A wisdom network computing power resource management system

Abstract

The present application relates to the technical field of computing power management, in particular to a wisdom network computing power resource management system, the system comprises a node response load carding module, a task computing power grade division module, a computing power scheduling adaptation module, a link load block detection module and a computing power flow direction reprogramming module. In the present application, by tracking the continuous change and reverse frequency of the node response direction, combining the computing power usage state and the cache occupation ratio, dynamically evaluating the node stability, extracting the task logic depth and the cache interaction deviation to divide the computing power grade, realizing the hierarchical matching of task demand and node capability, screening out the nodes with high cache load but unstable response, improving the accuracy of resource scheduling and the matching degree of task allocation, monitoring the abnormal growth trend and queuing backlog ratio of the link, identifying the transmission congestion risk in time, optimizing the replaceable node reconfiguration flow based on the idle computing power and the bearing ratio, and improving the agility of resource calling and the continuity of task execution in complex network.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computing power management, in particular to a wisdom network computing power resource management system. BACKGROUND

[0002] The computing power management technical field belongs to the key technical direction of the intersection of information technology and computer science, mainly involving related technologies for unified scheduling, allocation and optimized management of distributed computing resources, covering computing power scheduling strategies, resource perception mechanisms, computing node selection, load balancing methods, resource collaboration in heterogeneous environments, edge and cloud computing fusion scheduling and other core matters. The development of this technical field is committed to improving the use efficiency and response capability of computing power resources, supporting the efficient operation of complex computing tasks in a multi-source heterogeneous environment, and is a basic guarantee technology for supporting artificial intelligence, big data analysis, Internet of Things applications and intelligent network systems.

[0003] Among them, the traditional wisdom network computing power resource management system refers to a system that coordinates and controls cloud and edge computing resources in an intelligent network scenario. The technical matter addressed by this patent subject is how to effectively manage and schedule multi-source heterogeneous computing power resources in an intelligent network environment. The traditional wisdom network computing power resource management system usually adopts a static computing power description method based on a resource representation model, establishes a resource information library by predefining node characteristics and computing power indicators, and then combines a static scheduling algorithm based on the shortest task completion time or minimum energy consumption for task matching and resource allocation.

[0004] The existing system is based on a static model, and the node state description lacks the ability to capture real-time fluctuations and cache interaction dynamics. The task matching process cannot reflect the correlation between task levels and resource carrying capacity. In the face of frequent changes in node load or sustained response pressure drop in the environment, the scheduling logic is prone to matching errors, and the scheduling efficiency decreases significantly in the cache high-occupancy node scenario. The lack of link monitoring leads to a lag in identifying transmission bottlenecks, and the task dispatching does not form a feedback loop with the link state. In the case of blocked data channels, the original path output is still maintained, forming a phenomenon of task transmission backlog and resource idling coexistence, which limits the system's ability to adapt to task deployment and control efficiency in complex network structures. SUMMARY

[0005] The purpose of the present application is to solve the shortcomings in the prior art and to propose a wisdom network computing power resource management system.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: a wisdom network computing power resource management system comprises:

[0007] The node response load carding module references the node response fluctuation record in the wisdom network computing resource management, analyzes the continuous change direction, the adjacent response time reversal and the cumulative number, combines the node computing power usage, the cache read-write backlog and the input-output processing ratio, and generates a node computing power stable state table;

[0008] The task computing power grade division module extracts the task logic call depth, the instruction total number and the cache interaction number based on the node computing power stable state table, calculates the difference value and divides the grade, matches the task intensity, and generates a task computing power usage level list;

[0009] The computing power scheduling adaptation module references the task computing power usage level list, analyzes the online node computing power occupation and the cache occupation, filters the nodes matched with the computing power margin and the task intensity, eliminates the cache near-saturation nodes, and generates a task available computing power node set;

[0010] The link load blockage detection module monitors the transmission exception and the queuing amount based on the task available computing power node set, analyzes the continuously rising backlog proportion, and generates a link transmission blockage list;

[0011] The computing power flow direction reprogramming module references the link transmission blockage list and the unallocated node, filters the ratio node replacement connection, adjusts the task flow direction, and generates a computing power resource management result.

[0012] As a further scheme of the application, the node computing power stable state table includes the direction reversal number, the response time fluctuation frequency, the cache access processing ratio, and the node computing power usage proportion, the task computing power usage level list includes the logic call depth interval, the instruction call number grade, the cache interaction frequency range, and the call interaction difference value section, the task available computing power node set includes the computing power margin node, the cache load controllable node, and the online adjustable node, the link transmission blockage list includes the abnormal transmission frequency, the connection node queuing intensity, and the task backlog growth proportion, and the computing power resource management result includes the blockage node number, the replacement node number, and the adjusted task flow path.

[0013] The computing power margin and the task intensity matched node refers to the remaining computing power sufficient to meet the required intensity requirement of the specified task, and the cache occupation of the online node is not close to saturation.

[0014] The ratio node replacement connection node refers to the node selected from the unallocated node to replace the original connection node in the link blockage situation.

[0015] As a further scheme of the application, the node response load carding module includes:

[0016] The response fluctuation extraction submodule obtains node response fluctuation records, extracts continuous response directions, judges whether adjacent response time directions are consistent, records continuous direction sequences, and generates node direction change trend sequences;

[0017] The inversion identification submodule calls the node direction change trend sequences, compares whether adjacent directions are opposite, counts the number of direction inversions, combines the node computing power usage, and generates a direction inversion strength result;

[0018] The stable state generation submodule obtains the cache read-write backlog and input-output processing amount based on the direction inversion strength result, calculates the cache access and processing ratio, calls the inversion strength comparison frequency and ratio relationship, and generates a node computing power stable state table;

[0019] The node computing power usage refers to the total amount of computing power resources consumed by the node in executing tasks within a time period.

