Calculation power resource scheduling method and device, electronic equipment and storage medium

Abstract

The invention provides a computing power resource scheduling method and device, electronic equipment and a storage medium, and the method comprises the steps: responding to a business processing request of a business side, splitting a target business into a plurality of subtasks, and determining the task demand information of each subtask; the task demand information is used for indicating the index value of each sub-task for each task index; determining performance evaluation information of the alternative computing power nodes; the performance evaluation information is used for indicating an index value of a computing power evaluation index of the alternative computing power node; based on the task demand information and the performance evaluation information, determining the matching degree of the alternative computing power node and the subtask; and based on the matching degree, determining a target computing power node matched with the subtask in the alternative computing power nodes.

Description

Technical Field

[0001] The present disclosure relates to the technical field of computing resource scheduling, and in particular, to a computing resource scheduling method, device, electronic device and storage medium. Background Art

[0002] In the computing power network, heterogeneous computing power is ubiquitously distributed in different locations. Therefore, the core function of the computing power network is computing power resource matching and scheduling.

[0003] In the relevant technical scheme, by querying the distribution of computing power resources in the computing power network, the computing power resources and network resources that meet the customer's computing needs are found according to the distribution, and the computing power resource information and routing information that meet the customer's computing power needs are generated. The computing power resource information is notified to the business side, and the computing tasks are distributed by the business side.

[0004] The relevant technical solutions mainly select matching computing resources based on the computing resources requested by the business side. However, in actual business application scenarios, a completed business is completed by multiple different tasks. Since different tasks have different requirements for computing resources, the computing resources matched for the business in the relevant technical solutions may not necessarily meet the requirements of each task. Summary of the invention

[0005] The embodiments of the present disclosure at least provide a method, device, electronic device and storage medium for scheduling computing resources.

[0006] In a first aspect, an embodiment of the present disclosure provides a method for scheduling computing resources, including:

[0007] In response to a business processing request from the business side, split the target business into multiple subtasks, and determine task requirement information of each subtask; the task requirement information is used to indicate the indicator value of each subtask for each task indicator;

[0008] Determine performance evaluation information of the candidate computing power node; the performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node;

[0009] Determine the matching degree between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information;

[0010] Based on the matching degree, a target computing power node matching the subtask is determined among the candidate computing power nodes.

[0011] In an optional implementation manner, the determining the performance evaluation information of the candidate computing power node includes:

[0012] Obtain computing power node information of the candidate computing power node;

[0013] Determine the index value of each computing power evaluation index of the candidate computing power node based on the computing power node information;

[0014] The performance evaluation information is obtained after normalizing the index values ​​of each computing power evaluation index of the candidate computing power node.

[0015] In an optional implementation manner, determining the index value of each computing power evaluation index of the candidate computing power node based on the computing power node information includes:

[0016] Call indicator calculation script;

[0017] The computing power node information is calculated through the indicator calculation script, and the indicator value of the candidate computing power node for each computing power evaluation indicator is obtained after the calculation.

[0018] In an optional implementation, determining the degree of matching between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information includes:

[0019] Determine, based on the task requirement information, a weight value of each computing power evaluation indicator in the performance evaluation information relative to each of the subtasks;

[0020] Based on the weight value, the matching degree between the candidate computing power node and each of the subtasks is determined.

[0021] In an optional implementation, determining the weight value of each computing power evaluation indicator in the performance evaluation information relative to each subtask based on the task requirement information includes:

[0022] Determine the standard deviation of each computing power evaluation indicator of the candidate computing power node;

[0023] Based on the performance evaluation information and the task requirement information, determining the degree of correlation between each computing power evaluation indicator and each task indicator;

[0024] Based on the indicator correlation degree and the standard deviation, a weight value of each computing power evaluation indicator relative to the subtask is determined.

[0025] In an optional implementation, determining the indicator correlation degree between each computing power evaluation indicator and each task indicator based on the performance evaluation information and the task requirement information includes:

[0026] Determine the indicator correlation coefficient between each of the computing power evaluation indicators and each of the task indicators;

[0027] Based on the indicator correlation coefficient, determine the degree of indicator conflict between each of the computing power evaluation indicators and each of the task indicators;

[0028] The index conflict degree between the computing power evaluation index and each of the task indicators is summed to obtain the index correlation degree.

[0029] In an optional implementation, determining the weight value of each computing power evaluation indicator relative to the subtask based on the indicator correlation degree and the standard deviation includes:

[0030] Determining the information carrying capacity of each of the computing power evaluation indicators based on the indicator correlation degree and the standard deviation;

[0031] Based on the information carrying capacity, the weight value of each computing power evaluation indicator relative to the subtask is calculated.

[0032] In an optional implementation, determining a target computing node matching the subtask among the candidate computing nodes based on the matching degree includes:

[0033] Based on the matching degree, determining a candidate computing power node that matches the task requirement information of the subtask;

[0034] Determine the routing information of the matching candidate computing power node;

[0035] Based on the routing information of the matched candidate computing power nodes, the target computing power node is determined.

