Resource allocation method and system based on data elements

Abstract

The invention relates to the technical field of resource allocation analysis, in particular to a resource allocation method and system based on data elements, and the method comprises the steps: monitoring resource access, generating a space-time resource pulse sequence, and providing a data basis for subsequent analysis; the influence among data elements is quantified by constructing a space-time pulse response field, and the dynamic interaction relation in the system is visually displayed; a risk propagation path can be quickly positioned by identifying a strong interference data element pair and analyzing a causal link; a quantitative basis is provided for timely intervening excessive concentration of resources by calculating a resource polarization effect caused by pulse resonance; by tracking a polarization propagation trajectory and positioning key nodes, a resource bottleneck point which is most required to be intervened can be identified; pulse attenuation is started for key nodes exceeding a tolerance value, so that the risks of out-of-control resource competition and system collapse are effectively prevented; progressive resource reallocation is executed according to the instantaneous attenuation rate, and efficient system resource utilization and rapid balance recovery are ensured.

Description

Technical Field

[0001] This invention relates to the field of resource allocation analysis technology, and in particular to a resource allocation method and system based on data elements. Background Technology

[0002] In the field of modern data element resource allocation and management, with the widespread application of big data, cloud computing, and microservice architecture, the types and quantities of data elements within a system have increased dramatically. Different data service modules (such as data processing, data analysis, and data storage) need to share limited computing resources during execution. Currently, most resource allocation methods adopt static or semi-static strategies, allocating resources based on preset rules or historical average demand, which can maintain the basic operation of the system in an environment where resource demand is relatively stable.

[0003] However, in dynamic and ever-changing workload scenarios, especially when certain data service modules encounter sudden traffic surges or computationally intensive tasks, their resource demands can rise sharply. Existing resource allocation methods often lead to uneven resource allocation due to a lack of real-time assessment and dynamic adjustment capabilities for global resource contention. Excessive local resource concentration can quickly trigger resource contention across data service modules, which can easily amplify at each level in a highly coupled data processing environment, creating a Matthew effect in resource usage. Furthermore, intensified resource contention can cause the system to gradually evolve from local performance bottlenecks to global service delays or even system crashes. At the same time, traditional monitoring and scheduling methods are difficult to effectively predict and intervene in a timely manner, thus limiting the reliability and processing efficiency of data processing systems under high-concurrency and high-elasticity demand scenarios.

[0004] Therefore, this invention proposes a resource allocation method and system based on data elements. Summary of the Invention

[0005] The purpose of this invention is to solve the problems in the background art by proposing a resource allocation method and system based on data elements.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A resource allocation method based on data elements, comprising: S1. Monitor the resource access behavior of each data element and generate a resource pulse sequence; S2. Based on the resource pulse sequence, construct the spatiotemporal pulse response field between data elements, where the spatiotemporal pulse response field is used to quantify the spatiotemporal impact of the resource pulse of one data element on another data element. S3. Identify pairs of strongly interfering data elements in the spatiotemporal impulse response field and analyze the causal influence links between pairs of strongly interfering data elements; S4. When a pulse resonance event is detected on the causal link, the resource polarization effect caused by the pulse resonance event is calculated. S5. Based on the system resource topology, track the polarization propagation trajectory of the resource polarization effect and locate the key polarization convergence node in the polarization propagation trajectory where the intensity of the resource polarization effect is in a focused enhancement state. S6. If the intensity of the resource polarization effect of the key polarization convergence node exceeds the preset dynamic tolerance value, then the pulse attenuation strategy will be activated for the strong interference data elements that cause pulse resonance. S7. Based on the pulse decay strategy, obtain the instantaneous decay rate of the resource polarization effect, and perform gradual resource reallocation based on the instantaneous decay rate of the resource polarization effect.

[0007] Furthermore, the process of monitoring the resource access behavior of each data element and generating a resource pulse sequence includes: Record the timestamp and resource size of each data element that initiates a resource request within a unit of time; Each resource request is abstracted as a pulse event, which includes three dimensions: time, space, and energy. The time dimension is the timestamp of the pulse occurrence; the space dimension is the logical location coordinates defined by a preset logical topology mapping table; and the energy dimension is the pulse energy value calculated by the resource equivalent converter. All pulse events of each data element are combined in chronological order to form the resource pulse sequence of that data element.

[0008] Furthermore, the process of constructing the spatiotemporal impulse response field among data elements based on resource impulse sequences includes: Treating data elements as nodes, calculate the time-delay cross-correlation function of the resource pulse sequences of any two data element nodes; Extract the maximum peak value of the time-delay cross-correlation function and its corresponding time delay value, and take the peak value as the response intensity and the time delay as the propagation delay of the influence. Using response intensity and impact propagation delay as fields, a matrix describing the mutual influence relationships among all data elements is constructed, which is the spatiotemporal impulse response field.

[0009] Furthermore, the process of identifying pairs of strongly interfering data elements in the spatiotemporal impulse response field and analyzing the causal influence links between these pairs includes: In the spatiotemporal impulse response field, data element node pairs with response intensities exceeding a preset interference threshold are selected and marked as strong interference data element pairs; For each pair of strongly interfering data elements, the causal direction of the resource impact is determined based on the sign of the impact propagation delay. Connect all strongly interfering data elements that share the same causal direction and have direct or indirect interference relationships to form a causal influence chain.

[0010] Furthermore, when a pulse resonance event is detected on the causal link, the process of calculating the resource polarization effect caused by the pulse resonance event includes: Real-time monitoring of resource pulse sequences of each data element node in the causal impact chain; When multiple data element nodes on the causal link are detected to exhibit coordinated fluctuations that meet the preset resonance conditions, a pulse resonance event is determined to have occurred. Calculate the relative change in the statistical standard deviation of the resource occupancy rate distribution of each data element node in the causal impact chain caused by the pulse resonance event, and quantify the relative change as the intensity of the resource polarization effect.

