CN116302532A - A computing power node marking method and device - Google Patents

Abstract

The present application discloses a computing power node marking method and device, the method comprising: obtaining the first mark of the computing power node to be marked; the first mark is used to mark the resource utilization level of the computing power node; obtaining the computing power node The second mark; the second mark is used to measure the credibility of the first mark; the first mark and the second mark are used as scheduling marks of computing power nodes. In this way, the computing power nodes are marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels. In this way, when scheduling computing power nodes, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

Description

A computing power node marking method and device

technical field

The embodiments of the present application relate to the field of communication technologies, and in particular to a computing power node marking method, device, electronic equipment, and readable storage medium.

Background technique

At present, edge cloud, hybrid cloud, and distributed cloud have gradually become hot spots in the field of cloud computing, which puts forward higher requirements for efficient scheduling of computing power nodes. Wherein, the calculation node includes CPU (Central Processing Unit, central processing unit), GPU (Graphics Processing Unit, graphics processing unit), disk and memory and other computing power resources, and of course it is not limited thereto.

A good scheduling strategy can improve the scheduling efficiency of computing power nodes, so that users can get better cloud services. The current labeling scheme for computing power nodes is: label CPU, disk IO (Input/Output, input/output) and other computing power nodes with CPU labels, IO labels, etc., and at the same time, attach CPU labels and IO labels to tasks according to classification Or ordinary tags, and then enter a different tag queue. Furthermore, when scheduling computing power nodes, the scheduler assigns computing power nodes to corresponding tasks according to labels.

In the process of implementing this application, the inventor found that the current schemes are all based on the type or category of the computing power node as the label marking node, the granularity is relatively coarse, and the quantification is lacking. Scheduling based on such coarse-grained tags cannot accurately assign tasks to computing power nodes that meet the load conditions.

Contents of the invention

The embodiment of the present application provides a computing power node marking method, device, electronic equipment, and readable storage medium, so as to mark computing power nodes through quantized tags, so that there are finer-grained tags, so that when scheduling computing power nodes When , tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved. The technical solution is as follows:

According to an aspect of an embodiment of the present application, a method for marking computing power nodes is provided, and the method may include the following steps:

Obtain the first mark of the computing power node to be marked; the first mark is used to mark the resource utilization level of the computing power node;

Obtain the second mark of the computing power node; the second mark is used to measure the credibility of the first mark;

The first mark and the second mark are used as the scheduling marks of the computing power nodes.

Optionally, the step of obtaining the first mark of the computing power node to be marked may include:

Based on the resource utilization level determination formula, the first mark of the computing power node to be marked is obtained.

Optionally, the resource utilization rate determination formula may include:

Among them, Label_1 indicates the first label of the computing power node; P ₁ indicates that when M ₁ <v ₁ , the resource utilization level corresponding to the first label is P ₁ ; P _i+1 indicates that when v _i+1 >M ₁ ≥ When v _i , the resource utilization level corresponding to the first mark is P _i+1 ; P _n+1 means that when M ₁ ≥ v _n , the resource utilization level corresponding to the first mark is P _n+1 ;

M ₁ represents the average value of the m sampling data of the computing power node within the first sampling period; v ₁ , ..., _vi , ..., v _n represent the preset first classification threshold point, ..., the i Classification threshold point, ..., the nth classification threshold point; i, n, m are all positive integers.

Optionally, the step of obtaining the second mark of the computing power node may include:

Based on the calculation formula of the coefficient of variation, the second mark of the computing power node is obtained.

Optionally, the formula for calculating the coefficient of variation may include:

Among them, Label_2 represents the second label of the computing power node; CV represents the coefficient of variation; M ₂ represents the mean value of the p sampling data of the computing power node in the second sampling time; where the second sampling time is longer than the first sampling time; the first m sampling data are sampled within the sampling duration, and p is greater than m; x _k represents the kth sampling data among the p sampling data; p, k, and m are all positive integers.

Optionally, the step of using the first mark and the second mark as the scheduling mark of the computing power node may include:

Update the first mark according to the first sampling duration as an update period;

Update the second mark according to the second sampling duration as an update period; wherein, the second sampling duration is longer than the first sampling duration;

The updated first mark and the second mark are used as the scheduling marks of the computing power nodes.

