A computing power network container migration method based on probability language and resource similarity

Abstract

The application discloses a kind of power network container migration methods based on probability language and resource similarity, belong to the field of power network, including: according to container resource average demand quantity calculation container demand semantic probability, according to resource actual demand and declaration demand, construct multidimensional resource judgment matrix and calculate container demand semantic vector, calculate container resource demand probability language quantization vector.Power node resource semantic probability matrix is calculated, and the probability language matrix of the available resource of power node is constructed, and the probability language quantization matrix of the available resource of power node is calculated.According to container resource demand probability language quantization vector and the probability language quantization matrix of the available resource of power node, resource similarity vector is calculated, and the most similar power node is selected to migrate container.The application effectively quantizes the similarity of multidimensional data between container resource demand and power node resource condition, ensures migration result, meets the demand of user container and improves the stability and efficiency of power network operation.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of computing power network, and particularly relates to a computing power network container migration method based on probability language and resource similarity. BACKGROUND

[0002] With the development of the digital era, cloud computing concentrates large-scale computing capacity on a single device by establishing a large data center, provides computing services for users using virtualization technology, and efficiently utilizes computing resources and saves energy consumption using virtual machine migration technology. However, many specific application scenarios have specific requirements for the network throughput, concurrent computing and storage of the data center. On the other hand, the emergence of emerging technologies such as 5G / B5G brings massive data to the network edge, and accelerates the diffusion of computing power from a few data centers to the network edge and even terminal devices. In order to solve the problems of lack of effective coordination mechanism between computing nodes, imperfect computing task allocation and scheduling mechanism, and low resource utilization, the concept of computing power network emerges as the times require. The computing power network connects the computing power resources of dispersed computing nodes in the form of a computing power resource pool spanning the entire network, and provides flexible and high-quality computing services for various applications. Therefore, in view of the increase of resource demand dimensions in the computing power network, it is of great significance to select the appropriate computing power nodes for container migration to improve the coordination ability between computing power nodes and improve the resource utilization.

[0003] The probability language term set is a semantic set composed of a semantic set and a probability corresponding to the semantic set. It can not only express the hesitant situation of experts in scheme evaluation, but also show the preference degree of experts. Compared with other semantic evaluation methods, it can avoid the loss of expert preference information in the evaluation process. However, the term set depends on expert scoring to extract the importance of each index, and cannot directly obtain index information in the computing power network. There is a lack of quantitative method for the preferences of computing power network users and the service capabilities of computing power nodes. In the index of computing power network, the hesitation degree and uncertainty of different computing power network user term sets are different, so it is difficult to accurately and quickly measure and compare the matching degree of computing power user demand preferences and computing power service capabilities in the decision-making process. SUMMARY

[0004] This invention is a container migration method for computing power networks based on probabilistic language and resource similarity. The main ideas of this method include: calculating the semantic probability of container demand based on the average resource demand of containers; constructing a multi-dimensional resource judgment matrix and calculating a container demand semantic vector based on actual and declared resource demands; constructing a container resource demand probabilistic language vector; and calculating a container resource demand probabilistic language quantization vector. Next, calculating the computing power node resource semantic probability matrix, predicting the load balancing degree of each computing power node after migration as the computing power node resource semantic vector, constructing a computing power node available resource probabilistic language matrix, and calculating a computing power node available resource probabilistic language quantization matrix. Finally, calculating a resource similarity vector based on the container resource demand probabilistic language quantization vector and the computing power node available resource probabilistic language quantization matrix, and selecting the computing power node with the highest similarity for container migration.

[0005] The technical solution adopted in this invention is: a method for migrating computing network containers based on probabilistic language and resource similarity, the steps of which are as follows:

[0006] Step 1: Obtain the resource utilization list of computing nodes: UV = {U1, U2, ..., U i ,…,U n}, where U i ={u i1 ,u i2 ,u i3 ,u i4},u i1 ,u i2 ,u i3 ,u i4 These represent the utilization rates of the processor, memory, bandwidth, and disk of the i-th computing node, respectively. The total resource list of computing nodes is obtained as MV = {M1, M2, ..., M...}. i ,…,M n}, where M i ={m i1 ,m i2 ,m i3 ,m i4},m i1 ,m i2 ,m i3 ,m i4 These represent the total resources of the i-th computing node, including processor, memory, bandwidth, and disk.