[0011] Furthermore, based on the system resource topology, the process of tracing the polarization propagation trajectory of resource polarization effects and locating key polarization convergence nodes in the polarization propagation trajectory where the intensity of resource polarization effects is in a focused and enhanced state includes: The resource polarization effect is viewed as a fluctuation propagating along the system's resource topology network; Based on the propagation delay of the effect in the spatiotemporal impulse response field, the propagation path of the wave in the system resource topology network is simulated to form a polarized propagation trajectory. Identify data element nodes in the propagation trajectory that show a focused enhancement of resource polarization effect intensity and locate them as key polarization convergence nodes.

[0012] Furthermore, the process of identifying data element nodes in the propagation trajectory that exhibit a focused enhancement of resource polarization effect intensity includes: Calculate the second derivative of the resource polarization intensity of data element nodes within a continuous monitoring period; When the second derivative is greater than 0, the resource polarization intensity of the data element node is determined to be in a state of enhanced focus.

[0013] Furthermore, the pulse attenuation strategy includes: imposing resource request frequency limits and parallel processing quota constraints on the strongly interfering data elements that trigger pulse resonance; Among them, the frequency limit of resource requests is characterized by the frequency limit coefficient, and the parallel processing quota constraint is characterized by the quota constraint coefficient; the values ​​of the frequency limit coefficient and the quota constraint coefficient are dynamically adjusted according to the extent to which the intensity of resource polarization effect exceeds the preset dynamic tolerance value.

[0014] Furthermore, based on the pulse decay strategy, the instantaneous decay rate of the resource polarization effect is obtained, and the process of performing gradual resource reallocation based on the instantaneous decay rate of the resource polarization effect includes: Monitor the intensity of resource polarization effect at key polarization convergence nodes and calculate its instantaneous decay rate over time; When the instantaneous decay rate exceeds the preset decay rate trigger threshold, the polarization effect of the critical polarization convergence node is determined to enter the mitigation channel, and the resource reallocation operation is initiated. Identify low-polarity regions; establish a positive feedback control model for instantaneous decay rate and resource allocation speed; Continue to perform incremental reallocation until the intensity of resource polarization effect at critical polarization convergence nodes completely dissipates and the system returns to stability.

[0015] A resource allocation system based on data elements, comprising: The resource pulse sensing module is used to monitor the resource access behavior of each data element and generate a resource pulse sequence. The spatiotemporal impulse response field construction module is used to construct the spatiotemporal impulse response field between data elements based on resource impulse sequences; The causal link analysis module is used to identify pairs of strongly interfering data elements in the spatiotemporal impulse response field and to analyze the causal influence links between these pairs. The resource polarization effect calculation module is used to calculate the resource polarization effect caused by the pulse resonance event when a pulse resonance event is detected on the causal influence link. The critical polarization convergence node localization module is used to track the polarization propagation trajectory of resource polarization effect according to the system resource topology, and locate the critical polarization convergence node in the polarization propagation trajectory where the intensity of resource polarization effect is in a state of focused enhancement. The pulse attenuation control module is used to initiate a pulse attenuation strategy for strong interference data elements that cause pulse resonance when the intensity of the resource polarization effect at a key polarization convergence node exceeds a preset dynamic tolerance value. The dynamic resource allocation module is used to obtain the instantaneous decay rate of the resource polarization effect based on the pulse decay strategy, and to perform gradual resource reallocation based on the instantaneous decay rate of the resource polarization effect.

[0016] Compared with existing technologies, the present invention provides a resource allocation method and system based on data elements, the advantages of which are: 1) By monitoring the resource access behavior of each data element and generating resource pulse sequences with spatiotemporal attributes, we can improve the ability to finely perceive the resource usage of data elements and provide accurate basic data for subsequent resource allocation optimization. 2) By constructing a spatiotemporal impulse response field between data elements, the spatiotemporal impact of the resource impulse of one data element on another data element is quantified, thereby enhancing the system's ability to understand and analyze the interaction relationships between data elements; 3) By identifying pairs of strongly interfering data elements in the spatiotemporal impulse response field and analyzing their causal influence links, the system can accurately capture the key influence paths between data elements, providing a basis for resource conflict early warning and intervention; 4) When a pulse resonance event is detected on the causal link, the system can calculate the resource polarization effect caused by the event, thereby quantifying the negative effects of resource competition and providing quantitative support for timely resource allocation measures. 5) By tracking the polarization propagation trajectory of resource polarization effects and locating key polarization convergence nodes in the polarization propagation trajectory where the intensity of resource polarization effects is in a state of focused enhancement, the system can accurately identify resource bottlenecks and risk points, providing clear targets for implementing targeted intervention measures; 6) When the intensity of the resource polarization effect at the critical polarization convergence node exceeds the preset dynamic tolerance value, the system can activate the pulse attenuation strategy for the strong interference data elements that cause pulse resonance, effectively suppressing the spread and aggravation of the resource polarization effect and ensuring the stable operation of the system. 7) By obtaining the instantaneous decay rate of resource polarization effect based on the pulse decay strategy and performing gradual resource reallocation, the system can timely and reasonably reallocate resources after the resource polarization effect is alleviated, thereby improving the overall resource utilization efficiency and balance of the system. Attached Figure Description

[0017] Figure 1 This is a flowchart of a resource allocation method based on data elements proposed in this invention.