[0020] As a further scheme of the application, the task computing power level division module comprises:

[0021] The task parameter extraction submodule obtains the node computing power stable state table, extracts the logical call depth, instruction call total number and cache interaction number corresponding to each task, integrates the three data according to the task dimension, establishes a task execution structure item set, and generates a task execution index set;

[0022] The strength difference value calculation submodule calls the task execution index set, calculates the difference value of the logical call depth and cache interaction number in each record, matches the logical and cache difference value intervals according to the difference value, and generates a task strength level reference table;

[0023] The level list generation submodule filters and classifies the execution strength of differentiated level tasks according to the level labels in the task strength level reference table combined with the instruction call total number, constructs the computing power distribution records under the level tasks, and generates a task computing power usage level list;

[0024] The task execution structure item set refers to the integrated set of three core parameters of the logical call depth, instruction call total number and cache interaction number corresponding to each task;

[0025] The difference value matching logic refers to a matching rule for mapping the difference value of the logical call depth and cache interaction number to a preset task strength level;

[0026] The cache difference value interval refers to the strength level division section corresponding to the difference value of the logical depth and cache interaction according to the task type;

[0027] The execution strength refers to the degree of consumption of node computing power resources by the task in the running process.

[0028] As a further scheme of the present application, the computing power scheduling adaptation module comprises:

[0029] The node state extraction submodule obtains the task computing power usage hierarchy list, extracts the resource state of the online node, reads the computing power occupation value of the node, calculates the computing power occupation distribution index, integrates the associated index and the node identifier in structure, and generates an online resource occupation index set;

[0030] The resource matching judgment submodule calls the online resource occupation index set, compares the current computing power occupation value of the node with the task intensity value according to the computing power intensity level record in the task list, and generates a task adaptation node list;

[0031] The computing power node screening submodule calls the task adaptation node list, judges whether the cache occupation proportion corresponding to the node is in a saturated state, eliminates the node whose cache occupation is in a saturated interval, retains the remaining nodes as schedulable targets, generates a task available computing power node set, and outputs the task available computing power node set.

[0032] The computing power occupation value refers to the number and proportion of computing power resources occupied by the node being executed tasks;

[0033] The node whose cache occupation is in a saturated interval refers to a node whose cache resource utilization rate has reached and exceeded a preset upper threshold, causing a performance bottleneck.

[0034] As a further scheme of the present application, the computing power occupation distribution index calculation formula is specifically:

[0035] ;

[0036] Wherein, represents the computing power occupation distribution index of the kth node, represents the computing power usage value of the kth node on the mth task, represents the average computing power usage value of the kth node for all tasks, represents the total amount of cache occupation of the kth node, represents the total number of tasks currently processed by the kth node, represents a small constant to prevent division by zero.

[0037] As a further scheme of the present application, the link load blocking detection module comprises:

[0038] The abnormal record monitoring submodule calls the task available computing power node set, monitors the transmission abnormal number of connected nodes, records the abnormal event growth situation in time sequence, eliminates the record items without continuous change, and generates an abnormal event growth sequence;

[0039] The queuing percentage calculation submodule calls the abnormal event growth sequence, extracts the task queuing amount and the task capacity that the node can carry corresponding to the connected node, calculates the queuing percentage index value, determines whether the node ratio is rising continuously, filters the records of the upward trend, and generates the queuing percentage change trend.

[0040] The blocking list generation submodule identifies the node identifiers corresponding to continuously rising records based on the queuing ratio change trend, summarizes the abnormal sequences and queuing status, removes node items that do not meet the conditions, and generates a link transmission blocking list.

[0041] The record items with no continuous changes refer to node record data where the number of abnormal events does not continuously increase in adjacent time periods;

[0042] The upward trend record refers to the record item in which the queuing percentage index value shows an increasing trend in a continuous time point;

[0043] The node items that do not meet the conditions refer to node data that do not show a continuous upward trend in the abnormal growth and queuing ratio analysis and do not meet the blocking judgment criteria.

[0044] As a further aspect of the present invention, the formula for calculating the queuing ratio index is as follows:

[0045] ;

[0046] in, Representing the Each connection node in time Queue percentage at any given time. Representing the Each connection node in time The amount of tasks queued at any given time. Representing time Time of the first The weight of the tasks to be added to the team Representing time Time of the first The estimated processing time for each pending task to be added to the team. Representing the Each connection node in time Maximum task capacity at any given time. Indicates time The total number of all pending tasks associated with a given moment.

[0047] As a further aspect of the present invention, the computing power flow reprogramming module includes:

[0048] The blocking node positioning sub-module extracts the node number corresponding to each blocking record according to the link transmission blocking list, arranges the node number and the blocking information to form a corresponding mapping item, eliminates duplicate numbers and performs aggregation, and generates a blocking node number set;

[0049] The replacement node screening sub-module calls the blocking node number set, reads the idle computing power and the task carrying capacity of the current unallocated node in the intelligent network computing power resource management, calculates the computing power difference index of the unallocated node, compares the ratio of all unallocated nodes, screens the ratio dominant node for numbering, and generates a replacement candidate node set;

[0050] The task flow direction adjustment sub-module calls the replacement candidate node set, re-directs the corresponding task to the replacement node according to the connection path structure of the original blocking node, restructures the task scheduling relationship, updates the node connection and the task allocation path, and generates a computing power resource management result;

[0051] The repeated number refers to the same node number appearing multiple times in the link transmission blocking list and corresponding to multiple blocking records;

[0052] The ratio dominant node numbering refers to the individual with the highest computing power difference index ratio in the unallocated node, and the node number of the individual is marked.

[0053] As a further scheme of the present application, the computing power difference index calculation formula of the unallocated node is specifically:

[0054] ;

[0055] Wherein, represents the computing power difference index of the unallocated node numbered represents the idle computing power value of the unallocated node numbered in the dimension, represents the task carrying capacity of the unallocated node numbered represents the number of idle computing power dimensions involved in the calculation, represents the scheduling delay weight value corresponding to the unallocated node numbered is a positive offset factor.