[0036] In an optional implementation, determining, based on the matching degree, a candidate computing power node that matches the task requirement information of the subtask includes:

[0037] Determining task execution information for each of the subtasks;

[0038] Sorting the candidate computing power nodes according to the degree of association between the candidate computing power nodes and the task execution information to obtain a computing power sorting result;

[0039] Based on the matching degree, an alternative computing power node matching the task requirement information of the subtask is determined in the computing power sorting result.

[0040] In a second aspect, the present disclosure also provides a computing resource scheduling device, including:

[0041] A first determining unit is used to respond to a business processing request from a business side, split the target business into multiple subtasks, and determine task requirement information of each of the subtasks; the task requirement information is used to indicate requirement information of each of the subtasks for each task indicator;

[0042] A second determining unit is used to determine performance evaluation information of the candidate computing power node; the performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node;

[0043] A third determining unit, configured to determine a matching degree between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information;

[0044] The fourth determining unit is used to determine a target computing power node that matches the subtask among the candidate computing power nodes based on the matching degree.

[0045] In a third aspect, an embodiment of the present disclosure further provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are performed.

[0046] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are executed.

[0047] In an embodiment of the present application, first, in response to a business processing request from the business side, the target business is split into multiple subtasks, and the task requirement information of each subtask is determined, through which the indicator value of each subtask for each task indicator can be indicated; then, the performance evaluation information of the alternative computing power node can be determined, through which the indicator value of the computing power evaluation indicator of the alternative computing power node can be indicated; next, based on the task requirement information and the performance evaluation information, the degree of matching between the alternative computing power node and each subtask can be determined, thereby selecting a target computing power node that matches the subtask among the alternative computing power nodes based on the degree of matching.

[0048] In the above implementation, by splitting the target business into multiple subtasks and determining the task requirement information of each subtask, it is possible to achieve refined processing of computing power requirements; by determining the degree of matching between the alternative computing power nodes and each subtask, and then determining the target computing power node for each subtask based on the matching degree, it is possible to achieve more refined matching of computing power nodes, thereby improving the matching accuracy and efficiency of computing power nodes.

[0049] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without creative work.

[0051] Figure 1 A flowchart of a method for scheduling computing resources provided by an embodiment of the present disclosure is shown;

[0052] Figure 2 A flowchart showing a specific method for determining performance evaluation information of candidate computing power nodes in the computing power resource scheduling method provided in an embodiment of the present disclosure;

[0053] Figure 3 A flowchart showing a specific method for determining the degree of matching between candidate computing nodes and subtasks based on task requirement information and performance evaluation information in the computing resource scheduling method provided by an embodiment of the present disclosure;

[0054] Figure 4 A schematic diagram of a computing resource scheduling device provided by an embodiment of the present disclosure is shown;

[0055] Figure 5 A schematic diagram of an electronic device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0057] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0058] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0059] In the computing power network, heterogeneous computing power is ubiquitously distributed in different locations. Therefore, the core function of the computing power network is computing power resource matching and scheduling.

[0060] In the relevant technical scheme, by querying the distribution of computing power resources in the computing power network, the computing power resources and network resources that meet the customer's computing needs are found according to the distribution, and the computing power resource information and routing information that meet the customer's computing power needs are generated. The computing power resource information is notified to the business side, and the computing tasks are distributed by the business side.

[0061] Video application fields represented by ultra-high-definition video, industrial vision, AR / VR, etc. have high computing power density requirements and large computing power demands. The current computing power scheduling system mainly selects matching target computing power resources based on the computing power resources requested by the business side, and schedules different business sessions to the edge node where the computing power resources are located for business processing. However, in actual business application scenarios, different subtasks or services have different requirements for computing power resources. Therefore, resource scheduling based on business sessions may not necessarily meet the computing power requirements of different subtasks or services. Therefore, how to select matching computing power resources for the business has become a technical problem that needs to be solved urgently.

[0062] Based on the above research, the present disclosure provides a scheduling method, device, electronic device and storage medium for computing power resources. In an embodiment of the present application, first, in response to a business processing request from a business side, the target business is split into multiple subtasks, and the task requirement information of each subtask is determined, through which the index value of each subtask for each task index can be indicated; then, the performance evaluation information of the candidate computing power node can be determined, through which the index value of the computing power evaluation index of the candidate computing power node can be indicated; next, based on the task requirement information and the performance evaluation information, the degree of matching between the candidate computing power node and each subtask can be determined, thereby determining the target computing power node that matches the subtask in the candidate computing power node based on the matching degree.