Optionally, after using the first mark and the second mark as the scheduling marks of computing power nodes, it may also include:

When receiving a scheduling instruction for a computing power node, obtaining a first mark of each computing power node with a scheduling mark;

Select the computing power node with the lowest level of resource utilization among the obtained first marks;

Obtain the second mark of the computing power node with the lowest level of resource utilization as the target second mark;

When there are multiple target second marks, the computing power node with the highest reliability among the target second marks is selected as the computing power node for resource scheduling.

In the second aspect, the embodiment of the present application also provides a computing power node marking device, which may include:

The first acquisition module is used to obtain the first mark of the computing power node to be marked; the first mark is used to mark the resource utilization level of the computing power node;

The second acquisition module is used to obtain the second mark of the computing power node; the second mark is used to measure the credibility of the first mark;

The marking module is configured to use the first marking and the second marking as scheduling markings of computing power nodes.

Optionally, the first acquisition module may be specifically used for:

Based on the resource utilization level determination formula, the first mark of the computing power node to be marked is obtained.

Optionally, the resource utilization rate determination formula may include:

Among them, Label_1 indicates the first label of the computing power node; P ₁ indicates that when M ₁ <v ₁ , the resource utilization level corresponding to the first label is P ₁ ; P _i+1 indicates that when v _i+1 >M ₁ ≥ When v _i , the resource utilization level corresponding to the first mark is P _i+1 ; P _n+1 means that when M ₁ ≥ v _n , the resource utilization level corresponding to the first mark is P _n+1 ;

M ₁ represents the average value of the m sampling data of the computing power node within the first sampling period; v ₁ , ..., _vi , ..., v _n represent the preset first classification threshold point, ..., the i Classification threshold point, ..., the nth classification threshold point; i, n, m are all positive integers.

Optionally, the second acquisition module can be specifically used for:

Based on the calculation formula of the coefficient of variation, the second mark of the computing power node is obtained.

Optionally, the formula for calculating the coefficient of variation may include:

Among them, Label_2 represents the second label of the computing power node; CV represents the coefficient of variation; M ₂ represents the mean value of the p sampling data of the computing power node in the second sampling time; where the second sampling time is longer than the first sampling time; the first m sampling data are sampled within the sampling duration, and p is greater than m; x _k represents the kth sampling data among the p sampling data; p, k, and m are all positive integers.

Optionally, the marking module can include:

The first update unit is configured to update the first mark according to the first sampling duration as an update cycle;

The second updating unit is configured to update the second mark according to the second sampling duration as an updating period; wherein, the second sampling duration is longer than the first sampling duration;

The marking unit is configured to use the updated first marking and the second marking as scheduling markings of computing power nodes.

Optionally, the device may also include:

The third acquisition module is used to obtain the first and second marks of each computing power node with the scheduling mark when receiving a scheduling instruction for the computing power node after using the first mark and the second mark as the scheduling mark of the computing power node. mark;

The first selection module is configured to select the computing power node with the lowest level of resource utilization among the acquired first tags;

The fourth obtaining module is used to obtain the second mark of the computing power node with the lowest level of resource utilization as the target second mark;

The second selection module is configured to select the computing power node with the highest reliability among the target second markings as the computing power node for resource scheduling when there are multiple target second markings.

In the third aspect, the embodiment of the present application also provides an electronic device, the electronic device includes a processor and a memory, and a computer program is stored in the memory, and the computer program is loaded and executed by the processor to implement the method according to any one of the first aspect .

In a fourth aspect, the embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, and the computer program is loaded and executed by a processor to implement the method according to any one of the first aspect.

The technical solutions provided by the embodiments of the present application may include the following beneficial effects:

The computing power node marking scheme provided in the embodiment of the present application can first obtain the first mark of the computing power node to be marked, and the first mark is used to mark the resource utilization level of the computing power node. Then, the second mark of the computing power node can be obtained, and the second mark is used to measure the credibility of the first mark. Afterwards, the first mark and the second mark are used as the scheduling marks of the computing power nodes. In this way, the computing power nodes are marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels. In this way, when scheduling computing power nodes, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

Description of drawings

Fig. 1 is a flow chart of a method for marking a computing power node provided by an embodiment of the present application;

Fig. 2 is a flow chart of another computing power node marking method provided by the embodiment of the present application;

Fig. 3 is a structural block diagram of a computing power node marking device provided by an embodiment of the present application;

Fig. 4 is a structural block diagram of an electronic device provided by an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solution and advantages of the present application clearer, the implementation manners of the present application will be further described in detail below in conjunction with the accompanying drawings.

