Computing Power Scheduling Method, Device, Electronic Device and Storage Medium

Abstract

The present invention relates to the field of cloud computing technology, and specifically relates to a computing power scheduling method, device, electronic device, and storage medium. The method includes obtaining candidate computing power nodes, sampling data of target computing power resource monitoring metrics of the candidate computing power nodes within a preset time interval, and time attributes of the sampling data, where the preset time interval is the time interval between the current moment and the starting historical moment; performing volatility analysis on the sampling data to determine the volatility attribute of the sampling data; analyzing the operating status of the candidate computing power nodes based on the time attributes and the volatility attribute to determine the analysis result of the candidate computing power nodes; and performing computing power scheduling on the candidate computing power nodes based on the analysis result. By comprehensively considering the time attributes and volatility attributes of the sampling data, it is possible to improve the reliability of computing power scheduling without changing the parameter configurations of the monitoring systems of each manufacturer.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud computing, and particularly to a computing power scheduling method, apparatus, electronic device, and storage medium. Background Art

[0002] In recent years, edge cloud, hybrid cloud, distributed cloud, etc. have gradually become hotspots in the field of cloud computing. The deep integration of various resources of cloud, edge, and terminal has put forward higher requirements for computing power scheduling, especially for computing power scheduling between heterogeneous multi-clouds of different manufacturers. Monitoring data of computing power resources is an important basis for making computing power scheduling decisions. Since different parties may use cloud resource monitoring platforms of different manufacturers or different architectures during construction, there may be differences in the sampling frequencies of the same type of monitoring items. For example, some statistics data every 1 minute, and some every 5 minutes.

[0003] Existing computing power scheduling technical solutions mainly use the most recent sampling data or the average value of sampling data within a recent time interval as a reference basis. When the sampling timestamps of the same type of monitoring items on different cloud platforms are out of sync and vary greatly, when making scheduling decisions, the obtained sampling data may not be able to truly represent the running state of computing power resources at the current moment, and the sampling data of the same type of monitoring items between different computing power nodes is not comparable in time, resulting in low reliability of computing power scheduling based on this method. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a computing power scheduling method, apparatus, electronic device, and storage medium to solve the problem of low reliability of computing power scheduling.

[0005] According to a first aspect, an embodiment of the present invention provides a computing power scheduling method, including:

[0006] Obtain candidate computing power nodes, sampling data of target computing power resource monitoring metrics of the candidate computing power nodes within a preset time interval, and time attributes of the sampling data, where the preset time interval is the time interval between the current moment and the starting historical moment;

[0007] Perform volatility analysis on the sampling data to determine the volatility attribute of the sampling data;

[0008] Analyze the running state of the candidate computing power nodes based on the time attributes and the volatility attributes to determine the analysis result of the candidate computing power nodes;

[0009] Perform computing power scheduling on the candidate computing power nodes based on the analysis result.

[0010] The computing power scheduling method provided by the embodiments of the present invention comprehensively considers the time attribute and the fluctuation attribute of the sampling data when analyzing the operating status of candidate computing power nodes. Without changing the parameter configurations of the monitoring systems of each manufacturer, it can evaluate the computing power resource status of each candidate computing power node, thereby helping the computing power scheduling system make more reliable scheduling decisions and improving the reliability of computing power scheduling.

[0011] In some embodiments, obtaining the time attribute of the sampling data includes:

[0012] For each of the candidate computing power nodes, obtain the sampling time of the sampling data;

[0013] Calculate the time attribute using the sampling time.

[0014] Since the sampling time of the computing power scheduling method provided by the embodiments of the present invention is determined by the parameter configurations of each monitoring system, using the sampling time to determine the time attribute can ensure the reliability of the time attribute and there is no need to unify the time attributes of each candidate computing power node during computing power scheduling.

[0015] In some embodiments, the calculating the time attribute using the sampling time includes:

[0016] Calculate the average sampling time interval of the sampling data using the sampling time;

[0017] Extract the sampling time of the most recent sampling data in the sampling time;

[0018] Calculate the time difference between the sampling time of the most recent sampling data and the current moment to determine the time attribute, where the time attribute includes the average sampling time interval and the time difference.

[0019] In the computing power scheduling method provided by the embodiments of the present invention, the closer the time of the latest sampling data is to the current moment, the higher the credibility of the sampling data representing the current operating status of the computing power resources; the shorter the sampling interval and the higher the sampling frequency of the data, the higher the credibility of the sampling data representing the current operating status of the computing power resources; the smaller the volatility of the sampling data within a period of time, the more stable the resource operating status, and the higher the credibility of the sampling data representing the current operating status of the computing power resources. Therefore, by comprehensively considering the time difference between the latest sampling and the current moment, the sampling interval of the data, and the volatility of the sampling data, the reliability of computing power resource scheduling can be ensured. This method is particularly applicable to the cross-cloud platform computing power scheduling scenario.