[0063] In the above implementation, by splitting the target business into multiple subtasks and determining the task requirement information of each subtask, it is possible to achieve refined processing of computing power requirements; by determining the degree of matching between the alternative computing power nodes and each subtask, and then determining the target computing power node for each subtask based on the matching degree, it is possible to achieve more refined matching of computing power nodes, thereby improving the matching accuracy and efficiency of computing power nodes.

[0064] To facilitate understanding of this embodiment, a method for scheduling computing resources disclosed in an embodiment of the present disclosure is first introduced in detail. The execution subject of the method for scheduling computing resources provided in an embodiment of the present disclosure is generally a computer device with certain computing capabilities, and the computer device includes, for example, a terminal device or a server or other processing device. In some possible implementations, the method for scheduling computing resources can be implemented by a processor calling a computer-readable instruction stored in a memory.

[0065] See also Figure 1 FIG. 1 is a flowchart of a method for scheduling computing resources provided by an embodiment of the present disclosure, the method comprising steps S101 to S104, wherein:

[0066] S101: In response to a business processing request from a business side, split a target business into multiple subtasks, and determine task requirement information of each subtask; the task requirement information is used to indicate requirement information of each subtask for each task indicator.

[0067] Here, the business processing request may be a business for processing video streams or other types of data, and this application does not make any specific limitations on this; wherein the business processing request carries the task name of the target business. At this time, the target business can be split into multiple subtasks based on the task name in the business processing request.

[0068] Here, the task requirement information includes: computing power requirement information, network requirement information, and attribute information of each subtask.

[0069] The attribute information of the subtask includes the task type, task priority, type of computing node required for the task, etc. The computing power requirement information includes the computing power required for the subtask, and the network requirement information includes the bandwidth, latency, data throughput, etc. required for the subtask.

[0070] In this case, the task indicator may include at least one of the following: computing power, bandwidth, latency, and data throughput.

[0071] S102: Determine performance evaluation information of the candidate computing power node; the performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node.

[0072] Here, the computing power node information of each candidate computing power node can be determined in the computing power network, and the performance evaluation information of each candidate computing power node can be determined based on the computing power node information. Through the performance evaluation index, the index value of each computing power evaluation index of the candidate computing power node can be determined. For example, the computing power evaluation index includes but is not limited to the following: effective computing power evaluation index, communication capacity evaluation index such as network bandwidth, memory capacity index evaluation and storage capacity evaluation index.

[0073] S103: Based on the task requirement information and the performance evaluation information, determine the degree of matching between the candidate computing power node and the subtask.

[0074] After obtaining the task requirement information and performance evaluation information, the matching degree between the candidate computing power node and each subtask can be determined. The matching degree can be used to indicate the matching degree between each computing power evaluation index of the candidate computing power node and each subtask.

[0075] S104: Based on the matching degree, determine a target computing node among the candidate computing nodes that matches the subtask.

[0076] After determining the degree of matching, the candidate computing power node that matches each subtask among the candidate computing power nodes can be determined as the target computing power node based on the degree of matching between each computing power evaluation indicator of the candidate computing power node and each subtask.

[0077] In an embodiment of the present application, first, in response to a business processing request from the business side, the target business is split into multiple subtasks, and the task requirement information of each subtask is determined, through which the indicator value of each subtask for each task indicator can be indicated; then, the performance evaluation information of the alternative computing power node can be determined, through which the indicator value of the computing power evaluation indicator of the alternative computing power node can be indicated; next, based on the task requirement information and the performance evaluation information, the degree of matching between the alternative computing power node and each subtask can be determined, thereby selecting a target computing power node that matches the subtask among the alternative computing power nodes based on the degree of matching.

[0078] In the above implementation, by splitting the target business into multiple subtasks and determining the task requirement information of each subtask, it is possible to achieve refined processing of computing power requirements; by determining the degree of matching between the alternative computing power nodes and each subtask, and then determining the target computing power node for each subtask based on the matching degree, it is possible to achieve more refined matching of computing power nodes, thereby improving the matching accuracy and efficiency of computing power nodes.

[0079] The above steps will be described in detail below in conjunction with specific implementation methods.

[0080] It can be seen from the above description that in the embodiment of the present application, first, the business processing request initiated by the business side to the business platform is detected, and the target business is split into multiple subtasks, and the task requirement information of each subtask is determined.

[0081] It is assumed that the service processing request is a service of processing a video stream sent by a terminal device.

[0082] On this basis, after receiving the video stream sent by the terminal device, the business platform can determine the processor for frame extraction and decoding of the video stream according to the video encoding format of the video stream, such as whether the processor is a GPU or a CPU; then, the video stream is subjected to video frame extraction processing at a preset time interval. Next, the business platform obtains the task name and business processing request of the video processing task sent by the business side, and then splits the target business according to the business processing request to obtain multiple subtasks, and sets a corresponding task ID for each subtask, as well as determines the computing power requirement information and network requirement information for each subtask. Among them, the task requirement information includes: task ID, computing power requirement information and network requirement information.

