Gateway resource allocation method and device for financial service tenants, equipment and medium

Abstract

The invention relates to the field of financial science and technology, and discloses a gateway resource allocation method and device for financial service tenants, equipment and a medium, and the method comprises the steps: inputting obtained flow data of the financial service tenants into a trained flow feature extraction network model, and obtaining a flow feature sequence of each financial service tenant; classifying the traffic feature sequence of each tenant to obtain a resource demand traffic type and a traffic priority score; inputting the traffic feature sequence, the resource demand traffic type and the traffic priority score of each financial service tenant into a trained resource allocation network model to obtain a resource allocation result of a next preset time period; and allocating corresponding resources to the corresponding financial service tenants in a resource pool of the proxy gateway cluster according to a resource allocation result. The method solves the problem that the existing static resource allocation scheme is difficult to adapt to tenant requirements or significant fluctuation of traffic load due to allocation of fixed resources, and is easy to cause service performance reduction or resource waste.

Description

Technical Field

[0001] The present invention relates to the fields of data processing technology and financial technology, and in particular to a gateway resource allocation method, device, equipment and medium for financial business tenants. Background Art

[0002] In cloud computing, proxy gateway clusters serve as the core hub connecting multiple tenants in financial services with external service providers. Their resource allocation mechanisms directly impact system performance, security, and cost-effectiveness. Given the stringent requirements of financial services for high concurrency, low latency, and strong security, traditional static resource allocation methods are no longer able to meet these dynamic demands.

[0003] Static resource allocation allocates fixed computing, storage, and network resources to each tenant based on pre-defined rules or policies. In this model, resource allocation is determined at the initial stage, and each tenant's resource ceiling remains unchanged based on actual demand throughout system operation. Due to its lack of flexibility, this fixed allocation approach can lead to idle or insufficient resources. When tenant demand or traffic load fluctuates significantly, static resource allocation often fails to respond promptly, resulting in degraded service performance and wasted resources. Summary of the Invention

[0004] The present invention provides a gateway resource allocation method, device, equipment and medium for financial service tenants to solve the problem that existing static resource allocation schemes are difficult to adapt to tenant needs or significant fluctuations in traffic load due to the allocation of fixed resources, which easily leads to service performance degradation or resource waste.

[0005] In a first aspect, a gateway resource allocation method for a financial service tenant is provided, comprising: Obtain traffic data for N financial service tenants in the proxy gateway cluster during the current historical period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet. Inputting the traffic data of the financial service tenants into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources; By classifying the traffic feature sequences used to characterize the financial service resource demands of each of the financial service tenants, a resource demand traffic type and a traffic priority score corresponding to the resource demand traffic type are obtained; Inputting the traffic feature sequence, resource demand traffic type, and traffic priority score of each financial service tenant into the trained resource allocation network model to obtain the resource allocation result for the next preset time period of each financial service tenant; According to the resource allocation result of the next preset time period, corresponding resources are allocated to the corresponding financial service tenant in the resource pool of the proxy gateway cluster.

[0006] In a second aspect, a gateway resource allocation device for a financial service tenant is provided, comprising: An acquisition module is used to obtain the traffic data of N financial business tenants of the proxy gateway cluster in the current historical time period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet; A feature extraction module, configured to input the traffic data of the financial service tenants into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources; a classification module, configured to classify the traffic feature sequences characterizing the financial service resource demands of the financial service tenants to obtain resource demand traffic types and traffic priority scores corresponding to the resource demand traffic types; A resource allocation prediction module is used to input the traffic feature sequence, resource demand traffic type and traffic priority score of each financial service tenant into the trained resource allocation network model to obtain the resource allocation result of each financial service tenant in the next preset time period; The resource allocation module is used to allocate corresponding resources to the corresponding financial service tenant in the resource pool of the proxy gateway cluster according to the resource allocation result of the next preset time period.

[0007] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for allocating gateway resources to financial service tenants are implemented.

[0008] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned gateway resource allocation method for financial business tenants are implemented.

[0009] In the above-mentioned gateway resource allocation method, device, equipment and medium for financial business tenants, a network model is extracted through traffic characteristics, and traffic data is analyzed from multiple dimensions such as protocol type, packet size, and sending time to generate a traffic feature sequence for characterizing the demand for financial business resources of each financial business tenant; the traffic feature sequence is further classified to predict the traffic data type of the financial business tenant in the next time period, and the priority score is quantified to provide a scientific basis for subsequent resource allocation. By utilizing the resource allocation network model, combined with the traffic feature sequence of each tenant and its corresponding predicted traffic data type and priority score of the financial business tenant, the resource demand of each tenant in the next preset time period is predicted, and exclusive resources are dynamically allocated to each financial business tenant to ensure that high-priority traffic types obtain sufficient resources and low-priority traffic types are allocated on demand, thereby improving overall resource utilization and avoiding resource waste or shortage. Therefore, the present invention solves the problem that the existing static resource allocation scheme is difficult to adapt to tenant demand or significant fluctuations in traffic load due to the allocation of fixed resources, which easily leads to service performance degradation or resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0011] Figure 1 This is a schematic diagram of an application environment of a gateway resource allocation method for financial service tenants in one embodiment of the present invention; Figure 2 This is a flow chart of a method for allocating gateway resources to financial service tenants in one embodiment of the present invention; Figure 3 This is a structural diagram of a gateway resource allocation device for a financial service tenant in one embodiment of the present invention; Figure 4 It is a structural diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0012] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0013] The gateway resource allocation method for financial service tenants provided in the embodiment of the present invention can be applied in the following situations: Figure 1 In an application environment, the client communicates with the server through the network. The server obtains the traffic data of N financial business tenants of the proxy gateway cluster in the current historical time period; the traffic data includes the protocol type of the data packet, the size of the data packet and the sending time of the data packet; the traffic data of the financial business tenants is input into the trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand for financial business resources of each of the financial business tenants; by classifying the traffic feature sequence for characterizing the demand for financial business resources of each of the financial business tenants, the resource demand traffic type and the traffic priority score corresponding to the resource demand traffic type are obtained; the traffic feature sequence, resource demand traffic type and traffic priority score of each of the financial business tenants are input into the trained resource allocation network model to obtain the resource allocation result of each of the financial business tenants in the next preset time period; according to the resource allocation result of the next preset time period, corresponding resources are allocated to the corresponding financial business tenant in the resource pool of the proxy gateway cluster. In the present invention, a resource allocation network model is used, combined with the predicted traffic data type and priority score of the financial business tenant, to predict the resource demand for the next preset time period, and dynamically allocate exclusive resources to each financial business tenant to ensure that high-priority traffic types obtain sufficient resources and low-priority traffic types are allocated on demand, thereby improving overall resource utilization and avoiding resource waste or shortage. Among them, the client can be but is not limited to various personal computers, laptops, smart phones, tablets and portable wearable devices. The server can be implemented with an independent server or a server cluster composed of multiple servers. The present invention is described in detail below through specific embodiments.