[0020] In some embodiments, the analyzing the operating status of the candidate computing power node based on the time attribute and the fluctuation attribute to determine the analysis result of the candidate computing power node includes:

[0021] For the candidate computing power nodes, calculate a first ratio of the time attribute and a second ratio of the fluctuation attribute. The first ratio is the ratio of the time attribute of the candidate computing power node among the time attributes of all the candidate computing power nodes, and the second ratio is the ratio of the fluctuation attribute of the candidate computing power node among the fluctuation attributes of all the candidate computing power nodes;

[0022] Obtain a first weight of the time attribute and a second weight of the fluctuation attribute;

[0023] Calculate a weighted sum of the first weight and the first ratio and the second weight and the second ratio;

[0024] Determine an analysis result of the candidate computing power node based on the weighted sum.

[0025] The computing power resource scheduling method provided by the embodiments of the present invention determines the analysis result of the candidate computing power node by using the ratios of the time attribute and the fluctuation attribute, and fuses the time attribute and the fluctuation attribute of each sampling data of the candidate computing power node, thereby improving the reliability of the analysis result of the candidate computing power node.

[0026] In some embodiments, the time attribute includes an average sampling time interval and a time difference, and the following formula is used to calculate the analysis result of the candidate computing power node:

[0027] s i =w span *span_p i +w d *d_p i +w σ *σ_p i

[0028] wherein, s i is the analysis result of the i-th candidate computing power node, w span is the first weight of the average sampling time interval, span_p i is the first ratio of the average sampling time interval, w d is the first weight of the time difference, d_p i is the first ratio of the time difference, w σ is the second weight of the fluctuation attribute, and σ_p i is the second ratio of the fluctuation attribute.

[0029] In some embodiments, when the target computing power resource monitoring metrics include at least two, the analyzing the operating state of the candidate computing power node based on the time attribute and the fluctuation attribute to determine the analysis result of the candidate computing power node includes:

[0030] For each of the target computing power resource monitoring metrics, analyze the operating status of the candidate computing nodes based on the time attribute and the volatility attribute, and determine the analysis result corresponding to the target computing power resource monitoring metric;

[0031] Obtain the weights of each of the target computing power resource monitoring metrics;

[0032] Based on the weights of each of the target computing power resource monitoring metrics and the analysis results corresponding to the target computing power resource monitoring metrics, determine the analysis result of the candidate computing nodes.

[0033] In the computing power scheduling method provided by the embodiments of the present invention, when the target computing power resource monitoring metrics include at least two, the analysis results of all the target computing power resource monitoring metrics are integrated through the weights of each target computing power resource monitoring metric, ensuring the accuracy of the analysis result of the obtained candidate computing nodes.

[0034] In some embodiments, the computing power scheduling of the candidate computing nodes based on the analysis results includes:

[0035] Compare the magnitudes of the analysis results, and determine the candidate computing node corresponding to the smallest analysis result as the target computing node;

[0036] Perform computing power scheduling based on the target computing node.

[0037] According to a second aspect, an embodiment of the present invention provides a computing power scheduling device, including:

[0038] An acquisition module, configured to acquire candidate computing nodes, sampling data of target computing power resource monitoring metrics of the candidate computing nodes within a preset time interval, and the time attribute of the sampling data, where the preset time interval is the time interval between the current moment and the starting historical moment;

[0039] A first analysis module, configured to perform volatility analysis on the sampling data to determine the volatility attribute of the sampling data;

[0040] A second analysis module, configured to analyze the operating status of the candidate computing nodes based on the time attribute and the volatility attribute, and determine the analysis result of the candidate computing nodes;

[0041] A scheduling module, configured to perform computing power scheduling on the candidate computing nodes based on the analysis results.

[0042] According to a third aspect, an embodiment of the present invention provides an electronic device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the computing power scheduling method described in the first aspect or any one of the embodiments of the first aspect.

[0043] According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores computer instructions for causing a computer to execute the computing power scheduling method described in the first aspect or any one of the embodiments of the first aspect.