[0056] Compared with the prior art, the present application has the following advantages and positive effects:

[0057] ​​​In the application, by tracking the continuous change and reversal frequency of the response direction of the node, combining the algorithm power usage state and the cache occupation ratio, dynamically evaluating the node stability, extracting the task logic depth and cache interaction deviation to divide the algorithm power level, realizing the hierarchical matching of task demand and node capability, screening out the nodes with high cache load but unstable response, improving the accuracy of resource scheduling and the matching degree of task allocation, monitoring the abnormal growth trend of the link and the queuing backlog ratio, identifying the transmission congestion risk in time, optimizing the alternative node reconstruction flow based on the idle algorithm power and the bearing ratio, and improving the agility of resource calling and the continuity of task execution in the complex network. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 is a system flowchart of the application;

[0059] Figure 2 is a system block diagram of the application;

[0060] Figure 3 is a node response load combing module flowchart of the application;

[0061] Figure 4 is a task algorithm power level division module flowchart of the application;

[0062] Figure 5 is an algorithm power scheduling adaptation module flowchart of the application;

[0063] Figure 6 is a link load blocking detection module flowchart of the application;

[0064] Figure 7 is an algorithm power flow reprogramming module flowchart of the application. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical scheme and advantages of the application clearer, the application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0066] In the description of the application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application. In addition, in the description of the application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0067] Please refer to Figure 1 and Figure 2 A wisdom network computing resource management system comprises:

[0068] The node response load carding module obtains the node response fluctuation record in the wisdom network computing resource management, extracts the continuous change direction, judges the inversion phenomenon of adjacent response time, accumulates the number of direction inversion, reads the node computing power usage, extracts the cache read-write backlog and input-output processing amount, compares the cache access and input-output processing proportion, combines the fluctuation frequency and processing proportion, and generates a node computing power stable state table;

[0069] The task computing power level division module extracts the task logic call depth, instruction call total number and cache interaction times according to the node computing power stable state table, calculates the difference value of logic call and cache interaction, divides the level according to the difference value interval, matches the execution intensity and the level, and generates a task computing power usage level list;

[0070] The computing power scheduling adaptation module references the task computing power usage level list, extracts the online node resource state, reads the computing power occupation value and cache occupation proportion, judges the matching of computing power margin and task intensity, eliminates the nodes with nearly saturated cache occupation, and forms a task available computing power node set;

[0071] The link load blockage detection module calls the task available computing power node set, monitors the number of transmission exceptions, records the continuous abnormal growth, extracts the connection node task queuing amount, calculates the proportion of backlog tasks to the loadable amount, judges the continuous rising situation of the proportion, and generates a link transmission blockage list;

[0072] The computing power flow reprogramming module locates the blockage node number according to the link transmission blockage list, calls the unallocated node in the wisdom network computing resource management, reads the idle computing power and the load number, selects the node with large ratio value to replace the connection, adjusts the task flow direction, and generates a computing power resource management result.

[0073] The node computing power stable state table includes the number of direction inversion, the response time fluctuation frequency, the cache access processing proportion, and the node computing power usage proportion. The task computing power usage level list includes the logic call depth interval, the instruction call number level, the cache interaction frequency range, and the call interaction difference value section. The task available computing power node set includes the computing power margin node, the cache load controllable node, and the online adjustable node. The link transmission blockage list includes the abnormal transmission frequency, the connection node queuing strength, and the task backlog growth proportion. The computing power resource management result includes the blockage node number, the replacement node number, and the adjusted task flow path.

[0074] Please refer to Figure 3 and Figure 2 The node response load carding module comprises:

[0075] The response fluctuation extraction submodule obtains node response fluctuation records, extracts continuous response directions, judges whether adjacent response time directions are consistent, records continuous direction sequences, and generates node direction change trend sequences;

[0076] First, set the sampling period to every 0.5 seconds, periodically sample the response of the node, and continuously sample for 10 seconds to obtain 20 groups of response time data, for example, the response sequence is 12 ms, 14 ms, 14 ms, 13 ms, 11 ms, 10 ms, 10 ms, etc. After the system reads each group of response time, it compares the response difference with the previous group to determine whether there is a change. When the response time changes from 12 ms to 14 ms, the difference is positive, which is determined as a positive fluctuation. If it decreases from 14 ms to 13 ms, it is a negative fluctuation. When the response time of two groups is the same, the direction is not recorded. The system compares each group of sampling pairs in turn, marks each point with a change direction as "positive" or "negative" according to the change direction, and constructs a response direction sequence such as "positive, no, negative, negative, no". Among them, "no" indicates that the response time of the two consecutive points is the same, and the system skips this record. Only the "positive" and "negative" directions are continuously spliced into a direction sub-sequence to perform trend segment division. In trend segment identification, the first effective direction is recorded from the start of the sequence and the initial segment index is set. Then compare each direction with the previous direction. If they are consistent, they belong to the current segment. If they are not consistent, a new segment is started, and the start and end index values and direction labels of the previous segment are recorded. For example, three segments are formed: "positive for 2 times", "negative for 3 times", and "positive for 2 times". Each segment is marked by an index position and combined into a direction change trend sequence. Through the sequence, the continuous direction change of the node in different response stages can be identified and processed.

[0077] The reverse identification submodule calls the node direction change trend sequence, compares whether the adjacent directions are opposite, counts the number of direction reversals, combines the node computing power usage, and generates a direction reversal strength result;

[0078] The reverse recognition submodule calls the generated direction change trend sequence, compares the direction labels of the current segment and the next segment from the starting position of the sequence, for example, when the trend sequence is "positive 3 times", "reverse 2 times", "reverse 2 times", "positive 1 time", and "reverse 1 time", the system compares the directions of adjacent segments, such as the first and second segments, the second and third segments, and so on. If the directions are opposite, it is recorded as a reverse, and if they are the same, it is skipped. In the above sequence, the directions of the first and second segments are opposite, and one reverse is recorded. The directions of the second and third segments are the same, and are skipped. The directions of the third and fourth segments are different, and another reverse is recorded. In this way, the number of reverses is counted, and three direction reverses are identified. Then, the computing power usage in the current period of the node is called, which is obtained by the ratio of the number of tasks processed by the node and the maximum processing capacity set. For example, the maximum processing capacity of the node is 100 task units, and the current processing is 65 task units, so the usage is 0.65. The number of reverses is multiplied by the computing power usage to obtain the direction reverse intensity, that is, 3*0.65=1.95. This intensity value is used for subsequent stability determination. In order to ensure the consistency of the determination standard, the system presets four reverse intensity intervals, sets the maximum number of reverses to 10, and combines the maximum computing power usage of 1 to set the highest intensity to 10. Four intervals are divided in proportion: low intensity (0-2.5), medium intensity (2.5-5), high intensity (5-7.5), and extremely high intensity (7.5-10). The current value 1.95 is in the low intensity interval.

[0079] The stable state generation submodule obtains the cache read-write backlog and input-output processing capacity based on the direction reverse intensity result, calculates the cache access and processing ratio, calls the reverse intensity comparison frequency and ratio relationship, and generates the node computing power stability state table.

