the second target decoder can output resource demand prediction information; if the second target decoder is trained based on the application performance data, the second target decoder can output application performance prediction information. Therefore, the target decoder composed of the first target decoder and the second target decoder can realize both alarm and prediction of future resource demand and application performance based on the resource usage data and application performance data of the cloud platform.

[0121] In an exemplary embodiment, a target monitoring model is obtained according to a target encoder and a target decoder, including steps 1041 to 1045, wherein:

[0122] Step 1041, obtain a monitoring model according to the target encoder, target decoder and output layer.

[0123] The output layer converts the outputs of the first target decoder and the second target decoder into vector representations of warning data and vector representations of target prediction data.

[0124] It can be understood that the architecture of the monitoring model is the target encoder, target decoder, and output layer.

[0125] Step 1042, replace the output layer in the monitoring model with a linear layer to obtain a target reward model.

[0126] Optionally, the output layer in the monitoring model is replaced with a linear layer to obtain a reward model; the sample data of the numerical type is input into the monitoring model K times to obtain 2K output results of the monitoring model; based on the sample data of the numerical type input into the monitoring model and the 2K output results, a second sample data set is obtained; the reward model is trained based on the second sample data set to obtain a target reward model.

[0127] Among them, the linear layer is an initialized linear layer, which is used to generate a scalar output (output score), which represents the output quality. The higher the score, the higher the quality.

[0128] In one example, the target decoder includes a first target decoder and a second target decoder, so that by inputting the sample data of the numerical type into the monitoring model K times, 2K output results can be obtained, and the 2K output results include K alarm data and K target prediction data. It can be understood that when the sample data of the numerical type includes resource usage sample data, the target prediction data can be resource demand prediction data.

[0129] Based on the sample data of the numerical type and 2K output results of the input monitoring model, the format of the data in the second sample data set is: one sample data of the numerical type corresponds to K different alarm data and K different target prediction data. The 2K output results are sorted, that is, one sample data of the numerical type corresponds to two output sorts.

[0130] By inputting a sample data of a numerical type and its corresponding alarm data into the reward model, the alarm score output by the reward model can be obtained; by inputting a sample data of a data type and its corresponding target prediction data into the reward model, the target prediction score output by the reward model can be obtained, such as the resource demand prediction score. It can be understood that for a sample data of a numerical type, it is necessary to input 2K times into the reward model to obtain the corresponding scores of 2K output results, and the scores must satisfy the ranking of the second sample data set.

[0131] Specifically, the sample data of numerical type is represented as x, and the corresponding K warning data and K target prediction data are represented as y={y 1 ,…,y K}, where the rating sort order is y 1 >y 2 >…>y K , the high-score area is represented as y h ={y 1 ,…,y w}, the low-scoring area is represented by y d ={y w+1 ,…,y K}. K represents the number of outputs corresponding to each input, and the warning data and target prediction data each have K outputs. Therefore, both the warning data and the target prediction data will have Comparison. It is the reward value obtained by inputting numerical sample data and high-score output (warning data or target prediction data) into the reward model. It is the reward value obtained by inputting numerical sample data and low-score output (warning data or target prediction data) into the reward model. The purpose of training the reward model is to make the difference between high scores and low scores as large as possible. Therefore, in the process of training the reward model, the calculation formula of the loss function of the reward model is as follows:

[0132]

[0133] in, represents the expected value after calculating the cross entropy of the sample data x of the numerical type and the warning data / target prediction data selected from the second sample data set, The function represents the sigmoid function.

[0134] The trained target reward model can identify the quality of the output generated by the monitoring model, that is, based on the sample data of the input data type, the target reward model can judge the accuracy of the alarm data and target prediction data output by the monitoring model.

[0135] Step 1043, initialize the monitoring model, and input sample data of numerical type into the initialized monitoring model to obtain the output result of the initialized monitoring model.

[0136] Step 1044, inputting the sample data of the numerical type input into the initialized monitoring model and the output result of the initialized monitoring model into the target reward model to obtain the reward value output by the target reward model;

[0137] Step 1045 , adjusting the parameters of the initialized monitoring model based on the reward value output by the target reward model to obtain a target monitoring model.

[0138] The above steps 1043 to 1045 are to further strengthen the monitoring model based on the trained target reward model to obtain the target monitoring model. The specific strengthening process is as follows:

[0139] Initialize the monitoring model to obtain the initialized monitoring model, input the sample data x of the numerical type into the initialized monitoring model, obtain the output result of the initialized monitoring model, the output result includes the alarm data y and the target prediction data y, input the sample data x of the numerical type and the alarm data y into the target reward model, or input the sample data x of the numerical type and the target prediction data y into the reward value output by the target reward model , using the reward value The parameters of the monitoring model are adjusted to obtain the target monitoring model.

[0140] Reward value in use In the process of adjusting the parameters of the initialized monitoring model, KL divergence is used to restrict the monitoring model so that it cannot deviate greatly from the original monitoring model in order to obtain a higher score. The specific expression is:

[0141]

[0142] in, Represents the initialized monitoring model RLM; Represents the monitoring model MPMM; , is the adjustment factor; Indicates the sample data of the numerical type input to the initialized monitor model RLM and the output result of the initialized monitor model output.

