Management system based on enterprise data processing

Abstract

The invention relates to the technical field of data processing, in particular to a management system based on enterprise data processing, which comprises a feature extraction module, a dynamic resource portrait module and a self-adaptive routing decision module, the multi-dimensional feature data comprises a time delay sensitivity identification value and a calculation intensity identification value; the dynamic resource portrait module triggers load evaluation strategy switching in response to the time delay sensitivity identification value, and generates a resource capacity vector of the target node; the adaptive routing decision-making module adjusts a root node in the optimized node decision-making tree according to the time delay sensitivity identification value, evaluates a transmission path of the enterprise data flow in the optimized node decision-making tree in combination with a resource capacity vector of a target node, and generates an optimal transmission path; wherein the optimal data transmission path enables the high-delay sensitive enterprise data flow to preferentially occupy the computing node resources occupied by the low storage bandwidth, and is used for improving the enterprise data processing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and specifically to a management system based on enterprise data processing. Background Art

[0002] With the acceleration of the digital transformation of enterprises, the real-time performance of data processing, resource utilization efficiency, and task priority management have become the core challenges of enterprise information systems. Traditional enterprise data processing systems usually adopt static resource allocation and fixed routing strategies, which are difficult to adapt to the dynamically changing business requirements. For example, in the prior art, the load evaluation of computing nodes is mostly based on periodic resource sampling (such as CPU and memory utilization), lacking the ability to perceive business characteristics in real time, resulting in the inability to achieve differential scheduling when high-latency sensitive tasks (such as financial transactions and real-time monitoring) and low-sensitive tasks (such as batch report generation) compete for resources.

[0003] In addition, traditional routing strategies often rely on preset load balancing rules (such as round-robin and least connections), without optimizing the dynamic requirements of storage bandwidth and computing resources in combination with business types, which easily causes high-sensitive tasks to be delayed due to insufficient resource preemption, or low-sensitive tasks to over-occupy high-value node resources. Although there are hierarchical scheduling methods based on QoS (Quality of Service) in the prior art, their classification dimensions are single (such as only relying on business types or fixed priorities), unable to quantify dynamic characteristics such as the delay sensitivity and computing intensity of data streams, and lacking the ability to dynamically adapt to node capacity thresholds during resource profiling modeling (for example, not dynamically adjusting the available rate threshold of computing units according to task types).

[0004] At the same time, the real-time acquisition and feature extraction of multi-source heterogeneous data streams often introduce additional delays due to cumbersome processing frameworks (such as relying on offline batch processing), further exacerbating the response bottleneck of high-sensitive tasks. These problems lead to significant deficiencies in the resource utilization rate, task processing efficiency, and service quality guarantee of enterprise data processing systems. Summary of the Invention

[0005] The purpose of the present invention is to provide a management system based on enterprise data processing to solve the problems raised in the above background art, where the specific technical problem is the inability to dynamically distinguish the resource requirements of high-latency sensitive tasks and low-sensitive tasks, resulting in high-sensitive tasks being delayed due to insufficient resource preemption, while low-sensitive tasks may occupy high-value node resources, causing resource misallocation and a decline in critical business performance.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A management system based on enterprise data processing, including a feature extraction module, a dynamic resource profiling module, and an adaptive routing decision module, where:

[0007] The feature extraction module obtains enterprise data streams in real time and generates multi-dimensional feature data, where the multi-dimensional feature data includes a delay sensitivity identification value and a computing intensity identification value;

[0008] The dynamic resource profiling module triggers a load evaluation policy switch in response to the delay sensitivity identification value and generates a resource capacity vector of the target node;

[0009] The adaptive routing decision module adjusts the root node in the optimized node decision tree according to the delay sensitivity identification value, and evaluates the transmission path of the enterprise data stream in the optimized node decision tree in combination with the resource capacity vector of the target node, and generates an optimal transmission path, where the optimal data transmission path enables high-delay sensitive enterprise data streams to preferentially occupy the computing node resources with low storage bandwidth occupancy, so as to improve the enterprise data processing efficiency;

[0010] The calculation formula of the delay sensitivity identification value is as follows:

[0011]