The current scheme for marking computing power nodes is: to mark CPU, disk IO and other computing power nodes with CPU tags, IO tags, etc. At the same time, assign CPU tags, IO tags, or common tags to tasks according to the classification, and then enter different tag queues. Furthermore, when scheduling computing power nodes, the scheduler assigns computing power nodes to corresponding tasks according to labels.

In the process of implementing this application, the inventor found that the current schemes are all based on the type or category of the computing power node as the label marking node, the granularity is relatively coarse, and the quantification is lacking. Scheduling based on such coarse-grained tags cannot accurately assign tasks to computing power nodes that meet the load conditions.

In order to solve the above technical problems, embodiments of the present application provide a computing power node marking method, device, electronic equipment, and readable storage medium.

The method for marking computing power nodes provided by the embodiment of the present application will first be described below. The computing power node marking method provided in the embodiment of this application may include the following steps:

Obtain the first mark of the computing power node to be marked; the first mark is used to mark the resource utilization level of the computing power node;

Obtain the second mark of the computing power node; the second mark is used to measure the credibility of the first mark;

The first mark and the second mark are used as the scheduling marks of the computing power nodes.

By applying the computing power node marking scheme provided in the embodiment of the present application, the first mark of the computing power node to be marked can be obtained first, and the first mark is used to mark the resource utilization level of the computing power node. Then, the second mark of the computing power node can be obtained, and the second mark is used to measure the credibility of the first mark. Afterwards, the first mark and the second mark are used as the scheduling marks of the computing power nodes. In this way, the computing power nodes are marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels. In this way, when scheduling computing power nodes, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

The method for marking computing power nodes provided by the embodiment of the present application will be described in detail below with reference to FIG. 1 and FIG. 2 . Figure 1 is a flow chart of a method for marking a computing power node provided in an embodiment of the present application. Referring to Figure 1, the method for marking a computing power node may include the following steps:

S101: Obtain the first mark of the computing power node to be marked; the first mark is used to mark the resource utilization level of the computing power node;

It can be understood that the first mark may be used to mark the resource utilization level of the computing power node, and its expression form may be numbers, letters or Arabic numerals, but of course it is not limited thereto.

For example, resource utilization levels may be 0, 1, 2, 3, . . . , 9. Among them, the larger the number, the higher the resource utilization level of the computing power node. For another example, the resource utilization level may be A, B, C, D, E, F, G, H, I, J. Among them, the lower the letter is in the alphabet, the higher the resource utilization level of the computing power node.

The higher the resource utilization level of the computing power node, the less idle resources the computing power node has, and vice versa, the more idle resources. In this way, computing power nodes can be marked with quantized first tags, so that computing power nodes have labels with a finer granularity.

It can be understood that quantification in a narrow sense refers to representation by numerical value. Quantification in a broad sense refers to the form of expression that can be compared/compared, such as grade A, grade B and grade C, and grade A<B<C. Quantification in the embodiments of the present application can be quantification in a narrow sense or in a broad sense, all of which are reasonable.

Optionally, the first mark of the computing power node to be marked may be obtained based on a resource utilization level determination formula. The formula for determining the resource utilization level may be:

Among them, Label_1 indicates the first label of the computing power node; P ₁ indicates that when M ₁ <v ₁ , the resource utilization level corresponding to the first label is P ₁ ; P _i+1 indicates that when v _i+1 >M ₁ ≥ When v _i , the resource utilization level corresponding to the first flag is P _i+1 ; P _n+1 means that when M ₁ ≥ v _n , the resource utilization level corresponding to the first flag is P _n+1 . Wherein, P ₁ represents the first resource utilization rate level, ..., P _i represents the i-th resource utilization rate level, ..., P _n+1 represents the n+1-th resource utilization rate level.

M ₁ represents the average value of the m sampling data of the computing power node within the first sampling period; v ₁ , ..., _vi , ..., v _n represent the preset first classification threshold point, ..., the i Classification threshold point, ..., the nth classification threshold point; i, n, m are all positive integers. Wherein, those skilled in the art may pre-set the above classification threshold points according to experience and/or label fineness requirements, and of course it is not limited thereto.

It can be understood that the formula for determining the resource utilization level is not limited to the above content. Among them, those skilled in the art can also configure a weight coefficient for the above mean value according to the specific conditions of the computing power node, and of course it is not limited thereto.