[0080] The steady state generation submodule further reads the cache read-write backlog and input-output processing capacity data in the current period of the node based on the direction reversal intensity value, obtains the cache access and processing ratio, for example, the backlog is 120 MB, the processing capacity is 300 MB, the ratio is 120 / 300=0.4, the ratio value reflects the current cache data processing load situation, the smaller the ratio, the more timely the processing, otherwise it indicates that the cache load is heavy, the system compares the ratio with the direction reversal intensity value, uses a two-dimensional state determination table to complete the steady state recognition, wherein the cache ratio is divided into four intervals: low ratio (0-0.3), medium ratio (0.3-0.6), high ratio (0.6-0.8), and very high ratio (0.8-1.0). According to the actual setting, when the ratio is 0.4, it belongs to the medium ratio, and the aforementioned reversal intensity value is 1.95, which belongs to the low intensity. In the state matching table, the steady level corresponding to the low intensity and the medium ratio is "medium high", the system takes this level as the current node computing power steady state, and records the state together with the related parameter values in the calculation process to the steady state table to form a single record entry, including the reversal intensity value 1.95, the cache ratio value 0.4, and the steady state level "medium high". It is used for unified management and retrieval analysis of the subsequent node running state.

[0081] Please refer to Figure 4 and Figure 2 The task computing power level division module includes:

[0082] The task parameter extraction submodule obtains the node computing power steady state table, extracts the logical call depth, instruction call total number and cache interaction times corresponding to each task, integrates and records the three data according to the task dimension, establishes a task execution structure item set, and generates a task execution index set;

[0083] First, the computing power state field related to each specific task is extracted according to each record in the table, the task is associated with the corresponding record in the stable state table through the unique identification code of the task, and after obtaining the corresponding record of each task, the logical call depth, instruction call total and cache interaction times are read in turn. The logical call depth is identified by the recursive and nested number of layers of the task in the multi-layer function or module, for example, a task calls three nested function modules in turn and the recursive depth of each layer is 1, so the logical call depth is 3. The instruction call total is recorded by the number of assembly or intermediate instructions counted by the processor during the task execution period, for example, a task executes 15000 intermediate instructions during the execution process, and the instruction call total is recorded as 15000. The cache interaction times refer to the total number of read and write operations of the node cache system for the task, for example, the cache is read 90 times and written 60 times, so the cache interaction times of the task is 150 times. The system integrates the above three data into a structured record according to the task dimension, each record binds the task ID and puts the three data fields into a unified data structure to form a task execution structure item set. After all the task records are completed, all the structure items are batched and summarized to form a unified task execution index set. The index set is used for subsequent classification and comparison of different tasks in execution intensity and hierarchical division.

[0084] The intensity difference calculation submodule calls the task execution index set, calculates the difference between the logical call depth and the cache interaction times in each record, matches the difference value interval according to the difference value, and generates a task intensity level comparison table.

[0085] First, the logical call depth and cache interaction times fields in each record are read for numerical difference calculation. The difference value item is obtained by subtracting the cache interaction times from the logical call depth, for example, the logical call depth of a task is 8 layers and the cache interaction times is 6 times, the calculation result is 2, and if the logical call depth is 5 layers and the cache interaction times is 9 times, the result is−4. The system matches the preset logical and cache difference interval according to the difference value, sets the difference interval division as follows: interval A is difference value greater than or equal to 5, interval B is difference value between 2 and 5, interval C is difference value between−2 and 2, and interval D is difference value less than−2. Each interval corresponds to a task intensity level, respectively "high intensity", "medium intensity", "general intensity" and "low intensity". The difference value item is compared with the above interval, if the difference value is 6, it matches interval A, corresponding to "high intensity", and if the difference value is−3, it matches interval D, corresponding to "low intensity". The system labels the corresponding level label of the task in the record item after completing the matching of each record. After the difference calculation and level assignment of all task records are completed, the system generates a task intensity level comparison table. Each record in the table contains task ID, difference value and matching level fields, which are used for hierarchical classification of execution intensity in the next stage.

[0086] The level list generation submodule filters and classifies the execution intensity of the differentiated level tasks according to the level labels in the task intensity level table and the total number of instruction calls, constructs the computing power distribution record under the level task, and generates the task computing power use level list;

[0087] According to the level labels in the task intensity level table, the level identification field is traversed piece by piece, the task records under each level are extracted, and the total number of instruction calls is summarized and classified. In each level category, the system filters the record items with the same level label field in all task records, reads the corresponding total number of instruction calls parameters, and performs hierarchical statistics and task distribution construction. For example, there are 10 tasks under the "high intensity" level, and the total number of instruction calls is 13000, 14500, 16000, etc. The system calculates the average number of instruction calls and the distribution interval according to these values, and then groups them into computing power use distribution segments. For example, the total number of calls below 14000 is classified into the "low segment", 14000 to 16000 is classified into the "normal segment", and above 16000 is classified into the "high segment". The threshold is set based on the mean standard of the 95% confidence interval in the experience data or training set statistics. For example, the average total number of calls is 15000, and the standard deviation is 2000. The normal segment can be set to the range of 13000 to 17000. Below 13000 or above 17000 is an abnormal segment. Finally, the system arranges the call distribution of each level task into a task computing power use level list in a structured record form. Each record contains four fields: level label, task ID, total number of instruction calls, and distribution segment identifier, which are used for subsequent node resource scheduling and task allocation strategy reference.

[0088] Please refer to Figure 5 and Figure 2 The computing power scheduling adaptation module includes:

[0089] The node state extraction submodule obtains the task computing power use level list, extracts the resource state of the online node, reads the computing power occupation value of the node, calculates the computing power occupation distribution index, integrates the associated index and node identifier in a structure, and generates an online resource occupation index set.