[0083] Assume that after the video stream is processed by video frame extraction, m images are obtained. Assume that one video stream processes n subtasks in parallel, then the computing power requirement of the target business can be obtained as n channels, n×m images concurrently.

[0084] Furthermore, the task ID of each subtask can be set according to the rules. The task ID of each subtask can include 64 bits of data, which is globally unique and has a time trend increment, including 1 flag bit, 16 timestamp, 20 device identification, 5 task type identification, 5 priority identification, 5 computing power type identification, and 12 serial number identification.

[0085] Here, the flag bit is used to indicate the corresponding subtask, thereby distinguishing multiple subtasks. The timestamp is used to indicate the initiation time of the service processing request. The device identifier is used to indicate the device identifier of the initiator (ie, the service side) of the service processing request.

[0086] The task type identifier is used to indicate the task type of the subtask. For example, the task types include: low-latency video tasks, high-computing training video tasks, and high-interaction video tasks. For example, the task type identifier of a low-latency video task (e.g., an inference video task) can be set to 001, the task type identifier of a high-computing training video task (e.g., a training / rendering video task) can be set to 010, and the task type identifier of a high-interaction video task (e.g., a live video task) can be set to 100.

[0087] The priority flag is used to indicate the task priority of the subtask, for example, the task priority includes ordinary task, priority task, quick task, flash task, rapid task, etc. For example, the priority flag of ordinary task can be set to 000, the priority flag of priority task can be set to 001, the priority flag of quick task can be set to 010, the priority flag of flash task can be set to 011, and the priority flag of rapid task can be set to 100.

[0088] The computing power type identifier is used to indicate the computing power type information of the computing power node required for the subtask. For example, the computing power type information may include: small computing power, medium computing power, large computing power, and ultra-large computing power. For example, the computing power type identifier of small computing power (for example, small computing application scenarios) can be set to 001, the computing power type identifier of medium computing power (for example, reasoning application scenarios) can be set to 010, the computing power type identifier of large computing power (for example, most model training scenarios) can be set to 011, and the computing power type identifier of ultra-large computing power (for example, rendering scenarios) can be set to 100.

[0089] Here, assuming that there are n subtasks, the corresponding computing power demand information and network demand information can be defined for the n tasks, that is, the computing power and network demand matrix D (n×q-dimensional demand matrix) described below, where q is the number of indicators of the task indicator. For example, the task indicator may include: computing power size, bandwidth, latency, data throughput, at this time, the demand matrix D can be an n×4-dimensional demand matrix, for example, the demand matrix described by the following formula.

[0090]

[0091] In the above-mentioned demand matrix D, each row of data represents the computing power demand information and network demand information of each subtask, and each column represents a task indicator.

[0092] After determining the task requirement information of each subtask, the performance evaluation information of the candidate computing power nodes can be determined.

[0093] In an optional embodiment, if Figure 2 As shown, determining the performance evaluation information of the candidate computing power nodes specifically includes the following steps:

[0094] Step S11: Obtain computing power node information of the candidate computing power node;

[0095] Step S12: determining the index value of each computing power evaluation index of the candidate computing power node based on the computing power node information;

[0096] Step S13: normalizing the index values ​​of each computing power evaluation index of the candidate computing power node to obtain the performance evaluation information.

[0097] In the embodiment of the present application, it is assumed that the number of candidate computing power nodes is n. At this time, n computing power node information can be obtained, wherein each candidate computing power node corresponds to one computing power node information.

[0098] After obtaining the information of n computing power nodes, the index value of each candidate computing power node for each computing power rating index can be determined based on the information of n computing power nodes. Among them, the computing power evaluation index includes but is not limited to the following: effective computing power evaluation index, communication capacity evaluation index such as network bandwidth, memory capacity evaluation index and storage capacity evaluation index.

[0099] Here, the effective computing power of the candidate computing power node can be calculated using the following calculation formula (1):

[0100]

[0101] Among them, A c It is expressed as the index value of the effective computing power evaluation index, where the effective computing power is the computing power size of each candidate computing power node obtained by evaluating each candidate computing power node using mainstream benchmark test sets such as LINPACK, NPB, and IOzone; P si is the actual performance of the specified test set i on the benchmark computing node; P i is the actual performance of the specified test set i on each candidate computing power node. i is the weight of different benchmarks or actual application software in test set i. α is the adjustment coefficient, and the recommended constant value is 100.

[0102] Here, the method described in the above formula can be used to determine the index values ​​of communication capability evaluation indicators such as network bandwidth, memory capability evaluation indicators and storage capability evaluation indicators, which are denoted as A and d , A m and A s At this point, based on the index values ​​of the computing power evaluation index, the following evaluation index matrix P can be obtained:

[0103] Among them, k in the evaluation index matrix P represents the number of alternative computing power nodes.