[0018] Figure 2 This is a block diagram of a resource allocation system based on data elements proposed in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 The present invention provides a resource allocation method based on data elements, comprising: S1. Monitor the resource access behavior of each data element and generate a resource pulse sequence with spatiotemporal attributes; S2. Based on the resource pulse sequence, construct the spatiotemporal pulse response field between data elements, where the spatiotemporal pulse response field is used to quantify the spatiotemporal impact of the resource pulse of one data element on another data element. S3. Identify pairs of strongly interfering data elements in the spatiotemporal impulse response field and analyze the causal influence links between pairs of strongly interfering data elements; S4. When a pulse resonance event is detected on the causal link, the resource polarization effect caused by the pulse resonance event is calculated. S5. Based on the system resource topology, track the polarization propagation trajectory of resource polarization effect and locate the key polarization convergence node in the polarization propagation trajectory where the intensity of resource polarization effect is in a focused enhancement state; where, the system resource topology is based on the internal logical topology of the system, focuses on the relationship under the specific perspective of resource competition and influence transmission, and describes the potential influence path between data elements due to resource interaction. S6. If the intensity of the resource polarization effect of the critical polarization convergence node exceeds the preset dynamic tolerance value, a pulse attenuation strategy will be initiated for the strong interference data elements that cause pulse resonance. The dynamic tolerance value is used to determine whether the intensity of the resource polarization effect of the critical polarization convergence node exceeds the threshold of the acceptable range. The dynamic tolerance value is set when tracking the resource polarization effect and locating the critical polarization convergence node, and serves as the trigger condition for initiating the pulse attenuation strategy. The specific value is dynamically adjusted according to the system stability and resource allocation requirements. S7. Based on the pulse decay strategy, obtain the instantaneous decay rate of the resource polarization effect, and perform gradual resource reallocation based on the instantaneous decay rate of the resource polarization effect.

[0021] It should be further explained that, in the specific implementation process, the process of monitoring the resource access behavior of each data element and generating a resource pulse sequence includes: Record the timestamp and resource size of each data element that initiates a resource request within a unit of time; Each resource request is abstracted as a pulse event, which includes three dimensions: time, space, and energy. The time dimension is the timestamp of the pulse occurrence; the space dimension is the logical location coordinates defined by a preset logical topology mapping table; and the energy dimension is the pulse energy value calculated by the resource equivalent converter. All pulse events of each data element are combined in chronological order to form the resource pulse sequence of that data element; Specifically, monitoring resource access behavior includes: deploying lightweight monitoring probes at each resource access interface of the data elements, where resource access interfaces include, but are not limited to, application programming interface gateways, database connection pools, and computing task submission entry points; the monitoring probes record the instantaneous timestamp of each resource request, and also record the logical location identifier from which the resource request originates, including the business module number to which it belongs, the virtualization cluster identifier, and the name of the microservice it depends on; the monitoring probes record in detail the resource scale consumed by this resource request, where the resource scale is a multi-dimensional metric, including key indicators such as the number of CPU cores per second, the number of megabytes of memory used, the megabits per second of network bandwidth, and the number of I / O operations per second of storage; and generating an original record for each resource request containing a timestamp, logical location identifier, and other multi-dimensional resource scale metrics. The abstraction and normalization of pulse events includes: introducing a resource equivalent converter to unify resource requests from different sources and of different types into a pulse concept that can be compared and analyzed. The resource equivalent converter aggregates multi-dimensional resource vectors into a single scalar value based on preset resource weight coefficients for various resource types; this scalar value is called the pulse energy value. The pulse energy value is a normalized quantized representation of the resource scale. For example, the conversion rules are set as follows: a CPU core-second is set as the base unit of 1, 1 megabyte of memory usage is converted to 0.1 units, and 1 megabit per second of network bandwidth is converted to 0.05 units, etc. Through weighted accumulation, the resource scale of a single resource request can be converted into the corresponding pulse energy value. Each original record is abstracted into a standard pulse event, and the data structure of this event... The pulse consists of: {timestamp of pulse occurrence, logical location coordinates, and pulse energy value}. The logical location coordinates are defined using a pre-defined logical topology mapping table. This table is constructed based on pre-defined logical relationships between data elements (such as data flow dependencies, service call chains, or predefined business groups), thus forming an internal logical topology describing the connections between data elements. A hash value generated based on the internal resource topology is used for rapid logical location within the system's communication network or logical addressing system. In essence, the internal logical topology defines the inherent, logical connections and dependencies between data elements, predetermined by system design or business logic, for example, through data flow dependency graphs or microservice call chains. Its function is to provide logical location coordinates for each data element. The construction of resource pulse sequences and feature extraction includes: maintaining a time-series-arranged event pool for each uniquely identified data element; storing all pulse events belonging to that data element in the event pool according to their timestamps, thus forming a unique resource pulse sequence for that data element; performing real-time feature extraction on the resource pulse sequence of each data element within a sliding feature extraction time window (e.g., every 60 seconds); calculating features including the average pulse energy within the feature extraction time window, the number of pulse events occurring per unit time (i.e., pulse frequency), and the variance of pulse energy; using the real-time statistical feature values ​​as the basic input data for subsequent detection of abnormal pulse activity; the entire process transforms multidimensional heterogeneous resource access behavior into discrete event sequences carrying time, space, and energy labels, laying the data foundation for subsequent construction of higher-level spatiotemporal pulse response fields and effect analysis.