[0044] It should be noted that for the corresponding beneficial effects of the computing power scheduling device, electronic device, and computer-readable storage medium provided by the embodiments of the present invention, please refer to the description of the corresponding beneficial effects of the computing power scheduling method above, and details will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 is a block diagram of the computing power scheduling method according to an embodiment of the present invention;

[0047] Figure 2 is a flowchart of the computing power scheduling method according to an embodiment of the present invention;

[0048] Figure 3 is a flowchart of the computing power scheduling method according to an embodiment of the present invention;

[0049] Figure 4 is a flowchart of the computing power scheduling method according to an embodiment of the present invention;

[0050] Figure 5 is a structural block diagram of the computing power scheduling device according to an embodiment of the present invention;

[0051] Figure 6 is a schematic hardware structure diagram of the electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0053] When performing computing power scheduling for multiple computing power nodes, generally, by analyzing the sampled data. In the scenario of cross-cloud platform computing power scheduling, there may be a problem of time asynchrony in the computing power resource monitoring data of the monitoring systems of different manufacturers, that is, there will be a problem of time asynchrony in the sampled data. On this basis, in order to analyze the sampled data, it is necessary to adjust the configuration parameters of the monitoring systems of different manufacturers to adjust the sampling frequency of the monitoring data and ensure that the sampled data used for scheduling analysis is synchronized in time. However, this method requires adjusting the configuration parameters of different common monitoring systems, resulting in low reliability of computing power scheduling.

[0054] Based on this, the embodiments of the present invention provide a computing power scheduling method. Without changing the configuration parameters of each monitoring system, it can utilize the existing historical monitoring data to evaluate its representation of the operating status of each computing power node at the current moment, helping the computing power scheduling system make more reliable decisions. Figure 1 Shows a processing flow of the computing power scheduling method. Based on the configuration parameters of each monitoring system, the sampled data of each candidate computing power node is obtained. On this basis, the operating status of each candidate computing power node is analyzed to determine the analysis result, and finally, the target computing power node is determined from the candidate computing power nodes for computing power scheduling.

[0055] According to the embodiments of the present invention, an embodiment of a computing power scheduling method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0056] In this embodiment, a computing power scheduling method is provided, which can be used in electronic devices such as scheduling platforms, servers, etc. Figure 2 Is a flowchart of the computing power scheduling method according to the embodiments of the present invention, as Figure 2 Shown, the process includes the following steps:

[0057] S11, obtain the candidate computing power nodes, the sampled data of the target computing power resource monitoring indicators of the candidate computing power nodes within a preset time interval, and the time attributes of the sampled data.

[0058] Among them, the preset time interval is the time interval between the current moment and the starting historical moment.

[0059] For each computing power node in the computing cluster, the electronic device obtains the operating parameters of each computing power node, such as CPU utilization rate, memory utilization rate, etc., compares the obtained operating parameters with the corresponding thresholds, and determines the candidate computing power nodes. Taking the CPU utilization rate as an example, a CPU utilization rate threshold is set, and the obtained CPU utilization rate is compared with this threshold. If it is less than this threshold, then this computing power node is determined as a candidate computing power node. Of course, other methods can also be used to determine the candidate computing nodes, and no limitation is imposed on it here, and it is specifically set according to actual requirements.

[0060] For example, the candidate computing power nodes obtained by the electronic device that meet the computing power scheduling conditions are represented in a list form, that is, P = {P1…,P i ,…,P n}.

[0061] For each candidate computing power node, it samples the original data obtained at the sampling frequency to obtain the sampled data. The electronic device obtains the sampled data of the target computing power resource monitoring index within the preset time interval from each candidate computing power node, where the target computing power resource monitoring index r is set according to actual computing power scheduling requirements, including but not limited to memory utilization rate, processing waiting duration, etc.

[0062] The preset time interval is set according to actual scheduling requirements. The starting point of the preset time interval is the starting historical moment, and the end point is the current moment. For example, the preset time interval T = [T s ,T e , where T s represents the starting historical moment, T e represents the current moment, and T e >T s .

[0063] Obtain the sampled data X = {X1…,X s ,…,X e} of the target computing power resource monitoring index r of each candidate computing power node in the list P within the preset time interval T = [T i ,…,X n , where, represents the sampled data of the target computing power resource monitoring index r of the i-th candidate computing power node P s ,T e within the time interval T = [T i , and i_m represents the number of sampling points, i_m≥2; represents the distance from the current moment T sThe most recent sampling, i.e., the latest sampling data; sequence X i The corresponding sampling time is

[0064] The time attributes of the sampling data include but are not limited to the sampling time of the sampling data, the sampling time interval, the sampling frequency, etc., which are specifically set according to actual requirements.

[0065] Among them, the candidate computing power nodes can belong to different cloud resource pools, and the sampling frequencies of the same type of monitoring metrics of different computing power nodes can be different, that is, the number of samples of the same type of monitoring metrics in different computing power nodes within the same time interval can be different. Optionally, the sampling time of each sampling data is represented by a Unix timestamp.

[0066] The specific details of this step will be analyzed in detail below.

[0067] S12, perform volatility analysis on the sampling data to determine the volatility attribute of the sampling data.