[0104] Among them, the general formula of the evaluation index matrix P can be expressed as:

[0105] In the evaluation index matrix P, k represents the number of candidate computing power nodes, and w represents the number or type of computing power evaluation indicators.

[0106] Among them, each row in the evaluation index matrix P represents the index value of each candidate computing power node for each computing power evaluation index, and each column in the evaluation index matrix P represents each computing power evaluation index.

[0107] In order to eliminate the influence of dimension, the index values ​​in the evaluation index matrix P are normalized. First, the maximum and minimum values ​​in the matrix P are taken:

[0108] max P i =max{A i1 A i2 … A in … A iw};

[0109] min P i =min{A i1 A i2 … A in … A iw};

[0110] Among them, maxP i represents the maximum value of the i-th row in the evaluation index matrix P, minP i represents the minimum value of the i-th row in the evaluation index matrix P, A in It is used to indicate the index value of the i-th row and n-th column of the evaluation index matrix P, that is, the index value of the n-th computing power evaluation index of the i-th candidate computing power node.

[0111] Next, each indicator value A can be calculated in The normalized value A in-less :

[0112] In this normalized calculation formula, n represents n=1, 2, ..., m.

[0113] In an optional implementation, the above steps determine the index value of each computing power evaluation index of the candidate computing power node based on the computing power node information, specifically including the following steps:

[0114] First, call the indicator operation script;

[0115] Secondly, the computing power node information is calculated through the indicator calculation script, and the indicator value of the candidate computing power node for each computing power evaluation indicator is obtained after the calculation.

[0116] In an embodiment of the present application, after obtaining information of n computing power nodes, an indicator calculation script can be called, and the computing power node information can be calculated through the indicator calculation script to obtain the indicator value of each computing power evaluation indicator.

[0117] From the above description, it can be seen that the computing power evaluation indicators include but are not limited to the following: effective computing power evaluation indicators, communication capability evaluation indicators such as network bandwidth, memory capability evaluation indicators and storage capability evaluation indicators.

[0118] Here, for each computing power evaluation indicator, a corresponding indicator operation script can be set. By running the indicator operation script of each computing power evaluation indicator, the indicator value of the corresponding computing power evaluation indicator can be obtained.

[0119] After the performance evaluation information of the candidate computing power node is determined, the matching degree between the candidate computing power node and the subtask can be determined based on the task requirement information and the performance evaluation information.

[0120] In an optional embodiment, if Figure 3 As shown, based on the task requirement information and the performance evaluation information, determining the matching degree between the candidate computing power node and the subtask specifically includes the following steps:

[0121] Step S21: determining the weight value of each computing power evaluation indicator in the performance evaluation information relative to each subtask based on the task requirement information;

[0122] Step S22: Based on the weight value, determine the matching degree between the candidate computing power node and each of the subtasks.

[0123] In an embodiment of the present application, the weight values ​​of each computing power evaluation indicator of each candidate computing power node relative to each subtask can be determined based on the computing power requirement information and network requirement information (i.e., the requirement matrix D) in the task requirement information; wherein, the larger the weight value, the greater the effect of the computing power evaluation indicator on the subtask; conversely, the smaller the weight value, the smaller the effect of the computing power evaluation indicator on the subtask.

[0124] The weight value can be used to determine the matching degree of each candidate computing power node relative to each subtask. The higher the matching degree, the greater the possibility that the candidate computing power node is determined as the target computing power node of the subtask.

[0125] Here, for the index value of each computing power evaluation index of each candidate computing power node, the matching degree between each computing power evaluation index of each candidate computing power node and the subtask can be determined. Then, a matching target computing power node can be determined for the subtask according to the matching degree.

[0126] In the above implementation, by determining the degree of matching between the candidate computing power nodes and each subtask, and then determining the target computing power node for each subtask based on the degree of matching, a more refined matching of the computing power nodes can be achieved, thereby improving the matching accuracy and efficiency of the computing power nodes.

[0127] In an optional implementation, the above step determines the weight value of each computing power evaluation indicator in the performance evaluation information relative to each subtask based on the task requirement information, specifically including the following steps:

[0128] Step S211: determining the standard deviation of each computing power evaluation index of the candidate computing power node;

[0129] Step S212: determining the indicator correlation degree between each computing power evaluation indicator and each task indicator based on the performance evaluation information and the task requirement information;

[0130] Step S213: Based on the indicator correlation degree and the standard deviation, determine the weight value of each computing power evaluation indicator relative to the subtask.

[0131] In the embodiment of the present application, the standard deviation of each computing power evaluation index of the candidate computing power node can be calculated. For example, for the effective computing power evaluation index, communication capability evaluation index such as network bandwidth, memory capability evaluation index and storage capability evaluation index, the standard deviation of each computing power evaluation index can be calculated.