[0022] It should be further explained that, in the specific implementation process, the process of constructing the spatiotemporal impulse response field among data elements based on the resource impulse sequence includes: Treating data elements as nodes, calculate the time-delay cross-correlation function of the resource pulse sequences of any two data element nodes; Extract the maximum peak value of the time-delay cross-correlation function and its corresponding time delay value, and take the peak value as the response intensity and the time delay as the propagation delay of the influence. Using response intensity and impact propagation delay as fields, a matrix describing the mutual influence relationships among all data elements is constructed, which is the spatiotemporal impulse response field; Specifically, the preprocessing and time alignment of the resource pulse sequences of data elements includes: selecting a sufficiently long and common analysis time window from the resource pulse sequences of all data elements, such as data within the most recent 30 minutes; performing time alignment on all selected resource pulse sequences: selecting a fixed time resolution (e.g., 100 milliseconds) as a time slot, and dividing the entire analysis time window into consecutive equal-length time slots; for each data element's resource pulse sequence, if no pulse event occurs within each time slot, its normalized pulse energy value in that time slot is recorded as 0; if multiple pulse events occur within a time slot, the pulse energy values ​​of the pulse events are accumulated to obtain the total pulse energy value for that time slot; finally, the resource pulse sequence of each data element is transformed into an energy time sequence with equal time intervals. The calculation of the time-delay cross-correlation function includes: for any two different data element nodes, calculating the standardized time-delay cross-correlation function between their respective energy time series; the calculation method of the time-delay cross-correlation function is as follows: for a series of different time delay values, calculate the covariance of one energy time series and another energy time series at the corresponding time delay, divide it by the product of the standard deviations of the two energy time series and the square root of the number of effective data points, and thus obtain the standardized correlation coefficient; the range of time delay values ​​is set between the maximum positive and negative time delays, for example, the number of time slots corresponding to ±5 seconds; traverse all time delay values ​​within the range to obtain the complete functional relationship curve between the energy time series of the two data element nodes with respect to time delay; Extracting key parameters from the function curve includes: scanning the standardized time-delay cross-correlation function curve to find its global maximum value, i.e., the global highest peak value on the curve; the magnitude of the global highest peak value is defined as the response intensity from the previous data element node to the next data element node, used to quantify the extent to which the impulse activity of the previous data element node can predict or correlate the impulse activity of the next data element node; simultaneously, the time delay value corresponding to the global highest peak value is recorded, which is defined as the propagation delay of the influence between the two data element nodes; if the time delay value is positive, it indicates that the activity of the previous data element node leads the activity of the next data element node, suggesting that there may be a causal influence direction from the previous data element node to the next data element node; if the time delay value is negative, it indicates that the influence direction may be opposite; Constructing the spatiotemporal impulse response field matrix involves: maintaining a matrix of size equal to the total number of data element nodes multiplied by the total number of data element nodes; this matrix is ​​the mathematical representation of the spatiotemporal impulse response field; each element in the matrix corresponds to a pair of data element nodes, and the content stored in this element is a tuple containing two parameters: the calculated response intensity and the influence propagation delay; the spatiotemporal impulse response field matrix comprehensively characterizes the response intensity and temporal relationship of the mutual influence between all data element nodes within the system; the spatiotemporal impulse response field matrix is ​​updated periodically, for example, every ten minutes, to capture dynamic changes in the system; to filter out noise and random correlations, a significance threshold is set (the significance threshold is set when constructing the spatiotemporal impulse response field, for example, set to 0.3, to ensure that the influence relationship between the analyzed data elements is statistically significant). Only when the response intensity of a pair of data element nodes exceeds the significance threshold is a significant influence relationship considered to exist between them, and this relationship is included in the subsequent analysis process.

[0023] It should be further explained that, in the specific implementation process, the process of identifying pairs of strongly interfering data elements in the spatiotemporal impulse response field and analyzing the causal influence links between these pairs of strongly interfering data elements includes: In the spatiotemporal impulse response field, data element node pairs with response intensities exceeding a preset interference threshold are selected and marked as strong interference data element pairs. The interference threshold is set when identifying strong interference data element pairs and is usually higher than the significance threshold, for example, set to 0.6, to ensure that the identified strong interference relationships have high intensity. For each pair of strongly interfering data elements, the causal direction of the resource impact is determined based on the sign of the impact propagation delay. Connect all strongly interfering data elements that have the same causal direction and a direct or indirect interference relationship to form a causal influence chain; Specifically, identifying strong interfering data element pairs includes: traversing each off-diagonal element in the spatiotemporal impulse response field matrix, i.e., each pair of distinct data element nodes; setting an interference threshold, which is higher than the significance threshold used for screening, for example, 0.6; for any pair of data element nodes, if the response intensity between them is greater than the set interference threshold, and the absolute value of their influence propagation delay is greater than the minimum delay threshold (the minimum delay threshold is set together with the interference threshold when identifying strong interfering data element pairs, for example, set to 50 milliseconds to exclude influences that may be caused by random factors), then the pair of data element nodes is marked as a candidate strong interfering data element pair; by checking the sign of their influence propagation delay, the causal direction of the influence is determined; if the delay value is positive, the causal direction is defined as from the first data element node to the second data element node; if the delay value is negative, the causal direction is defined as from the second data element node to the first data element node; simultaneously, a detailed data structure is created for each strong interfering data element pair with confirmed direction, recording the source data element node, target data element node, response intensity, and propagation delay information; Reconstructing causal influence links involves: treating all identified, strongly interfering data element pairs with clear directions as directed edges and constructing a directed interferometric graph of the entire system; executing a causal link discovery algorithm on the directed interferometric graph: for existing direct connections in the directed interferometric graph, i.e., direct directed edges, such as a direct path from data element node A to data element node B, a basic direct causal influence link is formed; for indirect causal influence links, directed paths of length two or more are sought, such as paths from data element node A to data element node B, and from data element node B to data element node C; verifying the causal effectiveness of indirect paths by calculating the partial correlation coefficient between data element node A and data element node C, i.e., after controlling for the influence of intermediate data element node B, the causal effectiveness of the indirect paths is verified. The remaining correlation between node A and data element node C is analyzed. If, after controlling for data element node B, the partial correlation coefficient between data element node A and data element node C becomes insignificant or decreases significantly, then the influence of data element node A on data element node C is considered to be mainly transmitted through data element node B, thus confirming that the path A to B to C is an effective indirect causal link. Simultaneously, the presence of directed loops in the directed interferogram is detected, i.e., loops where the path eventually returns to the starting point. If such loops are detected, they are marked as impulse resonance loops. Impulse resonance loops are unstable causal link structures that require close monitoring because they are highly prone to triggering subsequent impulse resonance events. Finally, the output is a set of causal links containing all direct links, verified indirect links, and impulse resonance loops.