[0068] The said volatility analysis includes but is not limited to the analysis of standard deviation, variance, coefficient of variation or range of the sampling data, etc. After performing volatility analysis on the sampling data, the volatility attribute of the sampling data is obtained. Among them, the volatility attribute can be the result of a single volatility analysis or the fusion of multiple volatility analysis results. For example, the electronic device can only perform standard deviation analysis on the sampling data and determine the analysis result of the standard deviation as the volatility attribute; the electronic device can also perform standard deviation, variance and coefficient of variation analysis on the sampling data respectively, and then combine the weights of the standard deviation, variance and coefficient of variation to fuse the three to obtain the volatility attribute of the sampling data.

[0069] S13, analyze the operating status of the candidate computing power nodes based on the time attribute and the volatility attribute to determine the analysis result of the candidate computing power nodes.

[0070] The time attribute represents the sampling attribute of the candidate computing power nodes, that is, it represents the attribute of the candidate computing power nodes themselves; the volatility attribute represents the volatility of the sampling data, and the greater the volatility, the more unstable the sampling data. When the electronic device analyzes the operating status of the candidate computing power nodes, it fuses the time attribute and the volatility attribute to determine the analysis result of the candidate computing power nodes.

[0071] For example, when fusing the time attribute and the volatility attribute, it can be weighted calculation by combining their respective weights; it can also be normalization processing of the time attribute and the volatility attribute, etc. The fusion method is not limited here and is specifically set according to actual requirements.

[0072] Fusing the time attribute and the fluctuation attribute for analysis to obtain the analysis result of the operating state of the candidate computing power nodes, and thus the analysis result of the candidate computing power nodes can be determined. Since the analysis result is represented by specific numerical values, the analysis result can also be referred to as the credibility.

[0073] The specific details of this step will be analyzed in detail below.

[0074] S14, perform computing power scheduling on the candidate computing power nodes based on the analysis result.

[0075] As described above, the analysis result is represented by specific numerical values. Therefore, the electronic device can sort the magnitudes of the analysis results and perform computing power scheduling on the candidate computing power nodes based on the sorting result. For example, use the candidate computing power node with the best operating state in the analysis result for computing power scheduling, or use the N candidate computing power nodes with the best operating states for computing power scheduling, and so on.

[0076] In the computing power scheduling method provided in this embodiment, when analyzing the operating state of the candidate computing power nodes, the time attribute and the fluctuation attribute of the sampling data are comprehensively considered. Without changing the parameter configuration of the monitoring systems of each manufacturer, it can evaluate the computing power resource status of each candidate computing power node, thereby helping the computing power scheduling system make more reliable scheduling decisions and improving the reliability of computing power scheduling.

[0077] In this embodiment, a computing power scheduling method is provided, which can be used in electronic devices such as scheduling platforms, servers, etc. Figure 3 is a flowchart of the computing power scheduling method according to an embodiment of the present invention, as Figure 3 shown, the process includes the following steps:

[0078] S21, obtain the candidate computing power nodes, the sampling data of the target computing power resource monitoring indicators of the candidate computing power nodes within a preset time interval, and the time attribute of the sampling data.

[0079] Wherein, the preset time interval is the time interval between the current moment and the starting historical moment.

[0080] Specifically, the above S11 includes:

[0081] S211, obtain the candidate computing power nodes and the sampling data of the target computing power resource monitoring indicators of the candidate computing power nodes within a preset time interval.

[0082] Continuing with the above example, the candidate computing power nodes are represented in a list form as P = {P1…, P i ,…, P n}; the preset time interval is T = [T s , T e; For each candidate computing power node, taking the candidate computing power node P i as an example, the sampling data X of the target computing power resource monitoring index i is expressed as:

[0083] S212. For each candidate computing power node, obtain the sampling time of the sampling data.

[0084] For the candidate computing power node P i the sampling data the corresponding sampling time is

[0085] S213. Calculate the time attribute using the sampling time.

[0086] The electronic device calculates the sampling time interval or sampling frequency, etc. using the sampling time. In some embodiments, the above S213 includes:

[0087] (1) Calculate the average sampling time interval of the sampling data using the sampling time.

[0088] (2) Extract the sampling time of the most recent sampling data in the sampling time.

[0089] (3) Calculate the time difference between the sampling time of the most recent sampling data and the current moment, and determine the time attribute, where the time attribute includes the average sampling time interval and the time difference.

[0090] The electronic device calculates the average sampling time interval Span of the sampling data of each candidate computing power node using the sampling time, that is, Span = {span1…, span i ,…, span n}. Specifically, the span is calculated using the following formula i :

[0091]

[0092] where 1 ≤ i ≤ n, i_m > 1.