[0007] Obtain the list of container resource requirements to be migrated: NV = {n1, n2, n3, n4}, where n1, n2, n3, and n4 represent the declared requirements for container processors, memory, bandwidth, and disk, respectively. Obtain the list of historical container requirement datasets: DATAS = {DATA1, DATA2, ..., DATA...} i ,…,DATA n}, where DATA i = {d i1 , d i2 , d i3 , d i4}, where d i1 , d i2 , d i3 , d i4 denote the actual demand of the container's processor, memory, bandwidth, and disk at the i-th time, respectively, and d n1 , d n2 , d n3 , d n4 denote the actual demand of the container's processor, memory, bandwidth, and disk at the current time, respectively.

[0008] Step 2: Calculate the container resource average demand vector HV = {h1,..., h i ,..., h4} according to the following formula, where h i denotes the average demand of the i-th resource of the container, and d ji denotes the demand of the i-th resource of the container at the j-th time.

[0009]

[0010] Calculate the container demand semantic probability vector NSV = {ns1,..., ns i ,..., ns4} according to the following formula, where ns i denotes the demand semantic probability of the i-th resource.

[0011]

[0012] Step 3: Calculate the container multi-dimensional resource judgment matrix PT = (pt ij ) 4×4 , where pt ij denotes the demand utilization ratio of the i-th resource to the j-th resource.

[0013] pt ij = h i *n j / h j *n i

[0014] Calculate the container demand semantic vector RNW = {rn1,..., rn i ,..., rn4} according to the following formula, where rn i denotes the demand semantic value of the i-th resource

[0015]

[0016] Step 4: Calculate the container resource demand probability linguistic vector A = (a1, ..., a2) i ,…,a4), where the probability language a i A probabilistic language representing the demand for the i-th type of resource:

[0017]

[0018] Where d ni ns represents the actual demand for the i-th type of resource in the container. i Let rn represent the semantic probability of the demand for the i-th type of resource. i This represents the semantic value of the resource requirement for the i-th class.

[0019] Step 5: Based on the container demand semantic vector RNW and the container demand semantic probability vector NSV, calculate the container resource demand probabilistic linguistic quantization vector NA = (na1, ..., na2) i ,…,na4), where na i The probabilistic linguistic quantization value representing the demand for the i-th type of resource:

[0020]

[0021] Step 6: Calculate the average available resource vector V = (v1, ..., v) of the computing power nodes according to the following formula. i ,…,v4), where v i This represents the average available value of the i-th type of resource on the computing node.

[0022]

[0023] The semantic probability matrix of computing node resources, WT, is calculated according to the following formula: WT = (wt ij ) n×4 wt ij This represents the semantic probability of the j-th type of resource on the i-th computing node.

[0024] wt ij =m ij *(1-u ij ) / v j

[0025] Step 7: Calculate the semantic vector of computing node resources L = (l1, l2, ..., l) according to the following formula. i ,…,l n ), where l i This represents the resource semantic value of the i-th computing node.

[0026]

[0027] Step 8: Calculate the available resource probability language matrix CN = (CN1, CN2, ..., CN) for computing power nodes.i ,…, CN n ), wherein CN i = (b i1 , b i2 , b i3 , b i4 ), represents the available resource probability language vector of the i-th computing power node, wherein b ij represents the probability language of the j-th type resource of the i-th computing power node:

[0028]

[0029] wherein l i represents the semantic value of the i-th computing power node resource, m ij represents the total amount of the j-th type resource of the i-th computing power node, u ij represents the current utilization rate of the j-th type resource of the i-th computing power node, and wt ij represents the semantic probability of the j-th type resource of the i-th computing power node.

[0030] Step 9: According to the computing power node resource semantic probability matrix WT and the computing power node resource semantic vector L, the available resource probability language quantization matrix NCN = (NCN1, NCN2, …, NCN i ,…, NCN n ) of the computing power node is calculated, wherein NCN i = (nb i1 , nb i2 , nb i3 , nb i4 ), represents the available resource probability language quantization vector of the i-th computing power node, wherein nb ij represents the probability language quantization value of the j-th type available resource of the i-th computing power node:

[0031]

[0032] Step 10: According to the container resource demand probability language quantization vector NA and the computing power node available resource probability language quantization matrix NCN, the similarity vector RES = {res1, res2, …, res i ,…, res n} is calculated according to the following formula, wherein res i represents the similarity of the available resource and the container demand of the i-th computing power node.

[0033]

[0034] Step 11: Traverse the similarity vector RES = {res1, res2, …, res n}, and get the maximum value resmax , output the corresponding computing power node number max as the destination of container migration.