Among them, through the determination formula of the resource utilization rate, the preset grading threshold point can be compared with the mean value of the m sampling data of the computing power node in the first sampling period, and mapped to the corresponding resource utilization rate level according to the comparison result . In this way, information about resource utilization levels of computing power nodes can be obtained, that is, idle resource information of computing power nodes can be obtained.

S102: Obtain the second mark of the computing power node; the second mark is used to measure the credibility of the first mark;

It can be understood that the second mark may be used to mark the credibility of the first mark of the computing power node. The higher the reliability, the higher the reliability of the determination result for the resource utilization level corresponding to the first mark. Wherein, Arabic numerals, percentage figures, letters, etc. may be used to represent the credibility of the first mark, and of course it is not limited thereto.

For example, the reliability may be 0, 1, 2, 3, . . . , 9. Wherein, the larger the Arabic numeral, the higher the reliability, that is, the higher the reliability of the determination result of the resource utilization level corresponding to the first mark. For another example, the confidence level may be 10%, 20% and so on. Among them, the higher the value of the percentage figure, the higher the reliability.

Of course, the second mark can also measure the credibility of the first mark of the computing power node through the coefficient of variation. Among them, the larger the coefficient of variation, the greater the degree of deviation, the greater the risk, and the lower the credibility; the smaller the coefficient of variation, the smaller the degree of deviation, the smaller the risk, and the higher the credibility. Optionally, the second mark of the computing power node can be obtained based on the calculation formula of the coefficient of variation. Wherein, the formula for calculating the coefficient of variation can be:

Among them, Label_2 represents the second label of the computing power node; CV represents the coefficient of variation; M ₂ represents the mean value of the p sampling data of the computing power node in the second sampling time; where the second sampling time is longer than the first sampling time; the first m sampling data are sampled within the sampling duration, and p is greater than m; x _k represents the kth sampling data among the p sampling data; p, k, and m are all positive integers.

It can be understood that the formula for calculating the coefficient of variation is not limited to the above content. Wherein, those skilled in the art can also configure a calibration coefficient for the above variation coefficient CV according to the specific conditions of the computing power node, and of course it is not limited thereto.

In this way, the coefficient of variation is calculated through the formula for calculating the coefficient of variation, and the degree of reliability of the resource utilization level corresponding to the first mark can be determined through the coefficient of variation.

S103: Use the first mark and the second mark as scheduling marks of computing power nodes.

It can be understood that the first mark and the second mark are used as scheduling marks of computing power nodes. In this way, the computing power nodes can be marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels.

Therefore, when scheduling computing power nodes, based on the first mark, computing power nodes with low resource utilization levels can be preferentially selected for scheduling, so that computing power nodes with more idle resource capabilities can be quickly screened out, making them as fast and efficient as possible work.

When there are multiple computing power nodes with the lowest level of resource utilization, a computing power node with high reliability can also be selected based on the second flag, so that it can be selected from multiple computing power nodes with the lowest level of resource utilization More reliable computing nodes. In this way, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

Since the resource status of computing power nodes is always changing dynamically, in order to obtain more accurate scheduling labels, the above-mentioned steps of using the first label and the second label as the scheduling labels of computing power nodes may specifically include:

Update the first mark according to the first sampling duration as an update period;

Update the second mark according to the second sampling duration as an update period; wherein, the second sampling duration is longer than the first sampling duration;

The updated first mark and the second mark are used as the scheduling marks of the computing power nodes.

In this way, the first flag and the second flag can be dynamically updated by using the first sampling duration and the second sampling duration as an update period, so that more accurate scheduling labels can be obtained.

Wherein, the second sampling duration is longer than the first sampling duration, that is, the first mark adopts sampling in a shorter time interval, and the second mark adopts sampling in a longer time interval. In this way, the data in a shorter time interval can more accurately represent the resource utilization level of the current node, and the data in a longer time interval can more accurately evaluate the credibility of the first mark. This processing method can make the obtained first mark and second mark more accurate.

In order to improve the scheduling efficiency, after using the first mark and the second mark as the scheduling mark of the computing power node, it can also include:

When receiving a scheduling instruction for a computing power node, obtaining a first mark of each computing power node with a scheduling mark;

Select the computing power node with the lowest level of resource utilization among the obtained first marks;

Obtain the second mark of the computing power node with the lowest level of resource utilization as the target second mark;

When there are multiple target second marks, the computing power node with the highest reliability among the target second marks is selected as the computing power node for resource scheduling.