[0132] Here, the standard deviation of each computing power evaluation indicator can be calculated by the following formula:

[0133] in, represents the mean value of the data in the jth column of the indicator matrix P. That is, the mean value of the indicator values ​​of each candidate computing power node for the same computing power evaluation indicator is calculated to obtain the mean value of each column of data, S j represents the standard deviation of the data in the jth column, that is, the standard deviation of the computing power evaluation index corresponding to the data in the jth column in the index matrix, x ij It represents the index value of the jth computing power evaluation index of the i-th candidate computing power node in the index matrix. In the above formula, k represents the number of candidate computing power nodes.

[0134] After determining the standard deviation of each computing power evaluation indicator, it is also necessary to determine the degree of correlation between each computing power evaluation indicator and each task indicator. Among them, through the degree of correlation, the degree of conflict between each computing power evaluation indicator and the task indicator can be determined.

[0135] In an optional implementation, determining the indicator correlation degree between each computing power evaluation indicator and each task indicator based on the performance evaluation information and the task requirement information specifically includes the following steps:

[0136] First, determine the indicator correlation coefficient between each of the computing power evaluation indicators and each of the task indicators;

[0137] Secondly, based on the indicator correlation coefficient, determine the indicator conflict degree between each computing power evaluation indicator and each task indicator;

[0138] Finally, the index conflict degree between the computing power evaluation index and each of the task indicators is summed up to obtain the index correlation degree.

[0139] In the embodiment of the present application, the correlation between the computing power evaluation index and the task index can be determined by a correlation algorithm, thereby obtaining the index correlation coefficient r between the computing power evaluation index and each task index. ij .

[0140] Here, we can use the indicator correlation coefficient r ij , determine the degree of conflict between each computing power evaluation indicator and each task indicator. For example, the formula 1-|r ij | Calculate the index conflict degree; then, sum the index conflict degrees between the computing power evaluation index and each of the task indicators to obtain the index correlation degree. The specific formula is as follows:

[0141] Among them, A j Indicates the correlation degree of the indicator, r ij Represents the correlation coefficient between the i-th task indicator and the j-th computing power evaluation indicator.

[0142] After determining the indicator correlation degree, the weight value of each computing power evaluation indicator relative to the subtask can be determined based on the indicator correlation degree and the standard deviation, which specifically includes the following steps:

[0143] First, based on the correlation degree of the indicators and the standard deviation, the information carrying capacity of each computing power evaluation indicator is determined;

[0144] Here, the formula Calculate the information carrying capacity of each computing power evaluation indicator. j The information carrying capacity can represent the role of the computing power evaluation index. The larger the information carrying capacity, the greater the role of the jth computing power evaluation index in the entire evaluation index system, and the more weight value should be allocated.

[0145] Secondly, based on the information carrying capacity, the weight value of each computing power evaluation indicator relative to the subtask is calculated.

[0146] Here, the formula The information carrying capacity is processed to obtain the weight value of each computing power evaluation indicator relative to the subtask.

[0147] In an optional implementation, the determining of the target computing node matching the subtask among the candidate computing nodes based on the matching degree specifically includes the following steps:

[0148] Step S31: Based on the matching degree, determining a candidate computing power node that matches the task requirement information of the subtask;

[0149] Step S32: Determine the routing information of the matching candidate computing power node;

[0150] Step S33: Determine the target computing power node based on the routing information of the matching candidate computing power node.

[0151] In an embodiment of the present application, first, the task execution information of each of the subtasks can be determined; for example, the task type, task priority, type of computing power node required for the task, and other information. Afterwards, the candidate computing power nodes can be sorted according to the degree of association between the candidate computing power nodes and the task execution information to obtain a computing power sorting result; then, based on the degree of matching, the candidate computing power nodes that match the task requirement information of the subtask are determined in the computing power sorting result. Here, the candidate computing power nodes that match the task requirement information of the subtask can be determined according to the computing power sorting result.

[0152] From the above description, it can be seen that the technical solution of this application decomposes the target business into subtasks, making the computing power demand more refined. By determining the weight value of the computing power evaluation index, the computing power evaluation index can be objectively evaluated, and resources can be allocated according to the weight value, so that a more refined matching of computing power nodes can be achieved, thereby achieving effective decentralization of computing power demand; by sorting computing power nodes according to task type, computing power demand, and service priority, a reasonable allocation of computing power demand and algorithm deployment requests is achieved, and the algorithm is deployed on the appropriate computing power node, which improves the matching accuracy and efficiency.