[0024] It should be further explained that, in the specific implementation process, when a pulse resonance event is detected on the causal link, the process of calculating the resource polarization effect caused by the pulse resonance event includes: Real-time monitoring of resource pulse sequences of each data element node in the causal impact chain; When multiple data element nodes on a causal link are detected to exhibit coordinated fluctuations that meet preset resonance conditions, a pulse resonance event is determined to have occurred. The coordinated fluctuations that meet the preset resonance conditions are abnormal link aggregation pulse frequency and high pulse synchronization. Calculate the relative change in the statistical standard deviation of the resource occupancy rate distribution of each data element node in the causal impact link caused by the pulse resonance event, and quantify the relative change as the intensity of the resource polarization effect; Specifically, monitoring resource pulse sequences includes: continuously listening to the real-time resource pulse sequences of each data element node on all identified causal links; maintaining a short-term monitoring window (e.g., a window length of ten seconds); within the short-term monitoring window, calculating two resonance indicators in parallel to determine whether pulse resonance has occurred; the first indicator is the link aggregate pulse frequency, calculating the average pulse frequency of all data element nodes on the entire causal link within the current short-term monitoring window to obtain the link average pulse frequency value; and comparing the calculated link average pulse frequency value with the historical average frequency baseline value calculated by each data element node on the link within a historical time period (e.g., the past hour); when the link average pulse frequency value of the entire causal link exceeds the statistical mean of the historical average frequency baseline values ​​of each data element node and reaches a preset frequency mutation threshold (the frequency mutation threshold is set when monitoring resource pulse sequences to determine pulse resonance events, for example, set to 150%), then... If significant frequency changes are detected, a frequency anomaly flag is triggered. The second indicator is the pulse synchronization between data element nodes. The root mean square error (RMSE) of the pulse events occurring at any two data element nodes on the causal link within the current short-term monitoring window is calculated. If the calculated RMS error is less than a preset synchronization threshold (the synchronization threshold is set when monitoring resource pulse sequences to determine pulse resonance events, for example, set to 200 milliseconds, to quantify the degree of synchronization of pulse activities between nodes), then the pulse activities between data element nodes are considered to be highly synchronized. When both the link aggregation pulse frequency anomaly and the pulse synchronization between data element nodes are satisfied, a pulse resonance event is determined to have occurred on the causal link. The start time of the pulse resonance, the specific causal link involved, and the initial intensity of the pulse resonance are recorded. The initial intensity of the pulse resonance is expressed as the percentage by which the link average pulse frequency value exceeds the historical average frequency baseline value. The quantification process of the resource polarization effect is as follows: Upon confirmation of the impulse resonance event, the calculation of the resource polarization effect is immediately initiated; a time period before the start of the impulse resonance event is selected as the baseline time window (e.g., 30 seconds before the impulse resonance occurs), and a time period of the same length after the start of the impulse resonance is selected as the observation time window; resource occupancy rate data for each data element node on the causal link are collected within the baseline and observation time windows, respectively. Resource occupancy rate typically refers to CPU utilization or memory utilization; the statistical standard deviation of the resource occupancy rate of all data element nodes on the link is calculated within the baseline and observation time windows, respectively; the intensity of the resource polarization effect is quantified as the observation window value. The relative increment of the statistical standard deviation relative to the baseline time window is calculated by subtracting the statistical standard deviation of the baseline time window from the statistical standard deviation of the observation time window, and then dividing by the statistical standard deviation of the baseline time window. When the calculated intensity of the resource polarization effect is positive, it indicates that the resource occupancy among the data element nodes on the link has become more uneven due to the impulse resonance event, resulting in a wealth disparity where resources are concentrated in a few data element nodes, i.e., polarization has occurred. The greater the intensity, the more severe the polarization effect, and the greater the damage to the balance of the entire system. By linking the impulse resonance phenomenon with precisely measurable changes in statistical distribution, the negative effects of resource competition can be objectively and accurately quantified.

[0025] It should be further explained that, in the specific implementation process, the process of tracing the polarization propagation trajectory of resource polarization effect according to the system resource topology, and locating the key polarization convergence node in the polarization propagation trajectory where the intensity of resource polarization effect is in a focused enhancement state, includes: The resource polarization effect is viewed as a fluctuation propagating along the system resource topology network. The system resource topology network is a dynamic instantiation and quantitative representation of the system resource topology at a specific moment. The system resource topology network is a computable network graph, where nodes are data elements, edges represent connections, and are accompanied by quantitative parameters (i.e., response intensity and impact propagation delay) calculated from the spatiotemporal impulse response field. The system resource topology network is the specific computational carrier and dynamic domain for simulating the propagation of the resource polarization effect and locating key polarization convergence nodes. Based on the propagation delay of the effect in the spatiotemporal impulse response field, the propagation path of the wave in the system resource topology network is simulated to form a polarized propagation trajectory. Identify data element nodes in the propagation trajectory that exhibit a focused enhancement of resource polarization effect intensity and locate them as key polarization convergence nodes: calculate the second derivative of the resource polarization intensity of the data element node within a continuous monitoring period; when the second derivative is greater than 0, it is determined that the resource polarization intensity of the data element node is in a focused enhancement state; Specifically, the simulation process of polarization propagation trajectory is as follows: The identified and quantified resource polarization effect is regarded as an excitation signal propagating in the system resource topology network, where the initial intensity of the excitation signal is the calculated intensity of the resource polarization effect. The excitation signal is applied to the starting data element node of the causal influence link where impulse resonance occurs. The spatiotemporal impulse response field matrix is ​​used as a model to describe the propagation of the excitation signal in the system resource topology network, simulating the dynamic propagation process of the excitation signal in the entire system resource topology network. The simulation process adopts a discrete-time stepping method: at the initial moment of the simulation, i.e., the zeroth time step, the intensity of the resource polarization effect is assigned to the starting data element node of the impulse resonance link. At each simulation time step, for each data element node in the system resource topology network, the intensity of the resource polarization effect it receives from all upstream data element nodes is equal to the intensity of the upstream data element node. The resource polarization effect intensity of a point in the previous time step is multiplied by the response intensity of that upstream data element node to the current data element node recorded in the spatiotemporal impulse response field matrix. The node aggregate polarization intensity of the current data element node in this time step is the sum of the resource polarization effect intensities received from all upstream data element nodes. To more realistically reflect the delay characteristics of influence propagation, the simulation process takes into account the influence propagation delay recorded in the spatiotemporal impulse response field matrix. The excitation signal needs to pass through a corresponding delay time step to propagate from one data element node to another. The arrival time of the excitation signal on different links is managed by maintaining a delay queue. The resource polarization effect intensity of each data element node in the system resource topology network is recorded in each simulation time step. The spatial distribution of the resource polarization effect intensity over time in the entire system resource topology network constitutes the polarization propagation trajectory of the resource polarization effect. Locating key polarization convergence nodes involves: during simulated propagation, analyzing the temporal variation trend of the resource polarization effect intensity of each data element node to locate nodes with abnormally concentrated polarization effects that may become system bottlenecks, i.e., key polarization convergence nodes, from the system resource topology network; real-time analysis of the sequence of resource polarization effect intensity changes of each data element node over time steps in the propagation trajectory; calculating the second-order central difference of this sequence in the most recent consecutive time steps as a discrete approximation of the acceleration of the resource polarization effect intensity change of that data element node; and when any data element node is detected to have its resource polarization effect intensity changes over multiple consecutive time steps (e.g., 3 time steps), its resources... When the second-order central differences of the source polarization effect intensity are all greater than 0, the data element node is determined to be in a state of enhanced focus. This indicates that the resource polarization effect intensity of the data element node is at a high level, and its growth rate is still accelerating. This suggests that the data element node is becoming a core for attracting and accumulating polarization effects, and the risk is accumulating rapidly. All data element nodes in the state of enhanced focus are marked as candidate convergence nodes. From the list of candidate convergence nodes, the top K data element nodes with the highest absolute value of the current resource polarization effect intensity (e.g., K=3) are selected and finally confirmed as key polarization convergence nodes that require immediate intervention.