In this way, by quantitatively marking the operating status of the computing power node, the label can more accurately and reliably reflect the resource status of the computing power node to which it belongs. Moreover, when scheduling computing power nodes, there is no need to spend a lot of time and computing resources to calculate the optimal computing power nodes. so. Tasks can be quickly and accurately assigned to computing power nodes that meet the load conditions, improving the resource utilization of computing power nodes.

The method for marking computing power nodes provided by the embodiment of the present application will be described in detail below with reference to specific examples. Fig. 2 is a flow chart of another computing power node marking method provided in the embodiment of the present application. Referring to Fig. 2, the computing power node marking method may include the following steps:

S201: Initialize the parameter configuration of the computing power node to be marked, and start the timer.

First, the first label Label_1=0 and the second label Label_2=0 of the computing nodes to be labeled may be initialized.

Then, the grading threshold point sequence V={v _{1 .} . . , v _i , . . . , v _n } of the preset monitoring index X is initialized. Wherein, v ₁ , ..., _vi , ..., v _n represent the preset first classification threshold point, ..., i-th classification threshold point, ..., n-th classification threshold point.

Afterwards, the hierarchical state set P={P ₁ , . . . , P _i , . . . , P _n+1 } is initialized. Wherein, P ₁ represents the first resource utilization rate level, ..., P _i represents the i-th resource utilization rate level, ..., P _n+1 represents the n+1-th resource utilization rate level. i and n are positive integers.

In addition, the timer timer=0 may be initialized and the time interval T for resetting the timer may be initialized, and the first sampling duration T1 and the second sampling duration T2 may be initialized, and T2>T1.

S202: When the time interval T for resetting the timer is reached, reset the timer; and obtain the first sampling data sequence within the first sampling duration T1 before the current moment and the second sampling data within the second sampling duration T2 sequence.

Specifically, the timer timer can be started, and when the time interval T is reached, the timer is reset; at the same time, the first sampling data sequence X1={x ₁ ...,x _k ,...,x _m } and the second sampling data sequence X2={x ₁ ...,x _k ,...,x _m ,...,x _p in the second sampling duration T2 time interval before the current moment }. p, k, and m are positive integers, and p is greater than m.

x ₁ ..., x _k , ..., x _m represent m sampling data in the first sampling data sequence; x ₁ ..., x _k , ..., x _m , ..., x _p represent p in the second sampling data sequence sample data.

S203: Calculate the mean value of the first sampled data sequence, and then compare it with each classification threshold point in the classification threshold point sequence to obtain a classification state corresponding to the classification state set, and use it as a first mark.

计算。M₁表示算力节点在第一采样时长内m个采样数据的均值。然后与分级阈值点序列V中的各分级阈值点进行比较，获得分级状态集对应的分级状态，并作为第一标记Label_1的值。其中，第一标记对应取值的计算方法如下，该计算方式也可称为资源利用率等级确定公式：Among them, the mean value of the first sampling data sequence X1 can be calculated by the formula

calculate. M ₁ represents the mean value of the m sampling data of the computing power node within the first sampling period. Then compare it with each grading threshold point in the grading threshold point sequence V to obtain the grading state corresponding to the grading state set, and use it as the value of the first label Label_1. Wherein, the calculation method of the value corresponding to the first flag is as follows, and this calculation method may also be called the resource utilization level determination formula:

P ₁ indicates that when M ₁ <v ₁ , the resource utilization level corresponding to the first mark is P ₁ ; P _i+1 indicates that when v _i+1 >M ₁ ≥ v _i , the resource utilization rate corresponding to the first mark The level is P _i+1 ; P _n+1 means that when M ₁ ≥ v _n , the resource utilization level corresponding to the first mark is P _n+1 ;

M ₁ represents the average value of the m sampling data of the computing power node within the first sampling period; v ₁ , ..., _vi , ..., v _n represent the preset first classification threshold point, ..., the i Classification threshold point, ..., the nth classification threshold point; i, n, m are all positive integers.

S204: Calculate the coefficient of variation corresponding to the second sampled data sequence as a second mark.

Specifically, the second marker can be calculated by the following coefficient of variation calculation formula:

M₂表示算力节点在第二采样时长内p个采样数据的均值。Among them, the mean value of the second sampling data sequence X2

M ₂ represents the mean value of the p sampling data of the computing power node within the second sampling period.