[0153] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0154] Based on the same inventive concept, the embodiments of the present disclosure also provide a scheduling device for computing power resources corresponding to the scheduling method for computing power resources. Since the principle of solving the problem by the device in the embodiments of the present disclosure is similar to the scheduling method for computing power resources in the embodiments of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0155] Reference Figure 4, which is a schematic diagram of a computing resource scheduling device provided by an embodiment of the present disclosure, the device includes: a first determining unit 10, a second determining unit 20, a third determining unit 30 and a fourth determining unit 40; wherein,

[0156] The first determining unit 10 is used to respond to the business processing request of the business side, split the target business into multiple subtasks, and determine the task requirement information of each subtask; the task requirement information is used to indicate the requirement information of each subtask for each task indicator;

[0157] The second determining unit 20 is used to determine the performance evaluation information of the candidate computing power node; the performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node;

[0158] A third determining unit 30 is used to determine the matching degree between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information;

[0159] The fourth determining unit 40 is used to determine a target computing power node that matches the subtask among the candidate computing power nodes based on the matching degree.

[0160] In the above implementation, by splitting the target business into multiple subtasks and determining the task requirement information of each subtask, it is possible to achieve refined processing of computing power requirements; by determining the degree of matching between the alternative computing power nodes and each subtask, and then determining the target computing power node for each subtask based on the matching degree, it is possible to achieve more refined matching of computing power nodes, thereby improving the matching accuracy and efficiency of computing power nodes.

[0161] In a possible implementation manner, the second determining unit is further configured to:

[0162] Obtain computing power node information of the candidate computing power node;

[0163] Determine the index value of each computing power evaluation index of the candidate computing power node based on the computing power node information;

[0164] The performance evaluation information is obtained after normalizing the index values ​​of each computing power evaluation index of the candidate computing power node.

[0165] In a possible implementation manner, the second determining unit is further configured to:

[0166] Call indicator calculation script;

[0167] The computing power node information is calculated through the indicator calculation script, and the indicator value of the candidate computing power node for each computing power evaluation indicator is obtained after the calculation.

[0168] In a possible implementation manner, the third determining unit is further configured to:

[0169] Determine, based on the task requirement information, a weight value of each computing power evaluation indicator in the performance evaluation information relative to each of the subtasks;

[0170] Based on the weight value, the matching degree between the candidate computing power node and each of the subtasks is determined.

[0171] In a possible implementation manner, the third determining unit is further configured to:

[0172] Determine the standard deviation of each computing power evaluation indicator of the candidate computing power node;

[0173] Based on the performance evaluation information and the task requirement information, determining the degree of correlation between each computing power evaluation indicator and each task indicator;

[0174] Based on the indicator correlation degree and the standard deviation, a weight value of each computing power evaluation indicator relative to the subtask is determined.

[0175] In a possible implementation manner, the third determining unit is further configured to:

[0176] Determine the indicator correlation coefficient between each of the computing power evaluation indicators and each of the task indicators;

[0177] Based on the indicator correlation coefficient, determine the degree of indicator conflict between each of the computing power evaluation indicators and each of the task indicators;

[0178] The index conflict degree between the computing power evaluation index and each of the task indicators is summed to obtain the index correlation degree.

[0179] In a possible implementation manner, the third determining unit is further configured to:

[0180] Determining the information carrying capacity of each of the computing power evaluation indicators based on the indicator correlation degree and the standard deviation;

[0181] Based on the information carrying capacity, the weight value of each computing power evaluation indicator relative to the subtask is calculated.

[0182] In a possible implementation manner, the fourth determining unit is further configured to:

[0183] Based on the matching degree, determining a candidate computing power node that matches the task requirement information of the subtask;

[0184] Determine the routing information of the matching candidate computing power node;

[0185] Based on the routing information of the matched candidate computing power nodes, the target computing power node is determined.

[0186] In a possible implementation manner, the fourth determining unit is further configured to:

[0187] Determining task execution information for each of the subtasks;

[0188] Sorting the candidate computing power nodes according to the degree of association between the candidate computing power nodes and the task execution information to obtain a computing power sorting result;

[0189] Based on the matching degree, an alternative computing power node matching the task requirement information of the subtask is determined in the computing power sorting result.

[0190] For descriptions of the processing flow of each module in the device and the interaction flow between each module, reference may be made to the relevant descriptions in the above method embodiment, which will not be described in detail here.

[0191] Corresponds to Figure 1 The present disclosure also provides an electronic device 500, such as Figure 5 FIG. 5 is a schematic diagram of the structure of an electronic device 500 provided in an embodiment of the present disclosure, including:

[0192] Processor 51, memory 52, and bus 53; memory 52 is used to store execution instructions, including memory 521 and external memory 522; memory 521 here is also called internal memory, which is used to temporarily store operation data in processor 51 and data exchanged with external memory 522 such as hard disk. Processor 51 exchanges data with external memory 522 through memory 521. When the electronic device 500 is running, the processor 51 communicates with the memory 52 through bus 53, so that the processor 51 executes the following instructions:

[0193] In response to a business processing request from the business side, split the target business into multiple subtasks, and determine task requirement information of each subtask; the task requirement information is used to indicate the indicator value of each subtask for each task indicator;

[0194] Determine performance evaluation information of the candidate computing power node; the performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node;

[0195] Determine the matching degree between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information;

[0196] Based on the matching degree, a target computing power node matching the subtask is determined among the candidate computing power nodes.