[0143] In order to facilitate the understanding of the above-mentioned enhanced training process of the monitoring model, the following is provided: Figure 5 The schematic diagram of the supervisory model reinforcement training process is shown in Figure 1. The reinforcement training process uses the PPO algorithm and RLHF.

[0144] In this embodiment, enhanced training of the monitoring model is implemented based on the target reward model, so that the obtained target monitoring model has higher reliability.

[0145] Based on the above cloud platform monitoring method, the target monitoring model obtained can not only output alarms when monitoring the cloud platform, but also predict future resource requirements and application performance trends, so as to give users prompts to expand resources in a timely manner, reduce service terminals, storage overflow, traffic congestion, etc. In addition, this application uses different types of data from multiple cloud platforms for training, covering most risk operations and types of attacks, and can more accurately identify the causes of risks, so that risks can be handled faster, greatly improving the reliability and stability of monitoring.

[0146] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0147] Based on the same inventive concept, the embodiment of the present application also provides a cloud platform monitoring device for implementing the cloud platform monitoring method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more cloud platform monitoring device embodiments provided below can refer to the limitations of the cloud platform monitoring method above, and will not be repeated here.

[0148] In an exemplary embodiment, Figure 6 As shown, a cloud platform monitoring device is provided, and the cloud platform monitoring device 600 includes: a processing module 601, a first training module 602, a second training module 603, a determination module 604 and a monitoring module 605, wherein:

[0149] The processing module 601 is used to process the historical data of multiple cloud platforms to obtain a first sample data set, where the first sample data set includes sample data of multiple different data types;

[0150] A first training module 602 is used to obtain positive sample pairs and negative sample pairs according to the sample data of the text type and the sample data of the numerical type in the first sample data set, and train the encoder in the to-be-trained model according to the positive sample pairs and the negative sample pairs to obtain a target encoder;

[0151] The second training module 603 is used to train the decoder in the training model according to the sample data of the numerical type and the sample data of the alarm type in the first sample data set to obtain a target decoder;

[0152] A determination module 604 is used to obtain a target monitoring model according to a target encoder and a target decoder;

[0153] The monitoring module 605 is used to monitor multiple cloud platforms based on the target monitoring model to obtain monitoring results of each cloud platform.

[0154] In one embodiment, historical data includes target data, log data and security data, and the target data includes resource usage data and / or application performance data; the processing module 601 is specifically used to superimpose different types of target data corresponding to the same timestamp to obtain sample data of a numerical type corresponding to each timestamp; vectorize the log data of each timestamp to obtain sample data of a text type corresponding to each timestamp; vectorize the security data of each timestamp to obtain sample data of an alarm type corresponding to each timestamp; and obtain a first sample data set based on the sample data of the numerical type, the sample data of the text type and the sample data of the alarm type corresponding to each timestamp.

[0155] In one embodiment, the decoder includes a first decoder and a second training module 603, which is specifically used to input sample data of a numerical type into a target encoder to obtain a first output result output by the target encoder, and input the first output result into a first decoder to obtain alarm data output by the first decoder; obtain an alarm loss value based on the alarm data and the sample data of the alarm type, train the first decoder based on the alarm loss value to obtain a first target decoder; and obtain a target decoder based on the first target decoder.

[0156] In one embodiment, the decoder also includes a second decoder, the first sample data set includes sample data of multiple numerical types corresponding to multiple timestamps, and the second training module 603 is specifically used to split the sample data of multiple numerical types into first sample data and second sample data; input the first sample data into a target encoder to obtain a second output result output by the target encoder, and input the second output result into a second decoder to obtain target prediction data output by the second decoder, the target prediction data including resource demand prediction data and / or application performance prediction data; obtain a target prediction loss value based on the target prediction data and the second sample data, and train the second decoder based on the target prediction loss value to obtain a second target decoder; obtain a target decoder based on the first target decoder and the second target decoder.

[0157] In one embodiment, the determination module is specifically used to obtain a monitoring model based on the target encoder, the target decoder and the output layer; replace the output layer in the monitoring model with a linear layer to obtain a target reward model; initialize the monitoring model, and input the sample data of the numerical type into the initialized monitoring model to obtain an output result of the initialized monitoring model; input the sample data of the numerical type input into the initialized monitoring model and the output result of the initialized monitoring model into the target reward model to obtain a reward value output by the target reward model; adjust the parameters of the initialized monitoring model based on the reward value output by the target reward model to obtain the target monitoring model.

[0158] In one embodiment, the determination module 604 is specifically used to replace the output layer in the monitoring model with a linear layer to obtain a reward model; input the sample data of the numerical type into the monitoring model K times to obtain 2K output results of the monitoring model, and the 2K output results include K alarm data and K target prediction data; obtain a second sample data set based on the sample data of the numerical type input into the monitoring model and the 2K output results; train the reward model based on the second sample data set to obtain a target reward model.