[0035] Compared with the prior art, the present application has the beneficial effects that:

[0036] 1、The present application firstly quantifies the sensitivity degree of resource demand users, calculates the stability of container resource demand as the semantic probability of the container, and quantifies the difference degree between the subjective resource demand of the user container and the actual one by calculating the container demand semantic vector. Then, the container resource demand probability language vector is constructed, the container resource demand probability language quantization vector is calculated, the difference between the subjective demand of the user container and the actual one is considered, the sensitivity degree of the user container to the resource is retained, and the importance of the resource demand of the user container is effectively quantified.

[0037] 2、The present application calculates the resource semantic probability matrix of the computing power node, highlights the characteristics of the resource distribution of each computing power node, calculates the resource semantic vector of each computing power node after migration, quantifies the load balancing of each computing power node after migration, constructs the available resource probability language matrix of the computing power node, calculates the available resource probability language quantization matrix of the computing power node, considers the current resource situation of each node, combines the load balancing degree of the node after migration, effectively quantifies the service ability of the current computing power node, reduces the probability of subsequent container migration, and improves the resource utilization rate of the computing power network.

[0038] 3、The present application calculates the resource similarity vector according to the container resource demand probability language quantization vector and the available resource probability language quantization matrix of the computing power node, effectively quantizes the similarity of multi-dimensional data between the container resource demand and the resource situation of the computing power node, and ensures that the migration result meets the demand of the user container and improves the stability and efficiency of the operation of the computing power network. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 is the flowchart of the present application.

[0040] Figure 2 is the application scenario diagram of the present application. DETAILED DESCRIPTION

[0041] The present application will be further explained in combination with the drawings and specific embodiments.

[0042] Figure 2 is the actual application scenario diagram of the present application. The computing power network scheduling center periodically collects the running information of each computing power node, wherein the overloaded computing power node reports the container list to be migrated and the historical running data, and the other computing power nodes report the node load condition, which is shown in detail in step 1.

[0043] For each container in the reported list of containers to be migrated, calculate the corresponding probability language quantization vector. According to the available computing power node load information, calculate the corresponding probability language quantization vector. The resource similarity of the to-be-migrated container and each computing power node is calculated, and the target node is selected, which is described in detail in step 2-11.

[0044] The technical solution of the present application: a computing power network container migration method based on probability language and resource similarity, comprising the following steps:

[0045] Step 1: Obtain the resource utilization rate list UV = {U1, U2, …, U i ,…,U n} of the computing power node, wherein U i ={u i1 ,u i2 ,u i3 ,u i4},u i1 ,u i2 ,u i3 ,u i4 respectively represent the utilization rate of the processor, memory, bandwidth and disk of the i-th computing power node. Obtain the resource total amount list MV = {M1, M2, …, M i ,…,M n} of the computing power node, wherein M i ={m i1 ,m i2 ,m i3 ,m i4},m i1 ,m i2 ,m i3 ,m i4 respectively represent the total amount of resources of the processor, memory, bandwidth and disk of the i-th computing power node.

[0046] Obtain the container resource demand list NV = {n1, n2, n3, n4} to be migrated, wherein n1, n2, n3, n4 respectively represent the declared demand amount of the container processor, memory, bandwidth and disk. Obtain the container historical demand data set list DATAS = {DATA1, DATA2, …, DATA i ,…,DATA n}, wherein DATA i ={d i1 ,d i2 ,d i3 ,d i4}, wherein d i1 ,d i2 ,d i3 ,d i4 respectively represent the actual demand amount of the processor, memory, bandwidth and disk of the container at the i-th moment, d n1,d n2 ,d n3 ,d n4 These represent the actual current requirements of the container for processor, memory, bandwidth, and disk space, respectively.

[0047] Step 2: Calculate the average demand vector of container resources HV = {h1,…,h} according to the following formula. i ,…,h4}, where h i d represents the average demand for the i-th type of resource in the container. ji This represents the demand for the i-th type of resource in the container at time j.

[0048]

[0049] The container demand semantic probability vector NSV is calculated according to the following formula: NSV = {ns1, ..., ns} i ,…,ns4}, where ns i This represents the semantic probability of the demand for the i-th type of resource.

[0050]

[0051] Step 3: Calculate the container's multidimensional resource judgment matrix PT = (pt ij ) 4×4 , of which pt ij This represents the ratio of the demand utilization rate of resource type i to that of resource type j.

[0052] pt ij =h i *n j / h j *n i

[0053] The container requirement semantic vector RNW = {rn1,…,rn} is calculated according to the following formula. i ,…,rn4}, where rn i This represents the semantic value of the resource requirement for the i-th class.