[0197] The present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for scheduling computing resources described in the above method embodiment are executed. The storage medium may be a volatile or non-volatile computer-readable storage medium.

[0198] The present disclosure also provides a computer program product that carries a program code. The program code includes instructions that can be used to execute the steps of the method for scheduling computing resources described in the above method embodiment. For details, please refer to the above method embodiment, which will not be repeated here.

[0199] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0200] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0201] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0202] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0203] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0204] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for scheduling computing resources, characterized in that: include: In response to a business processing request from the business side, split the target business into multiple subtasks and determine task requirement information of each of the subtasks; The task requirement information is used to indicate the indicator value of each subtask for each task indicator; Determine the performance evaluation information of candidate computing nodes; The performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node; Determine the matching degree between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information; Based on the matching degree, a target computing power node matching the subtask is determined among the candidate computing power nodes.

2. The method according to claim 1, characterized in that The performance evaluation information of the candidate computing power nodes is determined, including: Obtain computing power node information of the candidate computing power node; Determine the index value of each computing power evaluation index of the candidate computing power node based on the computing power node information; The performance evaluation information is obtained after normalizing the index values ​​of each computing power evaluation index of the candidate computing power node.

3. The method according to claim 2, characterized in that The determining, based on the computing power node information, the index value of each computing power evaluation index of the candidate computing power node includes: Call indicator calculation script; The computing power node information is calculated through the indicator calculation script, and the indicator value of the candidate computing power node for each computing power evaluation indicator is obtained after the calculation.

4. The method according to claim 1, characterized in that: The determining, based on the task requirement information and the performance evaluation information, a degree of matching between the candidate computing power node and the subtask includes: Determine, based on the task requirement information, a weight value of each computing power evaluation indicator in the performance evaluation information relative to each of the subtasks; Based on the weight value, the matching degree between the candidate computing power node and each of the subtasks is determined.

5. The method according to claim 4, characterized in that The determining, based on the task requirement information, a weight value of each computing power evaluation indicator in the performance evaluation information relative to each subtask includes: Determine the standard deviation of each computing power evaluation indicator of the candidate computing power node; Based on the performance evaluation information and the task requirement information, determining the degree of correlation between each computing power evaluation indicator and each task indicator; Based on the indicator correlation degree and the standard deviation, a weight value of each computing power evaluation indicator relative to the subtask is determined.

6. The method according to claim 5, characterized in that The determining, based on the performance evaluation information and the task requirement information, the degree of correlation between each computing power evaluation indicator and each task indicator comprises: Determine the indicator correlation coefficient between each of the computing power evaluation indicators and each of the task indicators; Based on the indicator correlation coefficient, determine the degree of indicator conflict between each of the computing power evaluation indicators and each of the task indicators; The index conflict degree between the computing power evaluation index and each of the task indicators is summed to obtain the index correlation degree.

7. The method according to claim 5, characterized in that Determining the weight value of each computing power evaluation indicator relative to the subtask based on the indicator correlation degree and the standard deviation includes: Determining the information carrying capacity of each of the computing power evaluation indicators based on the indicator correlation degree and the standard deviation; Based on the information carrying capacity, the weight value of each computing power evaluation indicator relative to the subtask is calculated.

8. The method according to claim 1, characterized in that The determining, based on the matching degree, a target computing node among the candidate computing nodes that matches the subtask includes: Based on the matching degree, determining a candidate computing power node that matches the task requirement information of the subtask; Determine the routing information of the matching candidate computing power node; Based on the routing information of the matched candidate computing power nodes, the target computing power node is determined.

9. The method according to claim 8, characterized in that The determining, based on the matching degree, a candidate computing power node that matches the task requirement information of the subtask includes: Determining task execution information for each of the subtasks; Sorting the candidate computing power nodes according to the degree of association between the candidate computing power nodes and the task execution information to obtain a computing power sorting result; Based on the matching degree, an alternative computing power node matching the task requirement information of the subtask is determined in the computing power sorting result.

10. A computing resource scheduling device, characterized in that: include: A first determining unit, configured to respond to a business processing request from a business side, split a target business into a plurality of subtasks, and determine task requirement information of each of the subtasks; The task requirement information is used to indicate the requirement information of each subtask for each task indicator; A second determining unit, used to determine performance evaluation information of a candidate computing power node; The performance evaluation information is used to indicate the index value of the computing power evaluation index of the candidate computing power node; A third determining unit, configured to determine a matching degree between the candidate computing power node and the subtask based on the task requirement information and the performance evaluation information; The fourth determining unit is used to determine a target computing power node that matches the subtask among the candidate computing power nodes based on the matching degree.

11. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the method for scheduling computing resources as described in any one of claims 1 to 9 are performed.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of the method for scheduling computing resources as described in any one of claims 1 to 9.

Images