[0090] The computing power occupation distribution index calculation formula is specifically:

[0091] ;

[0092] Among them, represents the computing power occupation distribution index of the kth node, represents the computing power use value of the kth node on the mth task, represents the average computing power use value of all tasks of the kth node, represents the total amount of cache occupation of the kth node, Total number of tasks currently processed by the kth node, A small constant to prevent division by zero;

[0093] The computing power occupation distribution index is used to measure the balance of computing power usage and cache usage of a node in multiple tasks, which is a key indicator for evaluating the load state of the node. Its calculation formula integrates the computing power usage value, average usage value, cache occupation and task number of the node in different tasks;

[0094] The formula logic is:

[0095] First, square the sum of the difference between the computing power usage value of the node on all tasks and its average value (measure the degree of imbalance);

[0096] Then add the cache occupation of the node divided by the number of tasks as the second part (measure the resource density);

[0097] Finally, set a small constant to avoid division by zero;

[0098] The larger the result, the more uneven the computing power distribution of the node or the higher the cache pressure, and the scheduling system can identify the unreasonable node according to this and avoid its participation in high-intensity task allocation;

[0099] Calculate the computing power occupation distribution index of the kth node , you need to get the following parameters:

[0100] : The computing power usage value of the kth node on the mth task. Collect CPU usage rate data through node monitoring system (such as Prometheus), unit is percentage (%). For example, the CPU usage rate of node k on task 1 to task 4 is 65%, 70%, 60%, 75% respectively.

[0101] : The average computing power usage value of the kth node on all tasks. The calculation formula is:

[0102] ;

[0103] Substitute the above data:

[0104] ;

[0105] : The total cache occupation of the kth node. Get the cache usage through cache monitoring tool (such as RedisINFO command), unit is GB. Suppose the cache occupation of node k is 2.5GB.

[0106] n: the total number of tasks currently processed by the kth node. Through the task scheduling system (such as Kubernetes), it is known that node k is currently processing 4 tasks, i.e. n = 4.

[0107] δ: a small constant to prevent division by zero, set to 0.01.

[0108] Substitute the above values into the formula to calculate :

[0109] First part calculation:

[0110] ;

[0111] Second part calculation:

[0112] ;

[0113] Therefore, The calculation result is:

[0114] ;

[0115] The results show that the computing power of the kth node under the current task allocation is relatively balanced (the first part is 0), but the cache occupancy is relatively high (the second part is 1.2438), and attention should be paid to the allocation and use of cache resources to avoid potential performance bottlenecks.

[0116] The resource matching judgment submodule calls the online resource occupation index set, compares the current computing power occupation value of the node with the task intensity value according to the computing power intensity level record in the task list, and generates a task adaptation node list;

[0117] The resource matching judgment submodule calls the online resource occupation index set, reads the task intensity level record in each task list while traversing all online node records, compares the adaptability between the task level and the distribution level of the current computing power occupation value of the node, and matches the rules as follows: if the task is "high intensity", the node needs to be in the "low occupation" or "medium occupation" interval; if the task is "medium intensity", the node can be adapted to "medium occupation" or "high occupation"; and if the task is "low intensity", the node of any occupation level can be accepted. In this process, the system does not directly use fuzzy logic, but uses value interval division and task-node level comparison table for matching. Assuming that the task intensity level is "high intensity", the matching node computing power occupation value needs to be less than 0.7. If the node occupation value is 0.65, it is determined that the adaptation is successful, otherwise if the node occupation value is 0.85, it does not meet the condition, so the node is filtered out. After the system completes the adaptation judgment between all tasks and nodes, a corresponding task adaptation node list is generated for each task. Each record in the list includes the task ID, task intensity level, node ID and node computing power occupation value fields, which are used for subsequent cache state judgment and node screening.

[0118] The computing power node screening submodule calls the task adaptation node list, judges whether the cache occupation ratio of the node is in a saturated state, removes the nodes whose cache occupation is in the saturated interval, and retains the remaining nodes as schedulable targets to generate a task available computing power node set.

[0119] The computing power node screening submodule calls the task adaptation node list, further extracts the cache usage of the adaptation node under each task record, reads the cache occupation ratio value of each node in the current period, which is calculated by dividing the used capacity of the node cache by the maximum capacity of the cache. For example, if the total capacity of the cache of a node is 500MB and the current usage is 450MB, the occupation ratio is 0.9. The system sets the cache occupation interval as follows: normal interval (0-0.8), warning interval (0.8-0.9), and saturated interval (0.9-1). The interval is set in reference to the critical threshold of cache response timeout in the system running history. If the node cache occupation ratio is greater than or equal to 0.9, it is considered to be in a saturated state. The system traverses each node in the task adaptation node list in turn. If the node is in the saturated interval, it is removed from the candidate target node of the current task. The remaining nodes are retained as an effective target node set to form a task available computing power node set for each task. Each record in the set retains the task ID, node ID, computing power occupation value and cache occupation ratio fields, which are used by the scheduler to implement task allocation based on resource conditions.

[0120] Please refer to Figure 6 and Figure 2 The link load blocking detection module comprises:

[0121] The abnormal record monitoring submodule calls the task available computing power node set, monitors the transmission abnormal number of times of the connected node, records the abnormal event growth in time sequence, eliminates the record items without continuous change, and generates an abnormal event growth sequence;

[0122] The abnormal record monitoring submodule calls the task available computing power node set, the system collects abnormal data of each connected node in the task schedulable node list, the collection content is the transmission abnormal number of times of each node in each time period, the abnormal type is limited to the abnormal state event caused by communication packet loss, response delay or protocol interruption, the system collects the abnormal count value of each node once at a fixed time interval in the sampling period, for example, the abnormal number of times of node A in five consecutive periods is 1, 3, 4, 4 and 6, the system summarizes the abnormal number of times of the node in time sequence, then judges the difference value of adjacent data, if the difference value of the abnormal number of times of two consecutive sampling points is 0, it is regarded as a record item without change, for example, in the above example, from 3 to 4 is +1, from 4 to 4 is 0, and from 4 to 6 is +2, the system removes the time point corresponding to the middle pair of values (4, 4) from the sequence, and only retains the items with positive change, that is, the record item needs to meet the continuous growth condition, the record item constitutes an abnormal event growth sequence, if the abnormal number of times of the node appears a downward or stagnation trend, the record of the corresponding time point is not retained, in this way, the system traverses all the connected nodes in the task available node set one by one, extracts all the record items with abnormal growth trend, and groups and summarizes them according to the nodes to form an abnormal event growth sequence for subsequent queue state evaluation.