[0093] Calculate the time difference D between the sampling time of the most recent sampling data of each candidate computing power node in the list P within the time interval T and the current moment, that is, D = {d1…, d i ,…, d n}. Within the preset time interval T = [T s , T e , for the i-th candidate computing power node P i the calculation method of the time difference between the most recent sampling data of the target computing power resource monitoring index r and the current moment T and the current moment T e is:

[0094]

[0095] The closer the time of the latest sampled data is to the current moment, the higher the credibility of the sampled data representing the current operating state of the computing power resources; the shorter the sampling interval and the higher the sampling frequency of the data, the higher the credibility of the sampled data representing the current operating state of the computing power resources; the smaller the volatility of the sampled data within a period of time, the more stable the resource operating state, and the higher the credibility of the sampled data representing the current operating state of the computing power resources. Therefore, by comprehensively considering the time difference between the latest sampling and the current moment, the sampling interval of the data, and the volatility of the sampled data, the reliability of the computing power resource scheduling can be ensured. This method is particularly applicable to the scenario of cross-cloud platform computing power scheduling.

[0096] S22. Perform volatility analysis on the sampled data to determine the volatility attribute of the sampled data.

[0097] In this embodiment, the standard deviation is taken as an example to determine the volatility attribute of the sampled data. Specifically, within the preset time interval T = [T s , T e , for the sampled data X i of the target computing power resource monitoring index r of the i-th candidate computing power node P i , the calculation method of the standard deviation σ i is as follows:

[0098]

[0099] Among them,

[0100]

[0101] S23. Analyze the operating state of the candidate computing power nodes based on the time attribute and the volatility attribute to determine the analysis result of the candidate computing power nodes.

[0102] For details, please refer to Figure 2 S13 of the illustrated embodiment, which will not be elaborated here.

[0103] S24. Perform computing power scheduling on the candidate computing power nodes based on the analysis result.

[0104] For details, please refer to Figure 2 S14 of the illustrated embodiment, which will not be elaborated here.

[0105] For the computing power scheduling method provided in this embodiment, since the sampling time is determined by the parameter configurations of each monitoring system, using the sampling time to determine the time attribute can ensure the reliability of the time attribute and there is no need to unify the time attributes of each candidate computing power node during computing power scheduling.

[0106] In this embodiment, a computing power scheduling method is provided, which can be used in electronic devices such as scheduling platforms, servers, etc. Figure 4 It is a flowchart of the computing power scheduling method according to an embodiment of the present invention, as Figure 4 shown. The process includes the following steps:

[0107] S31, Obtain candidate computing power nodes, sampling data of target computing power resource monitoring indicators of candidate computing power nodes within a preset time interval, and time attributes of the sampling data.

[0108] Among them, the preset time interval is the time interval between the current moment and the starting historical moment.

[0109] For details, please refer to Figure 3 S21 of the embodiment shown, which will not be elaborated here.

[0110] S32, Perform volatility analysis on the sampling data to determine the volatility attribute of the sampling data.

[0111] For details, please refer to Figure 3 S22 of the embodiment shown, which will not be elaborated here.

[0112] S33, Analyze the operating status of candidate computing power nodes based on time attributes and volatility attributes to determine the analysis results of candidate computing power nodes.

[0113] Specifically, the above S33 includes:

[0114] S331, For a candidate computing power node, calculate the first ratio of the time attribute and the second ratio of the volatility attribute.

[0115] Among them, the first ratio is the ratio of the time attribute of the candidate computing power node among the time attributes of all candidate computing power nodes, and the second ratio is the ratio of the volatility attribute of the candidate computing power node among the volatility attributes of all candidate computing power nodes.

[0116] Specifically, the electronic device first calculates the sum of the time attributes of all candidate computing power nodes, and then calculates the ratio of each time attribute to the sum of the time attributes respectively, so as to determine the first ratio; correspondingly, the electronic device first calculates the sum of all volatility attributes, and then calculates the ratio of each volatility attribute to the sum of the volatility attributes respectively, so as to determine the second ratio.

[0117] In some embodiments, the first ratio of the calculated time attribute includes the first ratio of the average time interval and the first ratio of the time difference. Specifically, calculate the first ratio of the average time interval Span_P = {span_p1…,span_p i ,…,span_pn}, the first proportion D_P of the time difference = {d_p1…, d_p i , …, d_p n}. The second proportion of the fluctuation attribute is represented by the proportion of the standard deviation, that is, the standard deviation proportion Σ_P = {σ_p1…, σ_p i , …, σ_p n}. Among them,

[0118]

[0119]

[0120]

[0121] S332, obtain the first weight of the time attribute and the second weight of the fluctuation attribute.

[0122] The first weight and the second weight are set according to actual needs, and no limitations are imposed on them here.

[0123] S333, calculate the weighted sum of the first weight and the first proportion and the second weight and the second proportion.

[0124] The electronic device calculates the product of the weight and the corresponding proportion, and then calculates the sum of all products to obtain the weighted sum.

[0125] S334, determine the analysis result of the candidate computing power node based on the weighted sum.

[0126] As described above, if the analysis result is represented by a specific value, then the electronic device can determine the weighted sum as the analysis result of the candidate computing power node.