[0014] In one embodiment, see Figure 2 As shown, Figure 2 A flowchart of a method for allocating gateway resources to a financial service tenant provided in an embodiment of the present invention includes the following steps: S01: Obtain traffic data of N financial business tenants of the proxy gateway cluster in the current historical time period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet.

[0015] In this embodiment, the Gateway Proxy is a specialized proxy server that provides proxy services specifically for gateways. Acting as an intermediary between clients and backend servers, the Gateway Proxy receives client requests, forwards them to the backend servers, and then returns responses to the clients. A "Gateway Proxy Cluster" combines multiple Gateway Proxy servers and uses load balancing technology to achieve high availability and high performance. This deployment method primarily ensures that if a single Gateway node fails, traffic can be seamlessly transferred to other nodes without interruption.

[0016] In this embodiment, the method for obtaining the traffic data of N financial service tenants of the proxy gateway cluster in the current historical time period is to deploy a traffic collection tool in the proxy gateway cluster, which specifically includes the following methods: Traffic mirroring: Use the port mirroring function of a switch or load balancer to copy traffic to a traffic analysis device.

[0017] This is a built-in function of the proxy gateway. If the proxy gateway supports traffic logging (such as Nginx logs and HAProxy logs), it can be enabled directly.

[0018] Use third-party tools such as Wireshark, tcpdump, and Zeek (Bro) to capture traffic.

[0019] The present invention obtains the traffic data of N financial business tenants of the proxy gateway cluster in the current historical time period through the traffic mirroring method.

[0020] In network communications, traffic data is transmitted in packets. A packet is a unit of data encapsulated by the network layer, transport layer, or application layer protocols. It contains multiple fields and is used to ensure reliable data transmission and routing. It typically consists of the following main parts: 1) Header: The header of a data packet contains control information, such as the sending time of the data packet, source address, destination address, protocol type, data packet size, etc. This information is crucial for the routing and transmission of data packets in the network.

[0021] 2) Data: The actual content of the data packet, which can include text, images, audio, video, and other types of data. The size and content of the data depends on the specific network application.

[0022] 3) Trailer: Some protocols may add a checksum or other control information to the end of the packet to ensure data integrity and correctness. The trailer information is often used to verify data integrity after the packet reaches the destination.

[0023] S02: Inputting the traffic data of the financial service tenants into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources.

[0024] In this embodiment, before inputting the traffic data of the financial service tenant into the trained traffic feature extraction network model, the traffic data of the financial service tenant needs to be preprocessed. The specific processing process is as follows: Data cleansing is the process of reexamining and verifying data, aiming to remove duplicate information, correct existing errors, and provide data consistency verification. This example uses statistical methods to locate null values ​​and calculate the missing percentage for each column. Numerical features are filled using the median or linear interpolation. For time features, forward filling (ffill) or backward filling (bfill) are used. Forward filling fills missing values ​​with non-null values ​​from the previous traffic packet; backward filling fills missing values ​​with non-null values ​​from the next traffic packet. Character features are filled using the mode or a new "unknown" category is added.

[0025] Conversion processing converts the protocol type into an integer label and the timestamp into a date and time format, so that the subsequent model can extract features such as hour, day of the week, and whether it is a holiday. The numerical features are standardized / normalized.

[0026] Time series feature construction aggregates traffic data by fixed time windows (e.g., 10 seconds) to form a time series matrix. Each time window contains 10 traffic records, each with protocol characteristics and packet size.

[0027] In this embodiment, the traffic feature extraction network model uses deep learning technology to automatically extract key features of tenant traffic data from network traffic for tasks such as traffic classification, anomaly detection, and intrusion detection.

[0028] The preprocessed traffic data of financial tenants is fed into a trained traffic feature extraction network model. This model extracts and outputs information from each tenant's traffic data, including network protocol features, packet size features, temporal variations in traffic rate, and the time intervals between consecutive packet arrivals. This generates a traffic feature sequence that characterizes each financial tenant's demand for financial resources. Traffic feature sequences are key data that describe the temporal changes in resource demand of tenants (such as banks, securities firms, and insurance institutions). Analyzing these feature sequences enables dynamic resource allocation, capacity planning, and anomaly detection.

[0029] S03: Obtain resource demand traffic types and traffic priority scores corresponding to the resource demand traffic types by classifying the traffic feature sequences used to characterize the financial service resource demands of the financial service tenants.