[0012] is the delay sensitivity identification value, indicating the delay sensitivity level of the business to which the enterprise data stream belongs; is the current time point; 、 and correspond to the weight coefficients of the business type base value, jitter rate, and path dependence factor respectively; is the business type base value, which is a fixed sensitivity level mapped according to the business type to which the enterprise data stream belongs; represents the standard deviation of the packet arrival interval time of the enterprise data stream, which is used to quantify the jitter rate; represents the maximum allowable jitter threshold, which is used to normalize the jitter rate; represents the path dependence factor, indicating whether the enterprise data stream is on the critical service call chain;

[0013] By combining the delay sensitivity identification value with the adaptive routing decision, it is ensured that high-sensitive tasks preferentially occupy low-storage bandwidth resources, solving the problem of resource mismatch; the delay sensitivity identification value is calculated in real time, and the delay sensitivity (business type, jitter rate, path dependence) is quantified through the formula, avoiding misjudgment caused by single-dimensional (such as only business type) classification; by introducing the path dependence factor (such as whether it is on the critical service call chain), high-sensitive tasks are accurately identified, avoiding the deficiencies of traditional static classification;

[0014] According to the delay sensitivity identification value, the root node of the optimized node decision tree is divided into high, medium, and low sensitivity branches, and in combination with the resource capacity vector of the target node, a node path with low storage bandwidth occupancy is preferentially selected for high-sensitive tasks.

[0015] As a further improvement of this technical solution, the dynamic resource profiling module includes a delay policy switching unit, and the delay policy switching unit is used for switching the load evaluation policy, specifically including:

[0016] According to the value of the delay sensitivity identification determine the delay sensitivity level of the target node required by the enterprise data stream belonging to the service, where the delay sensitivity level includes low-sensitivity service, medium-sensitivity service, and high-sensitivity service;

[0017] According to the delay sensitivity level, trigger the corresponding load evaluation policy to generate the resource capacity vector of the target node, where the load evaluation policies are as follows:

[0018] When the delay sensitivity level is a high-sensitivity service, trigger millisecond-level real-time monitoring and adopt a hard real-time algorithm to ensure extremely low latency;

[0019] When the delay sensitivity level is a medium-sensitivity service, trigger second-level periodic sampling and allocate resources based on the weighted load balancing algorithm;

[0020] When the delay sensitivity level is a low-sensitivity service, trigger minute-level predictive monitoring and use the greedy algorithm to maximize resource utilization;

[0021] Through the switching of the load evaluation policy, high-sensitivity tasks can obtain millisecond-level resource preemption capabilities, low-sensitivity tasks can reduce monitoring overhead, and overall resource allocation efficiency is improved.

[0022] As a further improvement of this technical solution, the dynamic resource profiling module includes a threshold determination unit, and the threshold determination unit is used for generating the available rate threshold of the computing unit and the occupancy rate threshold of the storage bandwidth, specifically including:

[0023] According to the delay sensitivity level, limit the available rate of the computing unit and the occupancy rate of the storage bandwidth of the target node to obtain the available rate threshold of the computing unit and the occupancy rate threshold of the storage bandwidth of the target node. The specific process is as follows:

[0024] When the delay sensitivity level is a high-sensitivity service, and ;

[0025] When the delay sensitivity level is a medium-sensitivity service, and ;

[0026] When the delay sensitivity level is a low-sensitivity service, and ; where represents the available rate threshold of the computing unit, represents the occupancy rate threshold of the storage bandwidth, 、 and represents the available rate threshold of the historical computing unit, and ; 、 and represent the occupied rate thresholds of the historical storage bandwidth, and ; 、 and respectively represent the identifiers of high-sensitivity services, medium-sensitivity services, and low-sensitivity services;

[0027] Through this differential threshold constraint, high-sensitivity tasks are forced to be routed to nodes with low storage bandwidth occupancy (such as edge computing nodes), while low-sensitivity tasks are assigned to nodes with high bandwidth occupancy (such as centralized storage clusters), isolating critical and non-critical tasks at the physical resource level.