[0127] In some embodiments, the time attribute includes the average sampling time interval and the time difference, and the following formula is used to calculate the analysis result of the candidate computing power node:

[0128] s i = w span * span_p i + w d * d_p i + w σ * σ_p i

[0129] Among them, S i is the analysis result of the i-th candidate computing power node, w span is the first weight of the average sampling time interval, span_p i is the first proportion of the average sampling time interval, w d is the first weight of the time difference, d_p iis the first proportion of the time difference, w σ is the second weight of the fluctuation attribute, σ_p i is the second proportion of the fluctuation attribute.

[0130] S34. Perform computing power scheduling on candidate computing power nodes based on the analysis results.

[0131] Specifically, the above S34 includes:

[0132] S341. Compare the magnitudes of the analysis results, and determine the candidate computing power node corresponding to the smallest analysis result as the target computing power node.

[0133] The electronic device compares the magnitudes of the analysis results of each candidate computing power node, finds the smallest analysis result, i.e., Min(S), and determines the candidate computing power node corresponding to Min(S) as the target computing power node.

[0134] S342. Perform computing power scheduling based on the target computing power node.

[0135] The electronic device uses the target computing power node for computing power scheduling, that is, when a computing task is received, the computing task is assigned to the target computing power node for calculation. Subsequently, computing power scheduling is performed again to re-determine the target computing power node. The determination of the target computing power node can be analyzed at regular intervals or when a computing task is received, etc. There is no limitation on the analysis time of the computing power scheduling here, and it is specifically set according to actual requirements.

[0136] The computing power scheduling method provided in this embodiment determines the analysis results of candidate computing power nodes by using the proportions of the time attribute and the fluctuation attribute, and fuses the time attribute and the fluctuation attribute of each sampling data of the candidate computing power nodes, improving the reliability of the analysis results of the candidate computing power nodes.

[0137] In some other alternative embodiments, when there are at least two target computing power resource monitoring indicators, the above S33 includes:

[0138] (1) For each target computing power resource monitoring indicator, analyze the operating status of the candidate computing power nodes based on the time attribute and the fluctuation attribute, and determine the analysis result corresponding to the target computing power resource monitoring indicator.

[0139] (2) Obtain the weights of each target computing power resource monitoring indicator.

[0140] (3) Based on the weights of each target computing power resource monitoring indicator and the analysis results corresponding to the target computing power resource monitoring indicators, determine the analysis results of the candidate computing power nodes.

[0141] When it is necessary to analyze the running states of multiple target computing power resource monitoring metrics, different target computing power resource monitoring metrics are selected, and the above S31 - S33 are repeated to obtain the analysis results of each candidate computing power node on each target computing power resource monitoring metric. Then, weights are set for each target computing power resource monitoring metric, and the sum of the analysis results of all target computing power resource monitoring metrics of each candidate computing power node is calculated to obtain the analysis results of each candidate computing power node. Finally, S34 is executed to find the target computing power node.

[0142] When there are at least two target computing power resource monitoring metrics, the weights of each target computing power resource monitoring metric are used to comprehensively analyze the analysis results of all target computing power resource monitoring metrics, ensuring the accuracy of the analysis results of the candidate computing power nodes obtained.

[0143] In this embodiment, a computing power scheduling device is also provided. This device is used to implement the above - mentioned embodiments and preferred implementation manners, and those that have been described will not be repeated. As used hereinafter, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0144] This embodiment provides a computing power scheduling device, as Figure 5 shown, including:

[0145] An acquisition module 41, configured to acquire candidate computing power nodes, sampling data of target computing power resource monitoring metrics of the candidate computing power nodes within a preset time interval, and time attributes of the sampling data. The preset time interval is the time interval between the current moment and the starting historical moment;

[0146] A first analysis module 42, configured to perform volatility analysis on the sampling data to determine the volatility attribute of the sampling data;

[0147] A second analysis module 43, configured to analyze the running state of the candidate computing power nodes based on the time attribute and the volatility attribute to determine the analysis results of the candidate computing power nodes;

[0148] A scheduling module 44, configured to perform computing power scheduling on the candidate computing power nodes based on the analysis results.

[0149] In some implementation manners, the acquisition module 41 includes:

[0150] A first acquisition unit, configured to acquire the sampling time of the sampling data for each of the candidate computing power nodes;

[0151] A first calculation unit, configured to calculate the time attribute using the sampling time.

[0152] In some embodiments, the computing unit includes:

[0153] A first computing subunit, configured to calculate an average sampling time interval of the sampling data by using the sampling time;

[0154] An extraction subunit, configured to extract a sampling time of the most recent sampling data in the sampling time;

[0155] A second computing subunit, configured to calculate a time difference between the sampling time of the most recent sampling data and the current moment, and determine the time attribute, where the time attribute includes the average sampling time interval and the time difference.