[0030] In this embodiment, the classification model needs to predict multiple categories. To this end, a fully connected layer and a softmax function are used to classify the traffic feature sequences of the financial business resource demands of each financial business tenant. Specifically, if the traffic feature sequences of each financial business tenant are first input into the trained fully connected layer, its high-dimensional traffic feature sequence is mapped to the classification label space, and the unnormalized raw scores (Logits) are output. These scores reflect the model's confidence in each classification category, but have not yet been converted into probability values. Subsequently, the raw scores output by the fully connected layer are converted into probability distributions through the softmax function, and finally the prediction results are determined based on the category corresponding to the maximum probability. If the traffic feature extraction network model already includes a fully connected layer, the current classification model only needs to add a softmax function after the fully connected layer to convert the output raw scores into probability distributions.

[0031] In this embodiment, since the above-mentioned classification label space is fixed and does not change much, a fixed priority score is directly mapped according to the classification result.

[0032] S04: Input the traffic feature sequence, resource demand traffic type and traffic priority score of each financial business tenant into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

[0033] In this embodiment, the resource allocation network model is based on deep reinforcement learning. It dynamically predicts and allocates system resources (such as CPU, memory, bandwidth, and storage) based on input traffic feature sequences, traffic types, and priority information. The core goal of this model is to optimize resource allocation strategies to improve system performance, reduce resource waste, and ensure that critical tasks are prioritized.

[0034] Deep reinforcement learning models primarily consist of a policy network and a value network. During training, the policy network outputs the probability distribution of actions, while the value network evaluates the value of the current state to aid in policy optimization. Once the deep reinforcement learning model is trained, the policy network can be used directly to generate the probability distribution of actions.

[0035] In this embodiment, the traffic feature sequence of each tenant obtained in step S02, and its corresponding resource demand traffic type and priority score are integrated into a state vector S. For example, the state vector of each tenant can be expressed as: S = [traffic feature sequence, traffic type code, priority score], ensuring that the state vector dimension matches the input layer of the reinforcement learning model.

[0036] The state vector S of each tenant is passed into the trained policy network to obtain the probability distribution of actions that may correspond to the current state. The final action is selected from all possible actions output by the policy network as the resource allocation result for the next preset time period.

[0037] S05: Allocate corresponding resources to the corresponding financial service tenant in the resource pool of the proxy gateway cluster according to the resource allocation result of the next preset time period.

[0038] In this embodiment, it is ensured that resource allocation meets the following constraints: 1) Resource cap: The resource allocation of a single tenant must not exceed the total capacity of the resource pool.

[0039] 2) Fairness, to prevent some tenants from occupying too many resources for a long time.

[0040] 3) Priority: high-priority tenants (such as real-time transaction businesses) are given priority in allocating resources.

[0041] Obtain the resource pool status of the proxy gateway cluster in real time, calculate the resource allocation result of the absolute amount of resources allocated to the corresponding tenant according to the ratio (such as [0.6, 0.3, 0.1]) given in step S04, and calculate the resource allocation result of the absolute amount of resources allocated to the corresponding tenant. For example, the resource allocation of tenant A is: CPU = remaining CPU * 0.6 = 50 * 0.6 = 30 cores.

[0042] Memory = Remaining memory * 0.3 = 100 * 0.3 = 30GB.

[0043] Bandwidth = Remaining bandwidth * 0.1 = 5 * 0.1 = 0.5 Gbps.

[0044] The calculated resource allocation results are updated to the resource allocation table. Based on the resource allocation table, the resource allocation of the proxy gateway corresponding to tenant A is dynamically reconfigured.

[0045] The gateway resource allocation method for financial service tenants of this embodiment uses a traffic feature extraction network model to analyze traffic data from multiple dimensions such as protocol type, packet size, and transmission time to generate a traffic feature sequence for characterizing the demand for financial service resources of each financial service tenant; further classifies the traffic feature sequence, predicts the traffic data type of the financial service tenant in the next time period, and quantifies the priority score to provide a scientific basis for subsequent resource allocation. Utilizing the resource allocation network model, combined with the traffic feature sequence of each tenant and its corresponding predicted traffic data type and priority score of the financial service tenant, the resource demand of each tenant in the next preset time period is predicted, and exclusive resources are dynamically allocated to each financial service tenant to ensure that high-priority traffic types obtain sufficient resources and low-priority traffic types are allocated on demand, thereby improving overall resource utilization and avoiding resource waste or shortage. Therefore, the present invention solves the problem that the existing static resource allocation scheme is difficult to adapt to tenant demand or significant fluctuations in traffic load due to the allocation of fixed resources, which easily leads to service performance degradation or resource waste.

[0046] Optionally, in the above step S02, the traffic data of the financial service tenants is input into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources, including the following steps: S201: Input the traffic data of the financial business tenant into a trained convolutional neural network model for extracting network protocol features and packet size features from the traffic data to obtain traffic identification features of the financial business tenant.

[0047] In this embodiment, the Convolutional Neural Network (CNN) model is a deep learning model specifically used to process data with a grid structure (such as images and videos).

[0048] The CNN model consists of convolutional layers, pooling layers, and fully connected layers. The convolutional layers extract local temporal features (such as protocol modes and packet size variations). The pooling layers reduce the temporal dimension and retain key features. The fully connected layers map temporal features into fixed-dimensional traffic identification features (traffic identification features for financial tenants).

[0049] In this embodiment, the traffic data is converted into a tensor format according to the preprocessing method in step S02 and input into a trained CNN model. This model extracts network protocol features and packet size features from the traffic data, thereby obtaining traffic identification features for financial service tenants.