[0028] As a further improvement of this technical solution, the adaptive routing decision module includes a path determination unit, and the path determination unit is used to generate an optimal transmission path, specifically including:

[0029] According to the latency sensitivity identifier of the enterprise data stream , select the corresponding root node branch, and obtain the available rate threshold of the computing unit in the resource capacity vector defined in the threshold determination unit (202) and the occupied rate threshold of the storage bandwidth , select the nodes in the optimized node decision tree path that meet the available rate threshold of the computing unit and the occupied rate threshold of the storage bandwidth , and perform a weight evaluation on all path sets formed by this node, and select the path with the smallest weight as the optimal transmission path;

[0030] In the routing decision, combine the available rate threshold of the computing unit in the resource capacity vector and the occupied rate threshold of the storage bandwidth to screen paths, and select the path with the smallest weight (such as the shortest latency, the lowest number of hops) to further reduce the transmission latency of high-sensitivity tasks.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] By dynamically identifying the latency sensitivity identifier value and the computing intensity identifier value of the enterprise data stream, combined with the adaptive routing decision and the dynamic resource profiling mechanism, accurate resource isolation and priority scheduling of high-sensitivity tasks and low-sensitivity tasks are achieved;

[0033] High-sensitivity tasks are preferentially assigned to node paths with low storage bandwidth occupancy to reduce latency caused by resource competition; the collaborative adaptation of the dynamic load assessment strategy and the resource capacity threshold ensures that critical tasks exclusively occupy highly available resources in real time, while non-critical tasks flexibly utilize redundant resources to improve the overall resource utilization rate;

[0034] The management system based on enterprise data processing effectively solves the problem of resource mismatch caused by static resource allocation in traditional systems, significantly reduces the processing latency of high-sensitivity services, ensures the stability and real-time performance of critical service links, and comprehensively improves the efficiency of enterprise data processing and service quality. Brief Description of the Drawings

[0035] Figure 1 It is a schematic diagram of the overall module of the present invention;

[0036] Figure 2 It is a schematic diagram of the unit of the dynamic resource profiling module of the present invention;

[0037] Figure 3 It is a schematic diagram of the unit of the adaptive routing decision module of the present invention.

[0038] In the figure: 100, feature extraction module; 200, dynamic resource profiling module; 201, delay policy switching unit; 202, threshold determination unit; 300, adaptive routing decision module; 301, decision tree unit; 302, path determination unit. Detailed Embodiments

[0039] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] Next, please refer to Figure 1 , the present invention provides a technical solution: a management system based on enterprise data processing, including a feature extraction module 100, a dynamic resource profiling module 200, and an adaptive routing decision module 300.

[0041] The feature extraction module 100 uses streaming collection to obtain enterprise data streams in real time. By accessing multi-source heterogeneous data in real time (such as terminal device, server log, and API data), it performs batch processing on the original stream using dynamic window segmentation (such as a 100ms time window), executes data cleaning, feature extraction, calculates the latency sensitivity identification value and the computational intensity identification value in real time, and generates multi-dimensional feature data; combines a lightweight framework to achieve millisecond-level response. At the same time, through traffic classification (real-time / quasi-real-time / batch) and elastic resource allocation (independent thread pool, token bucket rate limiting), it ensures that high-sensitivity tasks are processed first, ensuring continuous and stable injection of data into the downstream decision-making module; the multi-dimensional feature data includes the latency sensitivity identification value and the computational intensity identification value, specifically including:

[0042] The latency sensitivity identification value is an indicator used to measure the degree of response time requirements for a specific task or data stream. In an enterprise environment, different business activities may have different levels of requirements for the system's response speed. For example, online trading services require extremely fast responses to ensure the user experience; in contrast, data analysis report generation can accept longer waiting times. Therefore, by setting corresponding latency sensitivity identifications for each task, it is used to allow the corresponding enterprise data to obtain processing resources first to ensure its quick completion; the calculation formula for the latency sensitivity identification value is as follows:

[0043]

[0044] is the latency sensitivity identification value, indicating the latency sensitivity of the business to which the enterprise data stream belongs. The larger the value, the higher the priority; is the current time point, indicating the real-time nature of dynamic calculation; 、 and correspond to the weight coefficients of the business type base value, jitter rate, and path dependence factor respectively; is the business type base value, which is the fixed sensitivity level mapped according to the business type to which the enterprise data stream belongs; represents the standard deviation of the packet arrival interval time of the enterprise data stream, used to quantify the jitter rate; represents the maximum allowable jitter threshold, used to normalize the jitter rate; represents the path dependence factor, which is a boolean variable (0 or 1), indicating whether the enterprise data stream is on the critical service call chain;