[0156] In some embodiments, the second analysis module 43 includes:

[0157] A second computing unit, configured to calculate a first ratio of the time attribute and a second ratio of the fluctuation attribute for the candidate computing power node, where the first ratio is a ratio of the time attribute of the candidate computing power node among the time attributes of all the candidate computing power nodes, and the second ratio is a ratio of the fluctuation attribute of the candidate computing power node among the fluctuation attributes of all the candidate computing power nodes;

[0158] A second obtaining unit, configured to obtain a first weight of the time attribute and a second weight of the fluctuation attribute;

[0159] A third computing unit, configured to calculate a weighted sum of the first weight and the first ratio and the second weight and the second ratio;

[0160] A first determining unit, configured to determine an analysis result of the candidate computing power node based on the weighted sum.

[0161] In some embodiments, the time attribute includes an average sampling time interval and a time difference, and the analysis result of the candidate computing power node is calculated by using the following formula:

[0162] s i = w span *span_p i + w d *d_p i + w σ *σ_p i

[0163] where s i is the analysis result of the i-th candidate computing power node, w span is the first weight of the average sampling time interval, span_p i is the first ratio of the average sampling time interval, w dis the first weight of the time difference, d_p i is the first proportion of the time difference, w σ is the second weight of the fluctuation attribute, σ_p i is the second proportion of the fluctuation attribute.

[0164] In some embodiments, when there are at least two target computing power resource monitoring metrics, the second analysis module 43 includes:

[0165] An analysis unit, configured to analyze the operating state of the candidate computing power nodes based on the time attribute and the fluctuation attribute for each of the target computing power resource monitoring metrics, and determine the analysis result corresponding to the target computing power resource monitoring metric;

[0166] A third acquisition unit, configured to acquire the weights of each of the target computing power resource monitoring metrics;

[0167] A third determination unit, configured to determine the analysis result of the candidate computing power nodes based on the weights of each of the target computing power resource monitoring metrics and the analysis result corresponding to the target computing power resource monitoring metric.

[0168] In some embodiments, the scheduling module 44 includes:

[0169] A comparison unit, configured to compare the magnitudes of the analysis results, and determine the candidate computing power node corresponding to the smallest analysis result as the target computing power node;

[0170] A scheduling unit, configured to perform computing power scheduling based on the target computing power node.

[0171] The computing power scheduling device in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0172] The further function descriptions of the above-mentioned respective modules are the same as those in the corresponding embodiments above, and will not be elaborated here.

[0173] The embodiment of the present invention also provides an electronic device having the above Figure 5 shown computing power scheduling device.

[0174] Please refer to Figure 6 , Figure 6 is a schematic structural diagram of an electronic device provided by an alternative embodiment of the present invention. As shown in Figure 6As shown, the electronic device may include: at least one processor 51, such as a CPU (Central Processing Unit), at least one communication interface 53, a memory 54, and at least one communication bus 52. Among them, the communication bus 52 is used to realize the connection and communication between these components. Among them, the communication interface 53 may include a display screen (Display) and a keyboard (Keyboard). Optionally, the communication interface 53 may also include a standard wired interface and a wireless interface. The memory 54 may be a high-speed RAM memory (Random Access Memory, volatile random access memory), or a non-volatile memory, such as at least one disk memory. Optionally, the memory 54 may also be at least one storage device located far from the aforementioned processor 51. Among them, the processor 51 may be combined with Figure 5 the device described, the memory 54 stores an application program, and the processor 51 calls the program code stored in the memory 54 to execute any of the above method steps.

[0175] Among them, the communication bus 52 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The communication bus 52 may be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 6 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0176] Among them, the memory 54 may include a volatile memory (English: volatile memory), such as a random access memory (English: random-access memory, abbreviation: RAM); the memory may also include a non-volatile memory (English: non-volatile memory), such as a flash memory (English: flash memory), a hard disk (English: hard disk drive, abbreviation: HDD) or a solid-state drive (English: solid-state drive, abbreviation: SSD); the memory 54 may also include a combination of the above types of memories.

[0177] Among them, the processor 51 may be a central processing unit (English: central processing unit, abbreviation: CPU), a network processor (English: network processor, abbreviation: NP), or a combination of a CPU and an NP.

[0178] Among them, the processor 51 may further include a hardware chip. The above-mentioned hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above-mentioned PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0179] Optionally, the memory 54 is further configured to store program instructions. The processor 51 may call the program instructions to implement the computing power scheduling method as shown in any embodiment of the present application.

[0180] The embodiment of the present invention further provides a non-transitory computer storage medium, which stores computer-executable instructions, and the computer-executable instructions can execute the computing power scheduling method in any of the above method embodiments. Among them, the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD), or a solid-state drive (SSD), etc.; the storage medium may also include a combination of the above types of memories.