[0050] S202: Input the traffic identification features of the financial business tenants into a trained long short-term memory network model for extracting the time-varying features of the traffic rate of the traffic data and the time interval features of the arrival of consecutive data packets, to obtain the traffic feature sequence for characterizing the demand for financial business resources of each financial business tenant.

[0051] In this embodiment, the Long Short-Term Memory (LSTM) network is a special recurrent neural network (RNN). By introducing memory cells and a gating mechanism, it effectively solves the vanishing or exploding gradient problems that traditional RNNs suffer when processing long sequences of data. This enables the LSTM network model to capture long-term dependencies in the data.

[0052] The LSTM network model consists of an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer receives input data after preprocessing and feature selection. The LSTM layer uses LSTM units to capture long-term dependencies in the data. The fully connected layer processes the output of the LSTM layer, converting it into a fixed-size vector. This vector can be used for further classification tasks. The output layer converts the output of the fully connected layer into the final classification result. In multi-classification tasks, the Softmax activation function is often used to generate probabilities for each class.

[0053] In this embodiment, the traffic identification features in step S201 are input into the trained LSTM network model, and the time-varying features of the traffic rate of the traffic data and the time interval features of the arrival of consecutive data packets are extracted to obtain the traffic feature sequence used to characterize the demand for financial business resources of each of the financial business tenants.

[0054] The gateway resource allocation method for financial service tenants in this embodiment uses a convolutional layer to extract local features from traffic data, such as protocol patterns (e.g., TCP / UDP features) and packet size distribution. Furthermore, a long short-term memory (LSTM) network is used to capture temporal dependencies in traffic data, such as temporal fluctuations in traffic rate and periodic changes in packet arrival intervals. The synergistic effect of CNN and LSTM enables deep feature extraction and modeling of financial service tenant traffic data, providing a scientific basis for subsequent resource allocation. This method effectively improves overall resource utilization and avoids resource waste or shortages.

[0055] Optionally, in the above-mentioned step S03, that is, classifying the traffic feature sequence used to characterize the financial service resource demand of each financial service tenant to obtain the resource demand traffic type and the traffic priority score corresponding to the resource demand traffic type, the following steps are included: S301: Classify the traffic feature sequence according to the data packet size characteristics and the time-varying characteristics of the traffic rate to obtain high-bandwidth demand traffic.

[0056] In this embodiment, the classification model needs to define a classification label space during training. For example, label 0 corresponds to critical business traffic, label 1 corresponds to low-latency sensitive traffic, label 2 corresponds to high-bandwidth demand traffic, and label 3 corresponds to normal traffic.

[0057] In this embodiment, when training the classification model, tenants' historical traffic data needs to be categorized and labeled. Labels for these data are generated based on threshold rules. Specifically, the classification and labeling rule for high-bandwidth demand traffic is as follows: if the average traffic rate over a certain period exceeds a certain rate threshold, and the size of multiple consecutive data packets exceeds a certain threshold, the traffic will be labeled as high-bandwidth demand traffic.

[0058] S302: Classify the traffic feature sequence according to the characteristics of the arrival time intervals of consecutive data packets to obtain low delay-sensitive traffic.

[0059] In this embodiment, the classification and labeling rule for low-delay-sensitive traffic is: if the time interval between the arrival of consecutive data packets within a certain period of time does not exceed a certain set time interval threshold, the traffic will be marked as low-delay-sensitive traffic.

[0060] S303: Classify the traffic feature sequence according to the network protocol feature or the importance mark to the tenant to obtain key business traffic.

[0061] In this embodiment, the classification and labeling rules for critical business traffic are as follows: if the network protocol types of the tenant traffic data packets within a certain period of time are all in the preset critical business traffic network protocol list, or the tenant is directly in the preset critical business traffic tenant list, then the traffic will be marked as critical business traffic.

[0062] S304: Obtain normal traffic based on the remaining traffic feature sequences that are neither high-bandwidth demand traffic nor low-delay-sensitive traffic nor critical business traffic.

[0063] In this embodiment, the remaining traffic that does not meet the classification marking rule of step S301, nor the classification marking rule of step S302 or the classification marking rule of step S303 is marked as normal traffic.

[0064] S305: The resource demand traffic type is any one of the high bandwidth demand traffic, the low delay sensitivity traffic, the critical business traffic and the common traffic.

[0065] S306: The traffic priority scores of the resource demand traffic types are arranged from large to small in the order of the traffic priority score of the critical business traffic, the traffic priority score of the low delay sensitivity traffic, the traffic priority score of the high bandwidth demand traffic, and the traffic priority score of the ordinary traffic.

[0066] The gateway resource allocation method for financial service tenants in this embodiment systematically categorizes tenant traffic data based on multiple dimensions, including packet size, temporal variations in traffic rate, packet arrival time intervals, network protocol characteristics, and tenant importance tags. Through precise classification, it is possible to clearly distinguish between high-bandwidth demanding traffic (e.g., big data transmission, video conferencing), low-latency sensitive traffic (e.g., real-time transactions, online payments), and critical business traffic (e.g., core database access, transaction system communications). This provides a reliable basis for subsequent resource allocation, enabling the scientific allocation of bandwidth and computing resources, significantly improving the overall utilization efficiency of network resources.

[0067] Optionally, in the above step S03, that is, after the traffic priority scoring corresponding to the resource demand traffic type, the following steps are included: S401: Priority scores are performed on different financial business tenants in the same resource demand flow type to obtain priority scores of the financial business tenants.

[0068] In this embodiment, the weighted sum method is used to calculate the priority score of the financial service tenant: Where n is the number of indicators, is the weight value of the i-th indicator, is the score of the i-th indicator.