[0123] The queue proportion calculation submodule calls the abnormal event growth sequence, extracts the task queue amount and the node bearable task capacity corresponding to the connected node, calculates the queue proportion index value, judges whether the node proportion is continuously rising, selects the rising trend record, and generates a queue proportion change trend;

[0124] The queue proportion index value calculation formula is specifically:

[0125] ;

[0126] Wherein, represents the queue proportion index value of the i-th connected node at time t, represents the task queue amount of the i-th connected node at time t, represents the weight of the i-th to-be-queued task at time t, represents the queue proportion index value of the i-th connected node at time t, represents the task queue amount of the i-th connected node at time t, represents the weight of the i-th to-be-queued task at time t, represents the queue proportion index value of the i-th connected node at time t, represents the task queue amount of the i-th connected node at time t, represents the weight of the i-th to-be-queued task at time t, represents the queue proportion index value of the i-th connected node at time t, represents the task queue amount of the i-th connected node at time t, The estimated processing time for each pending task to be added to the team. Representing the Each connection node in time Maximum task capacity at any given time. Indicates time The total number of all pending tasks associated with a given moment;

[0127] The queuing percentage metric reflects the ratio of a node's current task queuing status to its maximum task capacity, and is an important signal of link congestion. Its calculation considers factors such as the number of tasks, their weights, estimated processing time, and the node's maximum task capacity.

[0128] The calculation logic is as follows:

[0129] For all pending tasks, first calculate a weighted index based on their weight and expected processing time;

[0130] Then compare the number of queued tasks at the current node with its maximum processing capacity to determine the queuing level;

[0131] If this indicator rises at multiple consecutive time points, it indicates that the node is becoming increasingly congested.

[0132] This metric can help identify network bottleneck nodes so that the scheduler can adjust the flow in a timely manner to avoid congestion;

[0133] In time At that moment, for the first Each connection node collects and calculates the following parameters:

[0134] The current task queue size of a node, obtained through a real-time monitoring system, is expressed in the number of tasks.

[0135] The maximum task capacity that a node can handle is determined based on the node's hardware configuration and historical load data, and is expressed in the number of tasks.

[0136] : No. The estimated processing time for each pending task is based on historical processing data and task complexity assessment, and is expressed in seconds.

[0137] : No. The weight of tasks awaiting entry into the team is set according to the task's priority and resource consumption, and has no unit.

[0138] The total number of tasks currently awaiting joining the team is calculated in real time.

[0139] The specific values ​​are set as follows:

[0140] ;

[0141] ;

[0142] ;

[0143] 、 、 ;

[0144] 、 、 ;

[0145] Calculate the weighted square root processing duration of each task:

[0146] Calculate the weighted square root processing duration of each task:

[0147] ;

[0148] ;

[0149] ;

[0150] Sum up the total weighted square root processing duration:

[0151] ;

[0152] ;

[0153] ;

[0154] Sum up the total weighted square root processing duration:

[0155] ;

[0156] Substitute the formula to calculate the queuing ratio index value:

[0157] ;

[0158] The results show that at time , the queuing ratio index value of the th connection node is 0.424, indicating the relative difference between the current task queue amount and the node's task capacity. This index value can be used to judge the node load, and assist in formulating resource scheduling and load balancing strategies.

[0159] The blocking list generation submodule identifies the node identifier corresponding to the continuous rising record according to the queuing proportion change trend, summarizes the abnormal sequence and queuing state, eliminates the node items that do not meet the conditions, and generates a link transmission blocking list;

[0160] The blocking list generation submodule reads the node identifier in each record according to the queuing proportion change trend record, and cross-compares the period number in the abnormal event growth sequence and the period number in the queuing proportion sequence. If there is an overlapping period and the queuing proportion in the overlapping period segment continues to rise, the system will identify the node identifier as a blocking suspected target. Further, the abnormal growth frequency cumulative value of the node in the overlapping segment is read. The abnormal cumulative frequency threshold is set to 5 times, and the queuing proportion final value is greater than or equal to 0.85 as the blocking confirmation condition. If the abnormal growth frequency of a node reaches 6 times in 4 periods and the final value of the queuing proportion is 0.88, the above two conditions are met, and the system identifies the node as an actual transmission blocking node. Otherwise, if the cumulative abnormal frequency is less than the threshold or the proportion value does not exceed the set threshold, the record is eliminated and does not enter the final result. After the system completes the judgment of all nodes, the identifiers of the confirmed blocking nodes, the abnormal cumulative frequency, the queuing proportion final value, and the trend period segment number are summarized and integrated into a structured record to generate a link transmission blocking list. This list serves as an input reference for the system to determine the link performance bottleneck. Each record in the list provides complete and traceable link status.

[0161] Please refer to Figure 7 and Figure 2 The computing power flow direction reprogramming module includes:

[0162] The blocking node positioning submodule extracts the node number corresponding to each blocking record according to the link transmission blocking list, organizes the node number and blocking information to form a corresponding mapping item, eliminates duplicate numbers and summarizes them to generate a blocking node number set.

[0163] The blocking node location submodule, based on the link transmission blocking list, first reads the node identifier field marked in each record, extracts the node number as the primary key, and generates a preliminary number list. This list may contain multiple records with the same node number. The system performs deduplication on the number list, merging all duplicate node numbers using a hash structure, retaining only the unique identifier. For example, if the preliminary record is (N1, N2, N3, N1, N4, N2), the system deduplicates it to (N1, N2, N3, N4). Next, each node number is mapped and assembled with its corresponding blocking information in the original list. For example, node N1 corresponds to 6 cumulative anomalies and a queuing percentage of 0.88, while N2 corresponds to 5 cumulative anomalies and a queuing percentage of 0.9. Each number corresponds one-to-one with its corresponding blocking information, constructing a mapping structure. The system summarizes all mapping items to form a structured dataset. Each record contains fields such as node number, anomaly count, and the last value of the queuing percentage. Finally, the blocking node number set is generated, which is used to lock the target node range during subsequent node replacement.

[0164] The replacement node filtering submodule calls the blocking node number set, reads the idle computing power and task carrying capacity of the currently unallocated nodes in the intelligent network computing power resource management, calculates the computing power difference index of the unallocated nodes, compares the ratio of all unallocated nodes, filters the nodes with the superior ratio, marks them with numbers, and generates a replacement candidate node set.