S205: Update the computing power node corresponding to the first mark and the second mark as the updated scheduling mark of the computing power node; based on the timing of the timer, repeat the above steps.

For clarity, the computing power node marking method provided in the embodiment of the present application is described again by using the computing power node to be marked as a CPU as an example.

See Table 1. Table 1 shows the sampling data of CPU resource utilization of 5 CPU computing power nodes within the set first sampling period and the second sampling period.

Table 1

It can be seen from Table 1 that 5 sampling data of each computing power node are collected in the first sampling period, and 10 sampling data of each computing power node are collected in the second sampling period. Wherein, timing start times of the first sampling duration and the second sampling duration are the same. In addition, the sampling frequency is the sampling frequency set by the sampling device.

In addition, technicians can set the grading threshold point sequence as [10, 20, 30, 40, 50, 60, 70, 80, 90], and the grading state set as [0, 1, 2, 3, 4, 5 , 6, 7, 8, 9]. Therefore, the first mark and the second mark can be calculated based on the above resource utilization level determination formula and variation coefficient calculation formula, and the calculation results are listed in Table 2.

Table 2

It can be understood that the first mark can be further refined by setting a finer-grained classification threshold; or adopt a fine-grained threshold at a high position and a coarse-grained threshold at a low position, for example [20, 40, 50, 60, 70 , 75, 80, 82.5, 85, 87.5, 90] to further refine the first label, thereby providing a finer-grained first label. Furthermore, during the node scheduling process, candidate nodes can be quickly screened out by the first mark and the second mark.

Corresponding to the above method embodiment, the embodiment of the present application also provides a computing power node marking device, see Figure 3, the device may include:

The first obtaining module 301 is used to obtain the first mark of the computing power node to be marked; the first mark is used to mark the resource utilization level of the computing power node;

The second acquiring module 302 is configured to obtain a second mark of the computing power node; the second mark is used to measure the credibility of the first mark;

The marking module 303 is configured to use the first marking and the second marking as scheduling markings of computing power nodes.

Optionally, the first acquiring module 301 may be specifically configured to:

Based on the resource utilization level determination formula, the first mark of the computing power node to be marked is obtained.

Optionally, the resource utilization rate determination formula may include:

Among them, Label_1 indicates the first label of the computing power node; P ₁ indicates that when M ₁ <v ₁ , the resource utilization level corresponding to the first label is P ₁ ; P _i+1 indicates that when v _i+1 >M ₁ ≥ When v _i , the resource utilization level corresponding to the first mark is P _i+1 ; P _n+1 means that when M ₁ ≥ v _n , the resource utilization level corresponding to the first mark is P _n+1 ;

M ₁ represents the average value of the m sampling data of the computing power node within the first sampling period; v ₁ , ..., _vi , ..., v _n represent the preset first classification threshold point, ..., the i Classification threshold point, ..., the nth classification threshold point; i, n, m are all positive integers.

Optionally, the second obtaining module 302 may be specifically configured to:

Based on the calculation formula of the coefficient of variation, the second mark of the computing power node is obtained.

Optionally, the formula for calculating the coefficient of variation may include:

Among them, Label_2 represents the second label of the computing power node; CV represents the coefficient of variation; M ₂ represents the mean value of the p sampling data of the computing power node in the second sampling time; where the second sampling time is longer than the first sampling time; the first m sampling data are sampled within the sampling duration, and p is greater than m; x _k represents the kth sampling data among the p sampling data; p, k, and m are all positive integers.

Optionally, the marking module 303 may include:

The first update unit is configured to update the first mark according to the first sampling duration as an update period;

The second updating unit is configured to update the second mark according to the second sampling duration as an updating period; wherein, the second sampling duration is longer than the first sampling duration;

The marking unit is configured to use the updated first marking and the second marking as scheduling markings of computing power nodes.

Optionally, the device may also include:

The third acquisition module is used to obtain the first and second marks of each computing power node with the scheduling mark when receiving a scheduling instruction for the computing power node after using the first mark and the second mark as the scheduling mark of the computing power node. mark;

The first selection module is configured to select the computing power node with the lowest level of resource utilization among the acquired first tags;

The fourth obtaining module is used to obtain the second mark of the computing power node with the lowest level of resource utilization as the target second mark;

The second selection module is configured to select the computing power node with the highest reliability among the target second markings as the computing power node for resource scheduling when there are multiple target second markings.