[0069] The indicator system in this embodiment includes a business importance index, a resource demand urgency index, and a historical resource usage index. The business importance index is scored based on the proportion of large corporate clients among the financial tenant's clients. The higher the proportion, the higher the score. The resource demand urgency index is scored based on the tenant's business demand for low latency and high bandwidth. The higher the demand, the higher the score. The historical resource usage index is scored based on the tenant's actual utilization efficiency of the allocated resources in the past. The higher the utilization efficiency, the higher the score.

[0070] S402: Add the financial service tenant priority score and the corresponding traffic priority score of the financial service tenant to obtain a target financial service tenant priority score.

[0071] S403: Input the traffic feature sequence, resource demand traffic type and target financial business tenant priority score of each financial business tenant into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

[0072] The gateway resource allocation method for financial service tenants in this embodiment further introduces a financial service tenant priority score to reflect the tenant's business value, and a traffic priority score to reflect the urgency of the tenant's network needs. The combination of the two ensures that resource allocation considers both business value and network needs. This further improves resource utilization and avoids resource waste.

[0073] Optionally, in the above-mentioned step S04, that is, inputting the traffic feature sequence, resource demand traffic type, and traffic priority score of each financial service tenant into the trained resource allocation network model to obtain the resource allocation result of each financial service tenant in the next preset time period, includes the following steps: S501: Sequentially concatenate the traffic feature sequences, resource demand traffic types, and traffic priority scores of the financial service tenants to obtain a resource demand traffic type concatenation sequence.

[0074] In this embodiment, the traffic feature sequence extracted by the traffic feature extraction network model in step S02 for characterizing the demand for financial business resources of each of the financial business tenants, the dense vector generated by the resource demand traffic type obtained in step S03 after Embedding mapping, and the normalized traffic priority score are spliced ​​in sequence after the traffic feature sequence to generate a resource demand traffic type splicing sequence.

[0075] S502: Input the resource demand flow type splicing sequence into the trained resource allocation strategy network model to obtain the expected reward value corresponding to each possible action of increasing or decreasing resources in the resource pool under the current input state.

[0076] In this embodiment, the resource demand flow type sequence of each financial service tenant is input into the policy network trained in step S04, and the expected return value Q of each action (such as increasing or decreasing resources in the resource pool) is output. For example, the action space should be defined as a three-dimensional vector including computing resources, storage resources, and bandwidth resources. Each dimension corresponds to the adjustment range of a resource. Then the action space is: A=(a 计算 , a 存储 , a 带宽 ), where a 计算 Indicates the adjustment range of computing resources, a 存储 Indicates the adjustment range of storage resources, a 带宽Indicates the adjustment range of network bandwidth resources. For each dimension, the discrete action range design, for example, a 计算 =[-2, -1, 0, 1, 2], if the number of CPU cores for resource adjustment is used, then a 计算 The unit is piece, positive value indicates increase of resources, negative value indicates decrease of resources.

[0077] In this embodiment, the value function corresponding to the expected return value Q is: in, , , , are weight coefficients respectively.

[0078] Resource utilization rewards are given based on resource utilization. An ideal utilization range is set. When the actual utilization is within the ideal utilization range, a positive reward is given, and when it deviates, a negative reward is given.

[0079] Throughput rewards are given based on the amount of request data processed. The greater the ratio of actual throughput to theoretical maximum throughput, the greater the positive reward.

[0080] Traffic priority satisfaction rewards are given based on the service quality perceived by tenants. The higher the service evaluation score based on user feedback, the higher the positive reward.

[0081] Cost rewards are given based on the cost changes caused by resource allocation strategies. The lower the cost, the higher the reward.

[0082] S503: Selecting an action with the largest expected return value according to the expected return value, and obtaining a resource allocation result for the next preset time period.

[0083] The gateway resource allocation method for financial service tenants of this embodiment dynamically allocates resources through the above-mentioned reinforcement learning model and selects the optimal resource allocation result, thereby further significantly improving resource utilization efficiency, optimizing costs, and enhancing system stability.

[0084] Optionally, the above step S05, i.e., the pooling process of the resource pool of the proxy gateway cluster, includes the following steps: S601: The resources of the proxy gateway cluster include computing resources, storage resources and network resources.

[0085] S602: The computing resources of the physical server are abstracted into multiple containers through the container orchestration platform, where each container represents an independent computing unit, to obtain a computing resource pool.

[0086] In this embodiment, the dynamic allocation and isolation of computing resources is achieved through the container orchestration platform Kubernetes. Specifically, the computing resources of the physical server are abstracted into containers, each container acts as an independent computing unit, and all these computing units together form a computing resource pool.

[0087] S603: Abstracting the storage resources of the physical storage device into multiple distributed storage nodes through the distributed storage system to obtain a storage resource pool.

[0088] In this embodiment, the distributed storage system Ceph is used to implement pooled management of storage resources. Specifically, the resources of physical storage devices are abstracted into distributed storage pools, enabling flexible expansion and contraction on demand. Furthermore, data sharding and redundancy mechanisms effectively ensure high availability and reliability of stored data.

[0089] S604: Abstracting the bandwidth resources of the physical network device into multiple virtual networks through software-defined networking to obtain a network resource pool.

[0090] In this embodiment, dynamic allocation and isolation of network bandwidth are achieved through software-defined networking (SDN) technology. Specifically, the bandwidth resources of physical network devices are abstracted into a virtual network resource pool, thereby supporting flexible allocation and recycling based on demand.

[0091] S605: Obtain the resource pool of the proxy gateway cluster according to the computing resource pool, the storage resource pool, and the network resource pool.

[0092] The gateway resource allocation method for financial service tenants in this embodiment realizes dynamic allocation and isolation of resources by pooling the computing resources, storage resources and bandwidth resources in the proxy gateway cluster, thereby further improving resource utilization, service stability and operation and maintenance efficiency.