[0165] The specific formula for calculating the computing power difference index of unallocated nodes is as follows:

[0166] ;

[0167] in, Representative number is The computing power difference index of unassigned nodes, Representative number is The unassigned nodes in the first Idle computing power value per dimension Representative number is The number of tasks that unassigned nodes can handle. This represents the number of idle computing power dimensions involved in the computation. Representative number is The scheduling delay weight value corresponding to the unassigned node. A positive offset factor;

[0168] The computing power difference index is used to measure the resource advantage of the unassigned node, so as to select the optimal replacement node when the task is blocked. Its formula combines the idle computing power value of each dimension of the node (such as CPU, GPU, etc.), the current task carrying capacity, the scheduling delay weight and other factors, and reflects the ratio between the total amount of available computing power of the node and its load capacity. The calculation steps are:

[0169] Add the sum of the idle value and the delay weight of the node in all resource dimensions as the numerator;

[0170] Use the current task number of the node plus a small constant as the denominator;

[0171] The higher the index value, the more abundant the idle resources and the lighter the load of the node, which is suitable for use as a replacement node;

[0172] In this way, the system can dynamically select a replacement node when the network is blocked, maintaining scheduling continuity;

[0173] 1. Parameter acquisition and quantization method:

[0174] The idle computing power value of the unassigned node numbered in the dimension:

[0175] By monitoring the use of resources such as CPU, GPU, and memory of the node in real time, the idle computing power value of each dimension is obtained.

[0176] The task carrying capacity of the unassigned node numbered :

[0177] The number of tasks currently being processed by the node is counted as the task carrying capacity.

[0178] The number of idle computing power dimensions involved in the calculation:

[0179] According to the number of monitored resource dimensions, such as CPU, GPU, and memory.

[0180] The scheduling delay weight value corresponding to the unassigned node numbered :

[0181] By measuring the network delay, processing delay and other indicators of the node, the scheduling delay is calculated, and the corresponding weight value is assigned according to the delay degree.

[0182] The positive offset factor:

[0183] Set to a very small positive number, such as 0.01, to avoid the case of zero denominator.

[0184] Specific numerical settings:

[0185] Assume that the idle computing power of the node in three dimensions is:

[0186] (CPU idle computing power);

[0187] (GPU idle computing power);

[0188] (Memory idle computing power);

[0189] The number of tasks carried by the node ;

[0190] The number of idle computing power dimensions involved: ;

[0191] The scheduling delay weight value of the node ;

[0192] Positive offset factor: ;

[0193] Formula calculation process:

[0194] Calculate the first term:

[0195] ;

[0196] Calculate the second term:

[0197] ;

[0198] Add the two terms to get the value of :

[0199] ;

[0200] The result shows that the computing power difference index of the unallocated node is 12.883, which is used to measure the difference between the idle computing power of the node and the number of tasks carried, the larger the value, the greater the difference, which can be used to screen the ratio-dominant node for numbering marking, and generate a replacement candidate node set.

[0201] The task flow adjustment submodule calls the replacement candidate node set, re-directs the corresponding tasks to the replacement node according to the connection path structure of the original blocked node, and restructures the task scheduling relationship, updates the node connection and task allocation path, and generates a computing power resource management result;​​

[0202] First, read each node ID in the blocking node number set, and extract the upstream and downstream node relationship connected by the node in the topology graph structure, construct its original connection path structure, after the system identifies a complete path, for example, the path is composed of A→N1→B, and N1 is the blocking node, the system replaces the node identification at the position of N1 in the original path with the corresponding candidate node number, such as M3, and reconstructs the path as A→M3→B, in this replacement process, the system needs to read whether the candidate node currently exists with the task upstream and downstream nodes, if there is no connection, the system sends a connection instruction through the console, instructs the node to establish link communication, after the connection is successful, the system updates the scheduling mapping record in the task scheduling relationship table, modifies the original binding node N1 of the task to M3, the task number remains unchanged, only the target node field is updated, the path structure is updated synchronously, the node access table synchronously registers the new connection relationship, all changes are written into the computing power resource management system log synchronously, the system finally completes the scheduling target update of the task and the reorganization of the connection structure, and outputs the structure reorganization result record, forms the final computing power resource management result data set, each record contains the original blocking node, replacement node, original connection path, updated path and task number and other fields.

[0203] The above is only a preferred embodiment of the present application, and does not limit the present application in other forms. Any skilled person in the art can modify or change the above disclosed technical content to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made according to the technical essence of the present application to the above embodiments without departing from the technical solution content of the present application still belongs to the protection scope of the present application technical solution.

Claims

1. A smart network computing resource management system, characterized in that, The system includes: The node response load analysis module references node response fluctuation records from the intelligent network computing power resource management system, analyzes the direction of continuous change, the reversal of adjacent response times and the cumulative number of times, and generates a node computing power stability table by combining the node computing power usage, cache read / write backlog and input / output processing ratio. Based on the node computing power stability table, the task computing power level classification module extracts the task logic call depth, total number of instructions and cache interaction times, calculates the difference and classifies the levels, matches the task intensity, and generates a task computing power usage hierarchy list. The computing power scheduling and adaptation module references the task computing power usage hierarchy list, analyzes the computing power and cache usage of online nodes, filters nodes whose computing power reserves match the task intensity, removes nodes whose cache is close to saturation, and generates a set of nodes with available computing power for the task. The link load blocking detection module monitors transmission anomalies and queuing volume based on the set of available computing power nodes for the task, analyzes the continuous increase in the backlog ratio, and generates a list of link transmission blockages. The computing power flow reprogramming module references the link transmission blockage list and unallocated nodes, filters out nodes with a ratio to replace connected nodes, adjusts task flow, and generates computing power resource management results; The node computing power stability table includes the number of direction reversals, response time fluctuation frequency, cache access processing ratio, and node computing power utilization ratio. The task computing power utilization hierarchy list includes the logical call depth range, instruction call quantity level, cache interaction frequency range, and call interaction difference range. The task available computing power node set includes computing power reserve nodes, cache load controllable nodes, and online adjustable points. The link transmission blockage list includes abnormal transmission frequency, connection node queuing intensity, and task backlog growth ratio. The computing power resource management results include the blocking node number, the replacement node number, and the adjusted task flow path. The nodes whose computing power reserve matches the task intensity refer to online nodes whose remaining computing power is sufficient to meet the intensity requirements of the specified task and whose cache usage is not close to saturation. The node replacement connection of the ratio node: refers to replacing the original connection node with a node that has high task transmission efficiency selected from the unassigned nodes when the link is blocked. The link load blocking detection module includes: The anomaly recording monitoring submodule calls the set of available computing power nodes for the task, monitors the number of transmission anomalies of the connected nodes, records the growth of anomaly events in chronological order, removes records with no continuous changes, and generates an anomaly event growth sequence. The queuing percentage calculation submodule calls the abnormal event growth sequence, extracts the task queuing amount and the task capacity that the node can carry corresponding to the connected node, calculates the queuing percentage index value, determines whether the node ratio is rising continuously, filters the records of the upward trend, and generates the queuing percentage change trend. The blocking list generation submodule identifies the node identifiers corresponding to continuously rising records based on the queuing ratio change trend, summarizes the abnormal sequences and queuing status, removes node items that do not meet the conditions, and generates a link transmission blocking list. The record items with no continuous changes refer to node record data where the number of abnormal events does not continuously increase in adjacent time periods; The upward trend record refers to the record item in which the queuing percentage index value shows an increasing trend in a continuous time point; The node items that do not meet the conditions refer to node data that do not show a continuous upward trend in the abnormal growth and queuing ratio analysis and do not meet the blocking judgment criteria. The computing power flow reprogramming module includes: The blocking node location submodule extracts the node number corresponding to each blocking record based on the link transmission blocking list, organizes the node number and blocking information into a corresponding mapping item, removes duplicate numbers and summarizes them to generate a blocking node number set. The replacement node filtering submodule calls the blocked node number set, reads the idle computing power and task carrying capacity of the currently unallocated nodes in the intelligent network computing power resource management, calculates the computing power difference index of the unallocated nodes, compares the ratio of all unallocated nodes, filters the nodes with the superior ratio, marks them with numbers, and generates a replacement candidate node set. The task flow adjustment submodule calls the set of replacement candidate nodes, and according to the connection path structure of the original blocking node, redirects the corresponding task to the replacement node, restructures the task scheduling relationship, updates the node connection and task allocation path, and generates computing resource management results. The repeated number refers to the same node number that appears multiple times in the link transmission blocking list and corresponds to multiple blocking records; The node with the highest ratio is designated as the replacement node among the unassigned nodes, and its node number is identified.