[0026] It should be further explained that, in the specific implementation process, the pulse attenuation strategy includes: applying resource request frequency limits and parallel processing quota constraints to the strong interference data elements that cause pulse resonance; Among them, the frequency limit of resource requests is characterized by the frequency limit coefficient, and the parallel processing quota constraint is characterized by the quota constraint coefficient; the values ​​of the frequency limit coefficient and the quota constraint coefficient are dynamically adjusted according to the extent to which the intensity of resource polarization effect exceeds the preset dynamic tolerance value. Specifically, during the pulse attenuation strategy triggering and initial attenuation parameter calculation phase, if any critical polarization convergence node is confirmed and its resource polarization effect intensity exceeds the preset dynamic tolerance value, a strategy execution command is immediately sent to the pair of strongly interfering data elements (i.e., the source of the pulse resonance) identified as causing this pulse resonance; the preset dynamic tolerance value is then used; based on the magnitude by which the resource polarization effect intensity of the critical polarization convergence node exceeds the dynamic tolerance value (i.e., the excess value), the initial attenuation parameters are calculated using a predefined attenuation coefficient mapping function. This attenuation coefficient mapping function is piecewise linear. For example, when the excess is between 0 and 0.1, the resource request frequency limit coefficient is set to 0.8, allowing the maximum request frequency of the data element to drop to 80% of the normal level, while the parallel processing quota constraint coefficient is set to 0.7; when the excess is between 0.1 and 0.3, the frequency limit coefficient is adjusted to 0.5, and the quota constraint coefficient is adjusted to 0.4; when the excess is greater than 0.3, stricter restrictions are applied, with a frequency limit coefficient of 0.3 and a quota constraint coefficient of 0.2. During the pulse decay strategy execution phase, the calculated frequency limit coefficient and quota constraint coefficient are distributed to the traffic controller and resource scheduler responsible for managing the resource access of the data element. The traffic controller typically uses an algorithm such as the token bucket to set the upper limit of the resource request frequency of the data element as the frequency limit coefficient multiplied by its historical average request frequency (i.e., the average request frequency of the data element in its historical normal state). The resource scheduler limits the maximum computing resource unit (e.g., the number of container instances) that the data element can occupy at the same time to the calculated quota constraint coefficient multiplied by its rated quota. During the dynamic feedback adjustment phase, after the pulse attenuation strategy is initiated, a closed-loop control loop is established; the rate of change of the resource polarization effect intensity of the key polarization convergence node is continuously monitored; a preset strategy control period (e.g., 5 seconds) is implemented; at the end of each strategy control period, the attenuation coefficient applied to the strongly interfering data element pair is dynamically adjusted based on the absolute value and rate of change of the current resource polarization effect intensity of the key polarization convergence node: if the resource polarization effect intensity is still rising or decreasing very slowly, the attenuation intensity is increased by a certain step size (e.g., 0.1), i.e., the frequency limit coefficient and quota constraint coefficient are further reduced; if the resource polarization effect intensity begins to decrease rapidly, the current attenuation coefficient is maintained to observe the effect; if the resource polarization effect intensity has decreased below the dynamic tolerance value and remains stable, the attenuation coefficient is gradually relaxed by a smaller step size (e.g., 0.05) until the system returns to normal or a new pulse resonance event is triggered; the mechanism of adjusting the pulse attenuation strategy intensity in real time according to the effect gives the pulse attenuation strategy the ability to adapt to the system state, making it an intelligent controller that can smoothly and effectively suppress pulse resonance.