[0045] Among them, the computational intensity identification value reflects the amount of computing resources required to execute a certain task. Some tasks, such as image recognition and big data analysis, usually consume a large amount of CPU / GPU computing power, while other relatively simple query requests may only require very little computing power. By tagging each task with a computational intensity label, the scheduling algorithm can allocate limited computing resources more intelligently, thereby improving the operating efficiency of the entire system. The calculation formula of the computational intensity identification value is as follows:

[0046] , where represents the computational intensity identification value, indicating the consumption intensity of computing resources by the business to which the enterprise data stream belongs. The larger the value, the more intensive the task; is the basic value of the business type to which the enterprise data stream belongs, which is a fixed intensity score mapped according to the type of operation instruction; is the historical resource consumption pattern vector, indicating the average resource consumption of the business type to which the enterprise data stream belongs in historical data; is the current task resource demand vector, indicating the real-time resource demand of the business type to which the enterprise data stream belongs; is the load balancing coefficient; is the historical average CPU utilization rate of the business type to which the enterprise data stream belongs on the node; is the current CPU utilization rate of the business type to which the enterprise data stream belongs on the node.

[0047] The multi-dimensional feature data generated by the feature extraction module 100 is represented as .

[0048] The delay policy switching unit 201 in the dynamic resource profiling module 200 responds to the delay sensitivity identification value to trigger the switching of the load evaluation policy, specifically including:

[0049] According to the size of the delay sensitivity identification value , determine the delay sensitivity level of the target node required by the business to which the enterprise data stream belongs. The delay sensitivity levels include low-sensitivity services, medium-sensitivity services, and high-sensitivity services; among them:

[0050] The delay sensitivity level uses hierarchical clustering to analyze historical enterprise data streams. The historical enterprise data streams include the basic value of the business type, the jitter rate, and the path dependence factor. Group those with similar jitter rates and path dependencies into one category. Through hierarchical clustering, identify the natural groupings of different business types. Each group represents a delay sensitivity level. For example, data streams with a low jitter rate are classified as high-sensitivity services, and data streams with a high jitter rate are classified as low-sensitivity services; according to the clustering results, set the threshold of the delay sensitivity identification value , for example is a low-sensitivity service, Highly sensitive services are highly sensitive services;

[0051] According to the latency sensitivity level, trigger the corresponding load evaluation strategy to generate the resource capacity vector of the target node. The load evaluation strategy is as follows:

[0052] When the latency sensitivity level is a highly sensitive service, trigger millisecond-level real-time monitoring and adopt a hard real-time algorithm to ensure extremely low latency;

[0053] When the latency sensitivity level is a medium-sensitive service, trigger second-level periodic sampling and allocate resources based on the weighted load balancing algorithm;

[0054] When the latency sensitivity level is a low-sensitive service, trigger minute-level predictive monitoring and use the greedy algorithm to maximize resource utilization.

[0055] The threshold determination unit 202 in the dynamic resource profiling module 200 is used to generate the resource capacity vector of the target node. The resource capacity vector includes the available rate threshold of the computing unit and the occupancy rate threshold of the storage bandwidth, specifically including:

[0056] According to the latency sensitivity level, limit the available rate of the computing unit and the occupancy rate of the storage bandwidth of the target node to obtain the available rate threshold of the computing unit and the occupancy rate threshold of the storage bandwidth of the target node. The specific process is as follows:

[0057] When the latency sensitivity level is a highly sensitive service, and ;

[0058] When the latency sensitivity level is a medium-sensitive service, and ;

[0059] When the latency sensitivity level is a low-sensitive service, and ; where represents the available rate threshold of the computing unit, represents the occupancy rate threshold of the storage bandwidth, 、 and represent the historical available rate threshold of the computing unit, and ; 、 and represent the historical occupancy rate threshold of the storage bandwidth, and ; 、 and represent the highly sensitive service, medium-sensitive service, and low-sensitive service identifiers respectively.