[0159] Each module in the above-mentioned cloud platform monitoring device can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0160] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 7As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as a first sample data set, a second sample data set, and monitoring results. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a cloud platform monitoring method is implemented.

[0161] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0162] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of any one of the method embodiments described above are implemented.

[0163] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the above method embodiments are implemented.

[0164] In one embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of any one of the method embodiments described above.

[0165] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0166] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0167] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0168] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A cloud platform monitoring method, characterized in that: The method comprises: Processing historical data of multiple cloud platforms to obtain a first sample data set, wherein the first sample data set includes sample data of multiple different data types; According to the sample data of the text type and the sample data of the numerical type in the first sample data set, a positive sample pair and a negative sample pair are obtained, and according to the positive sample pair and the negative sample pair, an encoder in the to-be-trained model is trained to obtain a target encoder; According to the sample data of the numerical type and the sample data of the alarm type in the first sample data set, the decoder in the to-be-trained model is trained to obtain a target decoder; A target monitoring model is obtained according to the target encoder and the target decoder; and multiple cloud platforms are monitored based on the target monitoring model to obtain monitoring results of each cloud platform.

2. The method according to claim 1, characterized in that: The historical data includes target data, log data and security data, and the target data includes resource usage data and / or application performance data; the data processing of the historical data of multiple cloud platforms to obtain a first sample data set includes: Superimpose different types of target data corresponding to the same timestamp to obtain sample data of the numerical type corresponding to each timestamp; Performing vectorization processing on the log data of each timestamp to obtain sample data of the text type corresponding to each timestamp; Vectorizing the security data of each timestamp to obtain sample data of the alarm type corresponding to each timestamp; The first sample data set is obtained according to the sample data of the numerical type, the sample data of the text type, and the sample data of the alarm type corresponding to each time stamp.

3. The method according to claim 1, characterized in that: The decoder includes a first decoder, and the decoder in the to-be-trained model is trained according to the sample data of the numerical type and the sample data of the alarm type in the first sample data set to obtain a target decoder, including: Inputting the sample data of the numerical type into the target encoder to obtain a first output result output by the target encoder, and inputting the first output result into the first decoder to obtain the warning data output by the first decoder; Obtaining an alarm loss value according to the alarm data and sample data of the alarm type, and training the first decoder based on the alarm loss value to obtain a first target decoder; According to the first target decoder, the target decoder is obtained.

4. The method according to claim 3, characterized in that The decoder further includes a second decoder, the first sample data set includes sample data of numerical types corresponding to a plurality of timestamps, and the target decoder is obtained according to the first target decoder, including: Splitting the plurality of sample data of the numerical type into first sample data and second sample data; Inputting the first sample data into the target encoder to obtain a second output result output by the target encoder, and inputting the second output result into the second decoder to obtain target prediction data output by the second decoder, wherein the target prediction data includes resource demand prediction data and / or application performance prediction data; Obtaining a target prediction loss value according to the target prediction data and the second sample data, and training the second decoder based on the target prediction loss value to obtain a second target decoder; The target decoder is obtained according to the first target decoder and the second target decoder.

5. The method according to claim 1, characterized in that The step of obtaining a target monitoring model according to the target encoder and the target decoder comprises: Obtaining a monitoring model according to the target encoder, the target decoder and the output layer; Replacing the output layer in the monitoring model with a linear layer to obtain a target reward model; Initializing the monitoring model, and inputting the sample data of the numerical type into the initialized monitoring model to obtain an output result output by the initialized monitoring model; Inputting the sample data of the numerical type input into the initialized monitoring model and the output result of the initialized monitoring model into the target reward model to obtain the reward value output by the target reward model; The parameters of the initialized monitoring model are adjusted based on the reward value output by the target reward model to obtain the target monitoring model.

6. The method according to claim 5, characterized in that The step of replacing the output layer in the monitoring model with a linear layer to obtain a target reward model includes: Replacing the output layer in the monitoring model with a linear layer to obtain a reward model; Inputting the sample data of the numerical type into the monitoring model K times to obtain 2K output results output by the monitoring model; the 2K output results include K warning data and K target prediction data; Based on the sample data of the numerical type input into the monitoring model and the 2K output results, a second sample data set is obtained; The reward model is trained based on the second sample data set to obtain the target reward model.

7. A cloud platform monitoring device, characterized in that: The device comprises: A processing module, configured to process historical data of multiple cloud platforms to obtain a first sample data set, wherein the first sample data set includes sample data of multiple different data types; A first training module is used to obtain positive sample pairs and negative sample pairs according to the sample data of the text type and the sample data of the numerical type in the first sample data set, and train the encoder in the training model according to the positive sample pairs and the negative sample pairs to obtain a target encoder; A second training module is used to train the decoder in the to-be-trained model according to the sample data of the numerical type and the sample data of the alarm type in the first sample data set to obtain a target decoder; A determination module, configured to obtain a target monitoring model according to the target encoder and the target decoder; The monitoring module is used to monitor the multiple cloud platforms based on the target monitoring model to obtain monitoring results of each cloud platform.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images