[0027] It should be further explained that, in the specific implementation process, the process of obtaining the instantaneous decay rate of the resource polarization effect based on the pulse decay strategy, and then performing the gradual reallocation of resources based on the instantaneous decay rate of the resource polarization effect, includes: Monitor the intensity of resource polarization effect at key polarization convergence nodes and calculate its instantaneous decay rate over time; When the instantaneous decay rate exceeds the preset decay rate trigger threshold, the polarization effect of the critical polarization convergence node is determined to enter the mitigation channel, and the resource reallocation operation is initiated. The decay rate trigger threshold is set when monitoring the intensity of the resource polarization effect of the critical polarization convergence node to determine whether to initiate resource reallocation. The specific value is determined according to the system characteristics and requirements. Identify low-polarity regions; establish a positive feedback control model for instantaneous decay rate and resource allocation speed; wherein, the speed at which resources are allocated from the dynamic resource buffer pool to data elements in low-polarity regions is dynamically adjusted based on the instantaneous decay rate; Continue to perform gradual reallocation until the intensity of resource polarization effect at critical polarization convergence nodes completely dissipates and the system returns to stability; Specifically, the process of managing the dynamic resource buffer pool is as follows: During the period when the pulse decay policy is in effect, the computing resources reclaimed from the restricted data elements, such as the CPU cores and memory blocks that become idle as a result, are uniformly injected into a dedicated dynamic resource buffer pool; the capacity of the dynamic resource buffer pool changes dynamically with the intensity of the pulse decay policy, providing a source of resources for subsequent reallocation. The monitoring and allocation triggering conditions for determining the instantaneous decay rate include: continuously calculating the instantaneous decay rate of the resource polarization effect intensity of the critical polarization convergence node over time, i.e., the decrease in the resource polarization effect intensity per unit time; setting a decay rate trigger threshold; when the instantaneous decay rate of the critical polarization convergence node is not less than the decay rate trigger threshold, it indicates that the pulse decay strategy has taken effect and the resource polarization effect of the critical polarization convergence node is entering a rapid mitigation phase, at which point the resource reallocation operation is initiated; avoiding blindly allocating resources when the resource polarization effect is still high or mitigating slowly, which may lead to resource waste or even exacerbate system instability; The identification of low-polarity regions and the ranking of resource needs based on urgency include: real-time scanning of the current resource occupancy status of all data element nodes; a low-polarity region is defined as a set of data element nodes whose resource occupancy rate is lower than a certain percentage of the current overall average resource occupancy rate of the system (e.g., lower than 80% of the average). For data element nodes located in low-polarity regions, they are ranked according to their resource need urgency; the resource need urgency is determined by a combination of factors such as task queue backlog rate, service level agreement priority, and historical resource scarcity; the task queue backlog rate is the ratio of the number of pending tasks to its own processing capacity; the service level agreement priority is that data element nodes undertaking high-priority tasks will receive a higher urgency score; and the historical resource scarcity is the frequency with which resource requests have been rejected or delayed in the recent past. The execution process of the incremental reallocation strategy is as follows: Resource reallocation adopts an incremental strategy; a positive feedback model is established, making the resource allocation speed proportional to the instantaneous decay rate of the monitored key polarization convergence nodes—that is, the faster the decay, the faster the allocation speed; a resource allocation control period is set (e.g., two seconds); in each resource allocation control period, a certain amount of resources matching the current allocation speed is retrieved from the dynamic resource buffer pool; resources are allocated sequentially to low-polarity data element nodes at the top of the list, starting from the data element node with the highest urgency, according to the resource demand urgency ranking list; throughout the allocation process, the resource distribution balance of the entire system is continuously monitored (e.g., by calculating the Gini coefficient or standard deviation of the resource occupancy rate of all data element nodes). This resource distribution balance is negatively correlated with the intensity of the resource polarization effect, and the status of the key polarization convergence nodes is monitored simultaneously to ensure that this reallocation behavior is in the direction of reducing the overall imbalance of the system; once the resource polarization effect intensity of the key polarization convergence nodes completely disappears and the system state returns to stability, any remaining resources in the dynamic resource buffer pool are released back to the global resource pool, and the incremental reallocation process ends.

[0028] Please read Figure 2 This invention provides a resource allocation system based on data elements, comprising: The resource pulse sensing module is used to monitor the resource access behavior of each data element and generate a resource pulse sequence with spatiotemporal attributes. The spatiotemporal impulse response field construction module is used to construct the spatiotemporal impulse response field between data elements based on resource impulse sequences; The causal link analysis module is used to identify pairs of strongly interfering data elements in the spatiotemporal impulse response field and to analyze the causal influence links between these pairs. The resource polarization effect calculation module is used to calculate the resource polarization effect caused by the pulse resonance event when a pulse resonance event is detected on the causal influence link. The critical polarization convergence node localization module is used to track the polarization propagation trajectory of resource polarization effect according to the system resource topology, and locate the critical polarization convergence node in the polarization propagation trajectory where the intensity of resource polarization effect is in a state of focused enhancement. The pulse attenuation control module is used to initiate a pulse attenuation strategy for strong interference data elements that cause pulse resonance when the intensity of the resource polarization effect at a key polarization convergence node exceeds a preset dynamic tolerance value. The dynamic resource allocation module is used to obtain the instantaneous decay rate of the resource polarization effect based on the pulse decay strategy, and to perform gradual resource reallocation based on the instantaneous decay rate of the resource polarization effect.

[0029] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. The focus of each embodiment is on its differences from other embodiments. In particular, the apparatus embodiments are described simply because they are fundamentally based on the method embodiments; relevant details can be found in the descriptions of the method embodiments.

[0030] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0031] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can be implemented in one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs) containing computer-usable program code. The form of a computer program product implemented on ROM, optical memory, etc.

[0032] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0033] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0034] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0035] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A data element-based resource configuration method, characterized by: S1, monitoring the resource access behavior of each data element to generate a resource pulse sequence; S2, constructing a time-space pulse response field between data elements based on the resource pulse sequence, wherein the time-space pulse response field is used to quantify the time-space influence of the resource pulse of one data element on another data element; S3, identifying strong interference data element pairs in the time-space pulse response field and analyzing the causal influence links between the strong interference data element pairs; S4, when a pulse resonance event occurs on the causal influence link, calculating the resource polarization effect triggered by the pulse resonance event; S5, according to the system resource topology, tracking the polarization propagation trajectory of the resource polarization effect, and locating the key polarization convergence nodes in the polarization propagation trajectory where the resource polarization effect strength is in a focused enhanced state; S6, if the resource polarization effect strength of the key polarization convergence node exceeds the preset dynamic tolerance value, starting a pulse attenuation strategy for the strong interference data element pair that triggers the pulse resonance; S7, according to the pulse attenuation strategy, obtaining the instantaneous attenuation rate of the resource polarization effect, and performing gradual reallocation of resources according to the instantaneous attenuation rate of the resource polarization effect.