[0093] Optionally, in step S02 above, the traffic data of the financial service tenants is input into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources, including the following steps: S701: Obtain tenant identifiers of N financial service tenants of the proxy gateway cluster within the current historical time period.

[0094] In this embodiment, when a financial service tenant registers, a unique tenant ID is assigned to it, a tenant ID list is obtained, and each tenant ID in the tenant ID list is associated with other information of the corresponding tenant (such as IP address) and stored.

[0095] S702: Concatenate the tenant identifier of each financial service tenant with the data belonging to the corresponding tenant identifier in the traffic data to obtain traffic reconstruction data of N financial service tenants.

[0096] In this embodiment, the tenant identifier is associated and matched with the records belonging to the tenant in the traffic data. For example, each tenant identifier may correspond to a list of IP addresses or an IP address range belonging to the current tenant. If the IP address in the traffic data is within the IP address list or within the IP address range, the tenant identifier and the matching traffic data are concatenated into traffic reconstruction data in a unified format, ultimately generating traffic reconstruction data for N financial service tenants.

[0097] S703: Input the traffic reconstruction data of the N financial business tenants into the trained traffic feature extraction network model to obtain N traffic feature sequences for characterizing the demand of each financial business tenant for financial business resources; the traffic feature sequence includes the tenant identifier.

[0098] In this embodiment, after preprocessing the traffic reconstruction data of the N financial service tenants involved in the steps, it is uniformly input into a trained traffic feature extraction network model. This model can simultaneously extract features from the traffic reconstruction data of the N financial service tenants and output N traffic feature sequences that represent the demand for financial service resources by each financial service tenant.

[0099] This embodiment provides a gateway resource allocation method for financial service tenants. By appending each tenant's identifier to the corresponding tenant's traffic data, the traffic data is reconstructed, thereby generating traffic reconstructed data containing the tenant identifier. Subsequently, the reconstructed traffic data from N financial service tenants is input into a trained traffic feature extraction network model, and feature extraction is performed on these tenants' traffic reconstructed data simultaneously. This process significantly improves the efficiency of the traffic feature extraction network model in generating feature sequences.

[0100] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0101] In one embodiment, a gateway resource allocation device for a financial service tenant is provided, and the gateway resource allocation device for a financial service tenant corresponds one-to-one to the gateway resource allocation method for a financial service tenant in the above embodiment. Figure 3 As shown, the intelligent question-answering processing device includes an acquisition module 101, a feature extraction module 102, a classification module 103, a resource allocation prediction module 104, and a resource allocation module 105. The functional modules are described in detail as follows: The acquisition module is used to obtain the traffic data of N financial business tenants of the proxy gateway cluster in the current historical time period, where N>0; the traffic data includes the protocol type of the data packet, the data packet size, and the sending time of the data packet.

[0102] The feature extraction module is used to input the traffic data of the financial service tenants into the trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources.

[0103] The classification module is used to classify the traffic feature sequence used to characterize the financial service resource demand of each financial service tenant to obtain the resource demand traffic type and the traffic priority score corresponding to the resource demand traffic type.

[0104] The resource allocation prediction module is used to input the traffic feature sequence, resource demand traffic type and traffic priority score of each financial business tenant into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

[0105] The resource allocation module is used to allocate corresponding resources to the corresponding financial service tenant in the resource pool of the proxy gateway cluster according to the resource allocation result of the next preset time period.

[0106] Optionally, the feature extraction module 102 specifically includes: The local feature extraction submodule is used to input the traffic data of the financial business tenant into a trained convolutional neural network model for extracting network protocol features and packet size features from the traffic data to obtain the traffic identification features of the financial business tenant.

[0107] The entire feature extraction module is used to input the traffic identification features of the financial business tenants into a trained long short-term memory network model for extracting the time-varying features of the traffic rate of the traffic data and the time interval features of the arrival of consecutive data packets, so as to obtain the traffic feature sequence used to characterize the demand for financial business resources of each financial business tenant.

[0108] Optionally, the classification module 103 specifically includes: The high-bandwidth demand classification submodule is used to classify the traffic feature sequence according to the data packet size characteristics and the time-varying characteristics of the traffic rate to obtain high-bandwidth demand traffic.

[0109] The low delay sensitivity classification submodule is used to classify the traffic feature sequence according to the characteristics of the arrival time interval of the continuous data packets to obtain low delay sensitivity traffic.

[0110] The key business classification submodule is used to classify the traffic feature sequence according to the network protocol characteristics or the importance mark to the tenant to obtain key business traffic.

[0111] The ordinary traffic classification submodule is used to obtain ordinary traffic based on the remaining traffic feature sequences that are neither high-bandwidth demand traffic nor low-delay sensitive traffic or critical business traffic.

[0112] The traffic type submodule is used when the resource demand traffic type is any one of the high bandwidth demand traffic, the low delay sensitivity traffic, the critical business traffic and the ordinary traffic.

[0113] The priority scoring submodule is used to sort the traffic priority scores of the resource demand traffic types from large to small, namely, the traffic priority score of the critical business traffic, the traffic priority score of the low delay sensitivity traffic, the traffic priority score of the high bandwidth demand traffic, and the traffic priority score of the ordinary traffic.

[0114] Optionally, the classification module 103 specifically includes: The tenant priority scoring submodule is used to perform priority scoring on different financial business tenants in the same resource demand traffic type to obtain the financial business tenant priority score.