[0054]

[0055] Step 4: Based on the container resource demand stability and weight vector, use probabilistic language to calculate the container resource demand probabilistic language vector A = (a1, ..., a2) i The following formula, using probabilistic language, can not only express the difference between a user's subjective resource needs and actual needs, but also demonstrate the user's sensitivity to different resources. Wherein, the probabilistic language a i Represented as:

[0056]

[0057] Where ai Probability language representing the i-th type of resource requirement, d ni Actual requirement amount of the i-th type of resource of the container, ns i Probability of the i-th type of resource requirement semantics, rn i The i-th type of resource requirement semantics value.

[0058] Step 5: According to the container requirement semantics vector RNW and the container requirement semantics probability vector NSV, the container resource requirement probability language quantization vector NA=(na1, …, na i , na4), the following formula takes into account the subjectivity and objectivity of the weight, combines the subjective and objective situations and sensitivity of the user container requirement, and effectively quantizes the importance of the user container resource requirement, wherein na i Probability language quantization value of the i-th type of resource requirement:

[0059]

[0060] Step 6: According to the following formula, calculate the average available resource vector V=(v1, …, v i , v4) of the computing power node, wherein v i Average available value of the i-th type of resource of the computing power node.

[0061]

[0062] According to the following formula, calculate the computing power node resource semantics probability matrix WT=(wt ij ) n×4 , wt ij The i-th type of resource semantics probability of the j-th computing power node.

[0063] wt ij = m ij *(1-u ij ) / v j

[0064] Step 7: According to the following formula, calculate the computing power node resource semantics vector L=(l1, l2, …, l i , …, l n ), wherein l i The i-th type of resource semantics value of the computing power node.

[0065]

[0066] Step 8: According to the load balancing degree and available resource weight of the computing power node, use the probability language to calculate the available resource probability language matrix CN=(CN1, CN2, …, CN i , CN n ), CNi = (b i1 , b i2 , b i3 , b i4 ), represents the available resource probability language vector of the i-th computing node, the following formula uses probability language, not only expresses the characteristics of the resource distribution of each computing node, but also reflects the load of each computing node after migration, wherein:

[0067]

[0068] wherein b ij represents the probability language of the j-th type resource of the i-th computing node, l i represents the resource semantic value of the i-th computing node, m ij represents the total amount of the j-th type resource of the i-th computing node, u ij represents the current utilization rate of the j-th type resource of the i-th computing node, wt ij represents the semantic probability of the j-th type resource of the i-th computing node.

[0069] Step 9: According to the computing node resource semantic probability matrix WT and the computing node resource semantic vector L, the available resource probability language quantization matrix of the computing node NCN = (NCN1, NCN2, …, NCN i , …, NCN n ) is calculated, NCN i = (nb i1 , nb i2 , nb i3 , nb i4 ), represents the available resource probability language quantization vector of the i-th computing node, the following formula combines the current resource distribution of each computing node and the load balancing after migration, and effectively quantizes the service capability of the current each computing node, wherein nb ij represents the probability language quantization value of the j-th available resource of the i-th computing node:

[0070]

[0071] Step 10: According to the container resource demand probability language quantization vector NA and the computing node available resource probability language quantization matrix NCN, the similarity vector RES = {res1, res2, …, res i , …, res n} is calculated according to the following formula, wherein res i represents the similarity of the available resource of the i-th computing node and the container demand. The vector effectively quantizes the similarity of the multi-dimensional data between the container resource demand and the computing node resource probability language, simplifies the comparison process,

[0072]

[0073] Step 11: traverse the similarity vector RES = {res1, res2, …, res n} to get the maximum value res max , and output the corresponding computing power node number max as the destination of the container migration.