2. The method of claim 1, wherein, The process of monitoring the resource access behavior of each data element to generate a resource pulse sequence includes: Recording the timestamp and resource size of each data element initiating a resource request within a unit of time; Each resource request is abstracted as a pulse event, which contains three dimensions of time, space and energy; wherein the time dimension is the timestamp of the pulse occurrence; the space dimension is the logical position coordinates defined by the preset logical topology mapping table, and the energy dimension is the pulse energy value calculated by the resource equivalent converter; Combining all pulse events of each data element in chronological order to form the resource pulse sequence of the data element.

3. The method of claim 1, wherein, The process of constructing a time-space pulse response field between data elements based on the resource pulse sequence includes: Regarding data elements as nodes, calculating the time-lag cross-correlation function of the resource pulse sequence of any two data element nodes; Extracting the maximum peak value of the time-lag cross-correlation function and its corresponding time lag, taking the peak value as the response strength and the time lag as the influence propagation delay; Constructing a matrix describing the mutual influence relationship between all data elements with response strength and influence propagation delay as fields, which is the time-space pulse response field.

4. The method of claim 3, wherein, The process of identifying strong interference data element pairs in the time-space pulse response field and analyzing the causal influence links between the strong interference data element pairs includes: In the time-space pulse response field, filter out the data element node pairs with response strength exceeding the preset interference threshold, and mark them as strong interference data element pairs; For each strong interference data element pair, determine the causal direction of resource influence according to the positive and negative of the influence propagation delay; Connect all strong interference data element pairs with the same causal direction and direct or indirect interference relationship to form the causal influence link.

5. The method of claim 1, wherein, The process of calculating the resource polarization effect triggered by the pulse resonance event when a pulse resonance event occurs on the causal influence link includes: Real-time monitoring of the resource pulse sequence of each data element node on the causal influence link; When detecting that multiple data element nodes on the causal influence link appear to have coordinated fluctuations that meet the preset resonance condition, it is determined that a pulse resonance event has occurred; The statistical standard deviation of the resource occupancy rate distribution of each data element node on the causal influence link caused by the pulse resonance event is calculated, and the relative change is quantified as the resource polarization effect strength.

6. The method of claim 1, wherein, According to the system resource topology, the polarization propagation trajectory of the resource polarization effect is tracked, and the process of locating the key polarization convergence node in the polarization propagation trajectory where the resource polarization effect strength is in a focused enhancement state includes: Regarding the resource polarization effect as a fluctuation propagating along the system resource topology network; Based on the influence propagation delay in the time-space pulse response field, the propagation path of the fluctuation in the system resource topology network is simulated to form a polarization propagation trajectory; Identify the data element nodes in the propagation trajectory where the resource polarization effect strength is focused and enhanced, and locate them as key polarization convergence nodes.

7. The method of claim 6, wherein, The process of identifying the data element nodes in the propagation trajectory where the resource polarization effect strength is focused and enhanced includes: Calculate the second derivative of the resource polarization strength of the data element node in the continuous monitoring period; When the second derivative is greater than 0, it is determined that the resource polarization strength of the data element node is in a focused enhancement state.

8. The method of claim 1, wherein, The pulse attenuation strategy includes imposing resource request frequency limits and parallel processing quota constraints on the strong interference data element pair that triggered the pulse resonance; Wherein, the resource request frequency limit is represented by a frequency limit coefficient, and the parallel processing quota constraint is represented by a quota constraint coefficient; The values of the frequency limit coefficient and the quota constraint coefficient are dynamically adjusted according to the amplitude of the resource polarization effect strength exceeding the preset dynamic tolerance value.

9. The method of claim 1, wherein, According to the pulse attenuation strategy, the instantaneous attenuation rate of the resource polarization effect is obtained, and according to the instantaneous attenuation rate of the resource polarization effect, the process of performing gradual redistribution of resources includes: Monitor the resource polarization effect strength of the key polarization convergence node and calculate its instantaneous attenuation rate over time; When the instantaneous attenuation rate exceeds the preset attenuation rate trigger threshold, it is determined that the polarization effect of the key polarization convergence node enters the relief channel, and the resource redistribution operation is started; Identify the low polarization area; Establish a positive feedback control model of instantaneous attenuation rate and resource allocation speed; Continue to perform gradual redistribution until the resource polarization effect strength of the key polarization convergence node is completely dissipated and the system is restored to stability.

10. The data element based resource configuration system of claim 1, wherein, It includes: A resource pulse perception module for monitoring the resource access behavior of each data element and generating a resource pulse sequence; A time-space pulse response field construction module for constructing a time-space pulse response field between data elements based on the resource pulse sequence; A causal link analysis module for identifying strong interference data element pairs in the time-space pulse response field and analyzing the causal influence link between the strong interference data element pairs; A resource polarization effect calculation module for calculating the resource polarization effect caused by the pulse resonance event when a pulse resonance event is detected on the causal influence link; A key polarization convergence node positioning module is configured to track a polarization propagation track of the resource polarization effect according to a system resource topology, and locate a key polarization convergence node in which the resource polarization effect in the polarization propagation track is in a focused enhancement state; A pulse attenuation control module is configured to start a pulse attenuation strategy for a strong interference data element causing the pulse resonance, when the resource polarization effect of the key polarization convergence node exceeds a preset dynamic tolerance value. A dynamic resource allocation module is configured to obtain an instantaneous attenuation rate of the resource polarization effect according to the pulse attenuation strategy, and perform a gradual redistribution of resources according to the instantaneous attenuation rate of the resource polarization effect.

Images