[0115] a priority score processing submodule, configured to add the financial service tenant priority score to the traffic priority score corresponding to the financial service tenant to obtain a target financial service tenant priority score; The resource allocation prediction submodule is used to input the traffic feature sequence, resource demand traffic type and target financial business tenant priority score of each financial business tenant into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

[0116] Optionally, the resource allocation prediction module 104 specifically includes: The splicing submodule is used to sequentially splice the traffic feature sequence, resource demand traffic type and traffic priority score of each financial service tenant to obtain a resource demand traffic type splicing sequence.

[0117] The action prediction submodule is used to input the resource demand flow type splicing sequence into the trained resource allocation strategy network model to obtain the expected reward value corresponding to each possible action of increasing or decreasing resources in the resource pool under the current input state.

[0118] The action output submodule is configured to select an action with the largest expected return value according to the expected return value, and obtain a resource allocation result for the next preset time period.

[0119] Optionally, the resource allocation module 105 specifically includes: The proxy gateway cluster submodule uses resources of the proxy gateway cluster including computing resources, storage resources and network resources.

[0120] The computing resource pooling submodule is used to abstract the computing resources of the physical server into multiple containers through the container orchestration platform. Each container represents an independent computing unit, thus obtaining a computing resource pool.

[0121] The storage resource pooling submodule is used to abstract the storage resources of the physical storage device into multiple distributed storage nodes through the distributed storage system to obtain a storage resource pool.

[0122] The network resource pooling submodule is used to abstract the bandwidth resources of physical network devices into multiple virtual networks through software-defined networking to obtain a network resource pool.

[0123] The resource pooling submodule is used to obtain the resource pool of the proxy gateway cluster according to the computing resource pool, the storage resource pool and the network resource pool.

[0124] Optionally, the feature extraction module 102 specifically includes: The tenant identification acquisition submodule is used to obtain the tenant identifications of N financial business tenants of the proxy gateway cluster within the current historical time period.

[0125] The traffic reconstruction submodule is used to splice the tenant identification of each financial business tenant with the data belonging to the corresponding tenant identification in the traffic data to obtain traffic reconstruction data of N financial business tenants.

[0126] The traffic feature extraction submodule is used to input the traffic reconstruction data of N financial business tenants into the trained traffic feature extraction network model to obtain N traffic feature sequences used to characterize the demand of each financial business tenant for financial business resources; the traffic feature sequence includes the tenant identifier.

[0127] The present invention provides a gateway resource allocation device for financial service tenants. By extracting a network model based on traffic characteristics, the device analyzes traffic data from multiple dimensions, such as protocol type, packet size, and transmission time, to generate a traffic feature sequence that characterizes the demand for financial service resources by each financial service tenant. The device further classifies the traffic feature sequence to predict the traffic data type of the financial service tenant in the next time period, and quantifies the priority score to provide a scientific basis for subsequent resource allocation. The device utilizes a resource allocation network model, combines the traffic feature sequence of each tenant, and its corresponding predicted traffic data type and priority score of the financial service tenant, to predict the resource demand of each tenant in the next preset time period, and dynamically allocates exclusive resources to each financial service tenant, ensuring that high-priority traffic types obtain sufficient resources and low-priority traffic types are allocated on demand, thereby improving overall resource utilization and avoiding resource waste or shortage. Therefore, the present invention solves the problem that existing static resource allocation schemes, due to the allocation of fixed resources, are difficult to adapt to tenant demand or significant fluctuations in traffic load, which can easily lead to service performance degradation or resource waste.

[0128] For the specific definition of the gateway resource allocation device for financial business tenants, please refer to the definition of the gateway resource allocation method for financial business tenants above, which will not be repeated here. The various modules in the above-mentioned gateway resource allocation device for financial business tenants can be implemented in whole or in part by software, hardware and their combination. 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 of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0129] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or 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 network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the service side of a gateway resource allocation method for a financial business tenant.

[0130] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 4As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. 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 and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the client side of a gateway resource allocation method for a financial service tenant. In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed: Obtain traffic data for N financial service tenants of the proxy gateway cluster during the current historical time period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet.

[0131] The traffic data of the financial service tenants is input into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources.

[0132] By classifying the traffic feature sequences used to characterize the financial service resource demands of the financial service tenants, resource demand traffic types and traffic priority scores corresponding to the resource demand traffic types are obtained.

[0133] The traffic feature sequence, resource demand traffic type and traffic priority score of each financial business tenant are input into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

[0134] According to the resource allocation result of the next preset time period, corresponding resources are allocated to the corresponding financial service tenant in the resource pool of the proxy gateway cluster.

[0135] 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 following steps are implemented: Obtain traffic data for N financial service tenants of the proxy gateway cluster during the current historical time period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet.

[0136] The traffic data of the financial service tenants is input into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources.

[0137] By classifying the traffic feature sequences used to characterize the financial service resource demands of the financial service tenants, resource demand traffic types and traffic priority scores corresponding to the resource demand traffic types are obtained.

[0138] The traffic feature sequence, resource demand traffic type and traffic priority score of each financial business tenant are input into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

[0139] According to the resource allocation result of the next preset time period, corresponding resources are allocated to the corresponding financial service tenant in the resource pool of the proxy gateway cluster.

[0140] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0141] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. 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 memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0142] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0143] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A gateway resource allocation method for financial service tenants, characterized in that: include: Obtain traffic data for N financial service tenants in the proxy gateway cluster during the current historical period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet. Inputting the traffic data of the financial service tenants into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources; By classifying the traffic feature sequences used to characterize the financial service resource demands of each of the financial service tenants, a resource demand traffic type and a traffic priority score corresponding to the resource demand traffic type are obtained; Inputting the traffic feature sequence, resource demand traffic type, and traffic priority score of each financial service tenant into the trained resource allocation network model to obtain the resource allocation result for the next preset time period of each financial service tenant; According to the resource allocation result of the next preset time period, corresponding resources are allocated to the corresponding financial service tenant in the resource pool of the proxy gateway cluster.