By using the computing power node marking device provided in the embodiment of the present application, the first mark of the computing power node to be marked can be obtained first, and the first mark is used to mark the resource utilization level of the computing power node. Then, the second mark of the computing power node can be obtained, and the second mark is used to measure the credibility of the first mark. Afterwards, the first mark and the second mark are used as the scheduling marks of the computing power nodes. In this way, the computing power nodes are marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels. In this way, when scheduling computing power nodes, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

It should be noted that, when realizing the functions of the device provided by the above-mentioned embodiments, the division of the above-mentioned functional modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional modules according to the needs. The internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and the method embodiment provided by the above embodiment belong to the same idea, and the specific implementation process thereof is detailed in the method embodiment, and will not be repeated here.

An embodiment of the present application also provides an electronic device, see FIG. 4 , which is a structural block diagram of an electronic device provided by an embodiment of the present application. The electronic device includes a processor 401 and a memory 402, and a computer program is stored in the memory 402, and the computer program is loaded and executed by the processor 401 to implement any one of the methods for marking a computing power node as described above.

The processor 401 may include one or more processing cores, such as a 4-core processor, a 17-core processor, and the like. The processor 401 may be implemented in at least one hardware form among DSP (Digital Signal Processing, digital signal processing), FPGA (Field Programmable Gate Array, field programmable gate array), and PLA (Programmable Logic Array, programmable logic array). The processor 401 may also include a main processor and a coprocessor, the main processor is a processor for processing data in the wake-up state, and is also called a CPU (Central Processing Unit, central processing unit); Low-power processor for processing data in standby state. In some embodiments, the processor 401 may be integrated with a GPU, and the GPU is used for rendering and drawing the content that needs to be displayed on the display screen. In some embodiments, the processor 401 may further include an AI (Artificial Intelligence, artificial intelligence) processor, where the AI processor is configured to process computing operations related to machine learning.

Memory 402 may include one or more computer-readable storage media, which may be tangible and non-transitory. The memory 402 may also include high-speed random access memory and non-volatile memory, such as one or more magnetic disk storage devices and flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 402 stores a computer program, and the computer program is loaded and executed by the processor 401 to implement the above-mentioned computing power node marking method.

Those skilled in the art can understand that the structure shown in FIG. 4 does not constitute a limitation on the electronic device, and may include more or less components than shown in the figure, or combine certain components, or adopt different component arrangements.

Using the electronic device provided by the embodiment of the present application, the first mark of the computing power node to be marked can be obtained first, and the first mark is used to mark the resource utilization level of the computing power node. Then, the second mark of the computing power node can be obtained, and the second mark is used to measure the credibility of the first mark. Afterwards, the first mark and the second mark are used as the scheduling marks of the computing power nodes. In this way, the computing power nodes are marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels. In this way, when scheduling computing power nodes, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

The embodiment of the present application also provides a computer-readable storage medium, in which a computer program is stored, and the computer program is loaded and executed by a processor to implement any one of the above-mentioned computing power node labeling method embodiments. Methods.

Using the computer-readable storage medium provided by the embodiment of the present application, the first mark of the computing power node to be marked can be obtained first, and the first mark is used to mark the resource utilization level of the computing power node. Then, the second mark of the computing power node can be obtained, and the second mark is used to measure the credibility of the first mark. Afterwards, the first mark and the second mark are used as the scheduling marks of the computing power nodes. In this way, the computing power nodes are marked by using the quantized first mark and the second mark as labels, so as to have finer-grained labels. In this way, when scheduling computing power nodes, tasks can be accurately assigned to computing power nodes that meet the load conditions, and the time-consuming calculation of optimal computing power nodes can be reduced, and the utilization rate of computing power nodes can be improved.

Optionally, the computer-readable storage medium may include: ROM (Read-Only Memory, read-only memory), RAM (Random Access Memory, random access memory), SSD (Solid State Drives, solid-state hard disk) or optical discs, etc. Wherein, the random access memory may include ReRAM (Resistance Random Access Memory, resistive random access memory) and DRAM (Dynamic Random Access Memory, dynamic random access memory).

Wherein, the embodiments of the apparatus, electronic equipment, and computer-readable storage medium are the contents corresponding to the method embodiments, and related contents may refer to the method embodiments, and details are not repeated here.