2. The intelligent network computing resource management system according to claim 1, characterized in that: The node response load management module includes: The response fluctuation extraction submodule acquires node response fluctuation records, extracts continuous response directions, determines whether adjacent response time directions are consistent, records continuous direction sequences, and generates node direction change trend sequences. The reversal recognition submodule calls the node direction change trend sequence, compares whether adjacent directions are opposite, counts the number of direction reversals, and generates a direction reversal intensity result by combining the node's computing power usage. Based on the direction reversal strength result, the stable state generation submodule obtains the cache read / write backlog and input / output processing volume, calculates the cache access and processing ratio, calls the reversal strength comparison frequency and ratio relationship, and generates a node computing power stable state table. The node computing power usage refers to the total amount of computing resources consumed by a node in actually executing tasks within a time period.

3. The intelligent network computing resource management system according to claim 1, characterized in that: The task computing power level classification module includes: The task parameter extraction submodule obtains the node computing power stability status table, extracts the logical call depth, total number of instruction calls and cache interaction times corresponding to each task, integrates and records the three data items according to the task dimension, establishes a task execution structure item set, and generates a task execution indicator set. The intensity difference calculation submodule calls the task execution index set, calculates the difference between the logic call depth and the number of cache interactions in each record, and generates a task intensity level comparison table based on the difference matching logic and the cache difference range. The grade list generation submodule filters and categorizes the execution intensity of differentiated grade tasks based on the grade labels in the task intensity grade comparison table and the total number of instruction calls, constructs a computing power distribution record under the grade tasks, and generates a task computing power usage hierarchy list. The set of task execution structure items refers to the integrated set of three core parameters corresponding to each task: logical call depth, total number of instruction calls, and number of cache interactions. The difference matching logic refers to the matching rules used to map the difference between the logic call depth and the number of cache interactions to a preset task intensity level; The cache difference range refers to the segmentation of intensity levels corresponding to the logical depth and cache interaction difference preset according to the task type; The execution intensity refers to the degree to which a task consumes the computing resources of a node during its execution.

4. The intelligent network computing resource management system according to claim 1, characterized in that: The computing power scheduling and adaptation module includes: The node status extraction submodule obtains the task computing power usage hierarchy list, extracts the resource status of online nodes, reads the computing power occupancy value of the nodes, calculates the computing power occupancy distribution index, and structurally integrates the associated indexes with the node identifiers to generate an online resource occupancy index set. The resource matching and judgment submodule calls the online resource occupancy index set, compares the current computing power occupancy value of the node with the task intensity value according to the computing power intensity level record in the task list, and generates a task-adapted node list. The computing power node filtering submodule calls the task-matching node list, determines whether the cache occupancy ratio of the corresponding node is saturated, removes nodes whose cache occupancy is in the saturation range, retains the remaining nodes as schedulable targets, and generates a set of available computing power nodes for the task. The computing power occupancy value refers to the quantity and proportion of computing power resources occupied by the node for the tasks being executed. The nodes whose cache usage is in the saturation range refer to nodes whose cache resource utilization has reached or exceeded the preset upper limit threshold, resulting in a performance bottleneck.

5. The intelligent network computing resource management system according to claim 4, characterized in that: The specific formula for calculating the computing power occupancy distribution index is as follows: ； in, This represents the computing power distribution index of the k-th node. This represents the computing power usage of the k-th node on the m-th task. This represents the average computing power usage of all tasks on the k-th node. This represents the total cache usage of the k-th node. This represents the total number of tasks currently being processed by the k-th node. This represents a small constant that prevents division by zero.

6. The intelligent network computing resource management system according to claim 1, characterized in that: The specific formula for calculating the queuing percentage index is as follows: ； in, Representing the Each connection node in time Queue percentage at any given time. Representing the Each connection node in time The amount of tasks queued at any given time. Representing time Time of the first The weight of the tasks to be added to the team Representing time Time of the first The estimated processing time for each pending task to be added to the team. Representing the Each connection node in time Maximum task capacity at any given time. Indicates time The total number of all pending tasks associated with a given moment.

7. The intelligent network computing resource management system according to claim 1, characterized in that: The specific formula for calculating the computing power difference index of the unassigned nodes is as follows: ； in, Representative number is The computing power difference index of unassigned nodes, Representative number is The unassigned nodes in the first Idle computing power value per dimension Representative number is The number of tasks that unassigned nodes can handle. This represents the number of idle computing power dimensions involved in the computation. Representative number is The scheduling delay weight value corresponding to the unassigned node. It is a positive offset factor.

Images