[0060] The decision tree unit 301 in the adaptive routing decision module 300 adjusts and optimizes the root node in the node decision tree according to the delay sensitivity identification value Specifically, it includes:

[0061] The optimized node decision tree includes all data transmission paths and nodes. Each node represents a computing node or a network device, and the edges represent the connections between the nodes. Among them, the root node branches of the optimized node decision tree are divided according to the delay sensitivity level determined by the delay sensitivity identification value. According to the delay sensitivity level, including low-sensitivity services, medium-sensitivity services, and high-sensitivity services, the root node branches are divided into high-sensitivity branches, medium-sensitivity branches, and low-sensitivity branches.

[0062] The path determination unit 302 in the adaptive routing decision module 300 evaluates the transmission path of the enterprise data stream in the optimized node decision tree in combination with the resource capacity vector of the target node and generates the optimal transmission path. Specifically, it includes:

[0063] According to the delay sensitivity identification of the enterprise data stream , select the corresponding root node branch, and obtain the available rate threshold of the computing unit in the resource capacity vector limited by the threshold determination unit 202 and the storage bandwidth occupancy rate threshold , select the nodes in the path of the optimized node decision tree that meet the available rate threshold of the computing unit and the storage bandwidth occupancy rate threshold , evaluate the weights of all path sets formed by these nodes, and select the path with the smallest weight as the optimal transmission path.

[0064] By screening the nodes that meet the available rate of the computing unit and the storage bandwidth occupancy rate threshold, the system can preferentially allocate nodes with sufficient resources and stable loads to high-sensitivity services, avoid resource contention and overload risks. At the same time, combined with path weight evaluation (such as node processing capacity, network latency, etc.), dynamically select the optimal data transmission path, reduce the delay of critical services, improve resource utilization, and ensure that tasks with different sensitivity levels run efficiently in isolated resource pools, ultimately achieving global load balancing and precise guarantee of service quality (QoS); the optimal data transmission path enables high-delay-sensitive enterprise data streams to preferentially occupy the computing node resources with low storage bandwidth occupancy to improve the data processing efficiency of the system.

[0065] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. A management system based on enterprise data processing, characterized in that It includes a feature extraction module (100), a dynamic resource portrait module (200), and an adaptive routing decision module (300), where: The feature extraction module (100) obtains enterprise data streams in real time and generates multi-dimensional feature data, and the multi-dimensional feature data includes a delay sensitivity identification value and a computing intensity identification value; The dynamic resource portrait module (200) triggers a load evaluation policy switch in response to the delay sensitivity identification value and generates a resource capacity vector of a target node; The adaptive routing decision module (300) adjusts the root node in the optimized node decision tree according to the delay sensitivity identification value, and evaluates the transmission path of the enterprise data stream in the optimized node decision tree in combination with the resource capacity vector of the target node, and generates an optimal transmission path, where the optimal data transmission path enables high-delay sensitive enterprise data streams to preferentially occupy the computing node resources with low storage bandwidth occupancy, so as to improve the enterprise data processing efficiency.

2. The management system based on enterprise data processing according to claim 1, wherein The calculation formula of the delay sensitivity identification value is as follows: ; is the delay sensitivity identification value, indicating the delay sensitivity level of the service to which the enterprise data stream belongs; is the current time point; 、 and correspond to the weight coefficients of the basic service type value, jitter rate, and path dependence factor respectively; is the basic service type value, which is the fixed sensitivity level mapped according to the service type to which the enterprise data stream belongs; represents the standard deviation of the packet arrival interval time of the enterprise data stream, which is used to quantify the jitter rate; represents the maximum allowable jitter threshold, which is used to normalize the jitter rate; represents the path dependence factor, indicating whether the enterprise data stream is located on the critical service call chain.

3. The management system based on enterprise data processing according to claim 1, wherein, The calculation formula of the computing intensity identification value is as follows: , where represents the computing intensity identification value, indicating the consumption intensity of computing resources by the business to which the enterprise data stream belongs; is the basic value of the business type to which the enterprise data stream belongs, which is a fixed intensity score mapped according to the operation instruction type; is the historical resource consumption pattern vector, indicating the average resource consumption of the business type to which the enterprise data stream belongs in historical data; is the current task resource demand vector, indicating the real-time resource demand of the business type to which the enterprise data stream belongs; is the load balancing coefficient; is the historical average CPU utilization rate of the business type to which the enterprise data stream belongs on the node; is the current CPU utilization rate of the business type to which the enterprise data stream belongs on the node.