It should be understood that the "plurality" mentioned herein refers to two or more than two. "And/or" describes the association relationship of associated objects, indicating that there may be three types of relationships, for example, A and/or B may indicate: A exists alone, A and B exist simultaneously, and B exists independently. The character "/" generally indicates that the contextual objects are an "or" relationship. In addition, the numbering of the steps described herein only exemplarily shows a possible sequence of execution among the steps. In some other embodiments, the above-mentioned steps may not be executed according to the order of the numbers, such as two different numbers The steps are executed at the same time, or two steps with different numbers are executed in the reverse order as shown in the illustration, which is not limited in this embodiment of the present application. The foregoing embodiments may also be combined arbitrarily, and the combination solutions will not be repeated here.

The above are only exemplary embodiments of the application, and are not intended to limit the application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the application shall be included in the protection of the application. within range.

Claims (10)

1. A method of marking a computing node, comprising:

obtaining a first mark of a computing node to be marked; the first mark is used for marking the resource utilization rate grade of the computing power node;

obtaining a second signature of the computing node; the second mark is used for measuring the credibility of the first mark;

and taking the first mark and the second mark as scheduling marks of the computing nodes.

2. The method according to claim 1, wherein the step of obtaining a first signature of the computing force node to be signed comprises:

and determining a formula based on the resource utilization rate level, and obtaining a first mark of the computing power node to be marked.

3. The method of claim 2, wherein the resource utilization level determination formula comprises:

wherein, label_1 represents a first Label of the computing node; the P is ₁ When M is represented by ₁ <v ₁ When the resource utilization rate level corresponding to the first mark is P ₁ The method comprises the steps of carrying out a first treatment on the surface of the The P is _i+1 Representing when v _i+1 >M ₁ ≥v _i When the resource utilization rate level corresponding to the first mark is P _i+1 The method comprises the steps of carrying out a first treatment on the surface of the The P is _n+1 When M is represented by ₁ ≥v _n When the resource utilization rate level corresponding to the first mark is P _n+1 ；

The M is ₁ Representing m sample numbers of the computing force node in a first sample durationThe average value of the data; the v is ₁ 、……、v _i 、……、v _n Representing a preset first grading threshold point, … …, an ith grading threshold point, … … and an nth grading threshold point; and i, n and m are positive integers.

4. The method of claim 1, wherein the step of obtaining the second signature of the computing node comprises:

and obtaining a second mark of the computing force node based on a variation coefficient calculation formula.

5. The method of claim 4, wherein the coefficient of variation calculation formula comprises:

wherein, label_2 represents a second Label of the computing node; the CV represents the coefficient of variation; the M is ₂ Representing the average value of p sampling data of the computing force node in a second sampling time period; wherein the second sampling time period is longer than the first sampling time period; sampling m pieces of sampling data in a first sampling time length, wherein p is greater than m; the x is _k Representing the kth sampling data in the p sampling data; and p, k and m are positive integers.

6. The method of claim 1, wherein the step of using the first and second labels as dispatch labels for the computing nodes comprises:

updating the first mark according to the first sampling duration as an updating period;

updating the second mark according to the second sampling duration as an updating period; wherein the second sampling time period is longer than the first sampling time period;

and taking the updated first mark and the updated second mark as scheduling marks of the computing nodes.

7. The method of any of claims 1-6, further comprising, after the first marker and the second marker are used as scheduling markers for the computing node:

when a scheduling instruction for the power computing node is received, acquiring a first mark of each power computing node with a scheduling mark;

selecting the computing power node with the lowest resource utilization rate level in each obtained first mark;

acquiring a second mark of the computing power node with the lowest resource utilization rate level as a target second mark;

and when a plurality of target second marks exist, selecting the computing node with the highest credibility in the target second marks as the computing node for resource scheduling.

8. A computing force node marking device, comprising:

the first acquisition module is used for acquiring a first mark of the computing power node to be marked; the first mark is used for marking the resource utilization rate grade of the computing power node;

a second acquisition module for acquiring a second signature of the computing force node; the second mark is used for measuring the credibility of the first mark;

and the marking module is used for taking the first mark and the second mark as the scheduling marks of the computing nodes.

9. An electronic device comprising a processor and a memory, the memory having stored therein a computer program that is loaded and executed by the processor to implement the method of any of claims 1-7.

10. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a computer program, which is loaded and executed by a processor to implement the method of any of claims 1 to 7.

Images