[0074] The above description is only the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for migrating computing network containers based on probabilistic language and resource similarity, characterized in that, Includes the following steps: Step 1: Obtain the list of computing node resource utilization (UV), the list of total computing node resources (MV), the list of container resource requirements to be migrated (NV), and the list of container historical requirement datasets (DATAS). Step 2: Calculate the container resource average demand vector HV and the container demand semantic probability vector NSV; Step 3: Calculate the container multidimensional resource judgment matrix PT and the container demand semantic vector RNW; Step 4: Calculate the container resource demand probability linguistic vector A = (a1, ..., a2) i ,…,a4), where probability language Where d ni ns represents the actual demand for the i-th type of resource in the container. i Let rn represent the semantic probability of the demand for the i-th type of resource. i This represents the semantic value of the resource requirement for the i-th class; Step 5: Calculate the container resource requirement probability linguistic quantization vector NA based on the container requirement semantic vector RNW and the container requirement semantic probability vector NSV; Container resource requirement probabilistic language quantization vector NA = (na1, ..., na) i , ..., na4), where: Step 6: Calculate the average available resource vector V of computing power nodes and the semantic probability matrix WT of computing power node resources; Step 7: Calculate the semantic vector L of the computing node resources according to the following formula; The semantic vector of computing node resources is L = (l1, l2, ..., l i , ..., l n The calculation formula is as follows: Step 8: Calculate the probability language matrix CN of available resources for computing power nodes; Step 9: Calculate the available resource probabilistic language quantization matrix NCN of the computing node based on the computing node resource semantic probability matrix WT and the computing node resource semantic vector L; The available resource probabilistic language quantization matrix for computing power nodes is NCN = (NCN1, NCN2, ..., NCN). i ..., NCN n ), of which NCN i =(nb i1 nb i2 nb i3 nb i4 ), representing the probabilistic linguistic quantization vector of available resources for the i-th computing node, where, Step 10: Calculate the similarity vector RES based on the container resource demand probabilistic linguistic quantization vector NA and the computing node available resource probabilistic linguistic quantization matrix NCN; Step 11: Traverse the similarity vector RES = {res1,res2,...,res} n }, to obtain the maximum value res max Output the corresponding computing node number max as the destination for container migration.

2. The computing power network container migration method based on probabilistic language and resource similarity according to claim 1, characterized in that, The computing node resource utilization list UV = {U1, U2, ..., U} in step 1 is as follows: i ,…,U n }, where U i ={u i1 ,u i2 ,u i3 ,u i4 },u i1 ,u i2 ,u i3 ,u i4 Let M1, M2, ..., Mn represent the utilization rates of the processor, memory, bandwidth, and disk of the i-th computing node, respectively. Obtain the total resource list of the computing nodes: MV = {M1, M2, ..., Mn}. i ,…,M n }, where M i ={m i1 ,m i2 ,m i3 ,m i4 },m i1 ,m i2 ,m i3 ,m i4 These represent the total resources of the i-th computing node, including processor, memory, bandwidth, and disk, respectively. In step 1, the container resource requirement list NV = {n1, n2, n3, n4} to be migrated, where n1, n2, n3, and n4 represent the declared requirements for container processors, memory, bandwidth, and disk, respectively. The historical container requirement dataset list DATAS = {DATA1, DATA2, ..., DATA...} is then obtained. i ,…,DATA n }, DATA i ={d i1 ,d i2 ,d i3 ,d i4 }, where d i1 ,d i2 ,d i3 ,d i4 Let d represent the actual requirements of the container in terms of processor, memory, bandwidth, and disk space at time i. n1 ,d n2 ,d n3 ,d n4 These represent the actual current requirements of the container for processor, memory, bandwidth, and disk space, respectively.

3. The computing power network container migration method based on probabilistic language and resource similarity according to claim 1, characterized in that, In step 2, the average demand vector for container resources HV = {h1,…,h} i The formula for calculating h4 is as follows: The container demand semantic probability vector NSV in step 2 is {ns1,…,ns}. i The formula for calculating ns4 is as follows:

4. The computing power network container migration method based on probabilistic language and resource similarity according to claim 1, characterized in that, The container multidimensional resource judgment matrix PT = (pt) in step 3 ij ) 4×4 ,in pt ij =h i *n j / h j *n i The container requirement semantic vector RNW = {rn1,…,rn} in step 3 is... i The formula for calculating ,…,rn4} is as follows:

5. The computing power network container migration method based on probabilistic language and resource similarity according to claim 1, characterized in that, In step 6, the average available resource vector of computing nodes is V = (v1, ..., v1). i The calculation formula for (v4, ..., v4) is as follows: The semantic probability matrix WT of computing node resources in step 6 is (wt) ij ) n×4 ,in wt ij =m ij *(1-u ij ) / v j 。 6. The computing power network container migration method based on probabilistic language and resource similarity according to claim 1, characterized in that, The computing nodes in step 8 can be represented by the resource probability language matrix CN = (CN1, CN2, ..., CN). i , ..., CN n ), where CN i =(b i1 b i2 b i3 b i4 ), representing the available resource probability language vector of the i-th computing node, where: Among them l i Let m represent the resource semantic value of the i-th computing node. ij u represents the total amount of the j-th type of resources in the i-th computing node. ij wt represents the current utilization rate of the j-th type of resource on the i-th computing node. ij This represents the semantic probability of the j-th type of resource on the i-th computing node.

7. The computing power network container migration method based on probabilistic language and resource similarity according to claim 1, characterized in that, In step 10, the similarity vector RES = {res1,res2,…,res} i ,…,res n },in

Images