[0181] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations fall within the scope defined by the appended claims.

Claims

1. A computing power scheduling method, characterized in that, Including: Obtaining candidate computing power nodes, sampling data of target computing power resource monitoring metrics of the candidate computing power nodes within a preset time interval, and time attributes of the sampling data, where the preset time interval is the time interval between the current moment and the starting historical moment; Performing volatility analysis on the sampling data to determine the volatility attribute of the sampling data; Analyzing the operating status of the candidate computing power nodes based on the time attributes and the volatility attributes to determine the analysis results of the candidate computing power nodes, including: for the candidate computing power nodes, calculating a first proportion of the time attributes and a second proportion of the volatility attributes, where the first proportion is the proportion of the time attributes of the candidate computing power nodes among the time attributes of all the candidate computing power nodes, and the second proportion is the proportion of the volatility attributes of the candidate computing power nodes among the volatility attributes of all the candidate computing power nodes; obtaining a first weight of the time attributes and a second weight of the volatility attributes; calculating the weighted sum of the first weight and the first proportion and the second weight and the second proportion; determining the analysis results of the candidate computing power nodes based on the weighted sum; Performing computing power scheduling on the candidate computing power nodes based on the analysis results, including: comparing the magnitudes of the analysis results, and determining the candidate computing power node corresponding to the smallest analysis result as the target computing power node; performing computing power scheduling based on the target computing power node.

2. The method according to claim 1, characterized in that, Obtaining the time attributes of the sampling data includes: For each of the candidate computing power nodes, obtaining the sampling time of the sampling data; Calculating the time attributes using the sampling time.

3. The method according to claim 2, wherein The calculating the time attributes using the sampling time includes: Calculating the average sampling time interval of the sampling data using the sampling time; Extracting the sampling time of the most recent sampling data in the sampling time; Calculating the time difference between the sampling time of the most recent sampling data and the current moment to determine the time attributes, where the time attributes include the average sampling time interval and the time difference.

4. The method according to claim 1, wherein The time attributes include the average sampling time interval and the time difference, and the analysis results of the candidate computing power nodes are calculated using the following formula: s i = w span * span_p i + w d * d_p i + w σ * σ_p i where s i is the analysis result of the i-th candidate computing power node, w span is the first weight of the average sampling time interval, span_p i is the first proportion of the average sampling time interval, w d is the first weight of the time difference, d_p i is the first proportion of the time difference, w σ is the second weight of the fluctuation attribute, σ_p i is the second proportion of the fluctuation attribute.

5. The method according to claim 1, wherein When there are at least two target computing power resource monitoring metrics, the analyzing the operating status of the candidate computing power nodes based on the time attributes and the volatility attributes to determine the analysis results of the candidate computing power nodes includes: For each of the target computing power resource monitoring metrics, analyzing the operating status of the candidate computing power nodes based on the time attributes and the volatility attributes to determine the analysis results corresponding to the target computing power resource monitoring metrics; Obtaining the weights of each of the target computing power resource monitoring metrics; Determining the analysis results of the candidate computing power nodes based on the weights of each of the target computing power resource monitoring metrics and the analysis results corresponding to the target computing power resource monitoring metrics.

6. A computing power scheduling device, characterized in that, Including: An acquisition module, configured to acquire candidate computing power nodes, sampling data of target computing power resource monitoring metrics of the candidate computing power nodes within a preset time interval, and time attributes of the sampling data, where the preset time interval is a time interval between the current moment and the starting historical moment; A first analysis module, configured to perform volatility analysis on the sampling data to determine the volatility attribute of the sampling data; A second analysis module, configured to analyze the operating state of the candidate computing power nodes based on the time attributes and the volatility attributes, determine the analysis result of the candidate computing power nodes. For the candidate computing power nodes, calculate a first proportion of the time attributes and a second proportion of the volatility attributes. The first proportion is the proportion of the time attributes of the candidate computing power nodes among the time attributes of all the candidate computing power nodes, and the second proportion is the proportion of the volatility attributes of the candidate computing power nodes among the volatility attributes of all the candidate computing power nodes; obtain a first weight of the time attributes and a second weight of the volatility attributes; Calculate the weighted sum of the first weight and the first proportion and the second weight and the second proportion; determine the analysis result of the candidate computing power nodes based on the weighted sum; A scheduling module, configured to perform computing power scheduling on the candidate computing power nodes based on the analysis result, compare the magnitudes of the analysis results, and determine the candidate computing power node corresponding to the smallest analysis result as the target computing power node; Perform computing power scheduling based on the target computing power node.

7. An electronic device, characterized in that, Including: A memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the computing power scheduling method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the computing power scheduling method according to any one of claims 1-5.

Images