4. The management system based on enterprise data processing according to claim 1, characterized in that, The dynamic resource portrait module (200) includes a delay policy switching unit (201), and the delay policy switching unit (201) is used for switching the load evaluation policy, specifically including: According to the value of the delay sensitivity identification value determine the delay sensitivity level of the target node required by the business to which the enterprise data stream belongs, where the delay sensitivity level includes low-sensitivity services, medium-sensitivity services, and high-sensitivity services; According to the delay sensitivity level, trigger the corresponding load evaluation policy and generate a resource capacity vector of the target node, where the load evaluation policy is as follows: When the delay sensitivity level is a high-sensitivity service, trigger millisecond-level real-time monitoring and adopt a hard real-time algorithm to ensure extremely low latency; When the delay sensitivity level is a medium-sensitivity service, trigger second-level periodic sampling and allocate resources based on a weighted load balancing algorithm; When the delay sensitivity level is a low-sensitivity service, trigger minute-level prediction monitoring and use a greedy algorithm to maximize resource utilization.

5. The management system based on enterprise data processing according to claim 1, characterized in that, The resource capacity vector includes a computing unit availability rate threshold and a storage bandwidth occupancy rate threshold.

6. The management system based on enterprise data processing according to claim 1, wherein The dynamic resource portrait module (200) includes a threshold determination unit (202), and the threshold determination unit (202) is used for generating a computing unit availability rate threshold and a storage bandwidth occupancy rate threshold, specifically including: According to the delay sensitivity level, limit the computing unit availability rate and the storage bandwidth occupancy rate of the target node to obtain the computing unit availability rate threshold and the storage bandwidth occupancy rate threshold of the target node. The specific process is as follows: When the latency sensitivity level is a highly sensitive service, and ; When the latency sensitivity level is medium-sensitive services, and ; When the latency sensitivity level is for low-sensitivity services, and ; where represents the available rate threshold of the computing unit, represents the occupancy rate threshold of the storage bandwidth, 、 and represent the historical available rate threshold of the computing unit, and ; 、 and represent the historical occupancy rate threshold of the storage bandwidth, and ; 、 and represent high-sensitivity service, medium-sensitivity service, and low-sensitivity service identifiers respectively.

7. The management system based on enterprise data processing according to claim 1, characterized in that, The adaptive routing decision module (300) includes a decision tree unit (301), and the decision tree unit (301) is used for adjusting the root node in the optimized node decision tree, specifically including: The optimized node decision tree includes all data transmission paths and nodes, where each node represents a computing node and the edges represent the connections between nodes; among them, the root node branches of the optimized node decision tree are divided according to the delay sensitivity identification value The determined delay sensitivity levels are divided into low-sensitivity services, medium-sensitivity services, and high-sensitivity services according to the delay sensitivity level, and the root node branches are divided into high-sensitivity branches, medium-sensitivity branches, and low-sensitivity branches.

8. The management system based on enterprise data processing according to claim 6, characterized in that The adaptive routing decision module (300) includes a path determination unit (302), and the path determination unit (302) is used for generating an optimal transmission path, specifically including: Identify according to the delay sensitivity of the enterprise data stream , select the corresponding root node branch, and obtain the available rate threshold of the computing unit in the limited resource capacity vector in the threshold determination unit (202) and the storage bandwidth occupancy rate threshold , select the nodes in the optimized node decision tree path that meet the available rate threshold of the computing unit and the storage bandwidth occupancy rate threshold , evaluate the weights of all path sets formed by the nodes, and select the path with the smallest weight as the optimal transmission path.

9. The management system based on enterprise data processing according to claim 4, wherein The delay sensitivity level is determined by hierarchical clustering to determine low-sensitivity services, medium-sensitivity services, and high-sensitivity services.

10. The management system based on enterprise data processing according to claim 1, wherein The enterprise data streams are obtained by streaming acquisition. By real-time accessing multi-source heterogeneous data, batch processing of the original stream is performed by using dynamic window segmentation, and data cleaning, feature extraction, real-time calculation of the delay sensitivity identification value and the computing intensity identification value are executed to generate multi-dimensional feature data; Achieve millisecond-level response in combination with a lightweight framework. At the same time, through traffic classification and elastic resource allocation, ensure that high-sensitivity tasks are processed first.

Images