2. The gateway resource allocation method for financial service tenants according to claim 1, characterized in that: Inputting the traffic data of the financial service tenants into the trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources includes: Inputting the traffic data of the financial service tenant into a convolutional neural network model trained to extract network protocol features and packet size features from the traffic data to obtain traffic identification features of the financial service tenant; The traffic identification features of the financial business tenants are input into a trained long short-term memory network model for extracting the time-varying features of the traffic rate of the traffic data and the time interval features of the arrival of consecutive data packets, so as to obtain the traffic feature sequence for characterizing the demand for financial business resources of each financial business tenant.

3. The gateway resource allocation method for financial service tenants according to any one of claim 2, characterized in that: The classifying of the traffic feature sequences used to characterize the financial service resource demands of the financial service tenants to obtain resource demand traffic types and traffic priority scores corresponding to the resource demand traffic types includes: Classifying the traffic feature sequence according to the data packet size characteristics and the time-varying characteristics of the traffic rate to obtain high-bandwidth demand traffic; Classifying the traffic feature sequence according to the arrival time interval characteristics of the consecutive data packets to obtain low delay sensitivity traffic; Classifying the traffic feature sequence according to the network protocol features or the importance mark to the tenant to obtain key business traffic; Obtaining normal traffic based on the remaining traffic feature sequences that are neither high-bandwidth demand traffic nor low-delay sensitive traffic nor critical business traffic; The resource demand traffic type is any one of the high bandwidth demand traffic, the low delay sensitivity traffic, the critical business traffic and the ordinary traffic; The traffic priority scores of the resource demand traffic types are arranged from large to small as follows: the traffic priority score of the critical business traffic, the traffic priority score of the low delay sensitive traffic, the traffic priority score of the high bandwidth demand traffic, and the traffic priority score of the ordinary traffic.

4. The gateway resource allocation method for financial service tenants according to claim 3, characterized in that: After the traffic priority score corresponding to the resource demand traffic type is scored, the following steps are included: Priority scores are assigned to different financial business tenants in the same resource demand flow type to obtain priority scores for financial business tenants; Adding the financial service tenant priority score to the corresponding traffic priority score of the financial service tenant to obtain the target financial service tenant priority score; The traffic feature sequence, resource demand traffic type and target financial business tenant priority score of each financial business tenant are input into the trained resource allocation network model to obtain the resource allocation result of each financial business tenant in the next preset time period.

5. The gateway resource allocation method for financial service tenants according to claim 1, characterized in that: Inputting the traffic feature sequence, resource demand traffic type, and traffic priority score of each financial service tenant into a trained resource allocation network model to obtain a resource allocation result for each financial service tenant in the next preset time period includes: Sequentially splicing the traffic feature sequences, resource demand traffic types, and traffic priority scores of the respective financial service tenants to obtain a resource demand traffic type splicing sequence; Inputting the resource demand flow type splicing sequence into the trained resource allocation strategy network model to obtain the expected return value corresponding to each possible action of increasing or decreasing resources in the resource pool under the current input state; An action with the largest expected return value is selected according to the expected return value to obtain a resource allocation result for the next preset time period.

6. The gateway resource allocation method for financial service tenants according to claim 1, characterized in that: The pooling process of the resource pool of the proxy gateway cluster includes: The resources of the proxy gateway cluster include computing resources, storage resources and network resources; The container orchestration platform abstracts the computing resources of physical servers into multiple containers, each representing an independent computing unit, to form a computing resource pool. The storage resources of the physical storage device are abstracted into multiple distributed storage nodes through the distributed storage system to obtain a storage resource pool; Through software-defined networking, the bandwidth resources of physical network devices are abstracted into multiple virtual networks to obtain a network resource pool; The resource pool of the proxy gateway cluster is obtained according to the computing resource pool, the storage resource pool and the network resource pool.

7. The gateway resource allocation method for financial service tenants according to claim 2, characterized in that: Inputting the traffic data of the financial service tenants into the trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources includes: Get the tenant IDs of N financial business tenants in the proxy gateway cluster within the current historical time period; splicing the tenant identifier of each financial service tenant with the data belonging to the corresponding tenant identifier in the traffic data to obtain traffic reconstruction data of N financial service tenants; The traffic reconstruction data of N financial business tenants are input into a trained traffic feature extraction network model to obtain N traffic feature sequences for characterizing the demand of each financial business tenant for financial business resources; the traffic feature sequences include the tenant identifier.

8. A gateway resource allocation device for financial service tenants, characterized in that: include: An acquisition module is used to obtain the traffic data of N financial business tenants of the proxy gateway cluster in the current historical time period, where N>0; the traffic data includes the protocol type, size, and sending time of the data packet; A feature extraction module, configured to input the traffic data of the financial service tenants into a trained traffic feature extraction network model to obtain a traffic feature sequence for characterizing the demand of each financial service tenant for financial service resources; a classification module, configured to classify the traffic feature sequences characterizing the financial service resource demands of the financial service tenants to obtain resource demand traffic types and traffic priority scores corresponding to the resource demand traffic types; A resource allocation prediction module, configured to input the traffic feature sequence, resource demand traffic type, and traffic priority score of each of the financial service tenants into a trained resource allocation network model to obtain a resource allocation result for each of the financial service tenants in the next preset time period; The resource allocation module is used to allocate corresponding resources to the corresponding financial service tenant in the resource pool of the proxy gateway cluster according to the resource allocation result of the next preset time period.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the gateway resource allocation method for financial service tenants according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the gateway resource allocation method for financial service tenants according to any one of claims 1 to 7 are implemented.

Images