CN111338757A - Energy-optimized virtual machine deployment method and system - Google Patents

Abstract

An energy optimized virtual machine deployment method and system are disclosed. The method comprises the following steps: step 1: determining a virtual machine deployment target and constraint conditions thereof; step 2: determining a deployment solution calculation initialization parameter; and step 3: performing population initialization to enable the deployment solution of the goblet sea squirts to correspond to the deployment solution of the virtual machines one by one; and 4, step 4: calculating the individual fitness of each bottle sea squirt in the initialized population according to the deployment target of the virtual machine and the constraint conditions thereof; and 5: determining a food source, a leader and a follower to obtain an updated population; step 6: calculating the fitness of each individual of the ascidians in the updated population, and determining the food source; and 7: and 5-6, outputting the position of the food source in the current population as the final optimal solution for deploying the virtual machine. According to the invention, exploration behaviors and development behaviors in an iterative process are balanced through a group optimization model of the goblet sea squirts, and the deployment effect of the energy-optimized virtual machine is realized while the global property and diversity are ensured.

Description

Energy-optimized virtual machine deployment method and system

technical field

The present invention relates to the fields of cloud computing and intelligent swarm algorithms, and more particularly, to an energy-optimized virtual machine deployment method and system.

Background technique

Cloud computing can overcome the shortcomings of localized computing methods of traditional computing models for data center owners through its flexibility, security, high scalability, and quality of service assurance. It can configure a large number of heterogeneous servers of different types, and a large number of virtual machines can be further deployed on the server using virtualization technology, which is beneficial to the isolation of applications when virtual machines provide services. Virtual machines also have different load types and different requirements for resources in each dimension, which may lead to conflicts between resource utilization and high host energy consumption. The high energy consumption of the host will lead to an increase in cloud computing operating costs and adversely affect the environment. High energy consumption is closely related to the utilization of host resources. Therefore, how to deploy energy-efficient virtual machines will be a problem that cloud computing must solve.

Intelligent swarm algorithm is a typical algorithm for solving optimization problems. Among them, particle swarm algorithm, genetic algorithm, ant colony algorithm, bee colony algorithm and gravitational search algorithm have all been applied in the virtual machine deployment problem in cloud computing environment. As one of the optimization goals, the energy consumption problem has achieved many phased results. However, due to the inherent limitations of such meta-heuristic algorithms, such as falling into local optimality and relying on too many initial parameters, energy consumption-based virtual machine deployment optimization The problem already has certain limitations, and a meta-heuristic algorithm with better performance is needed to optimize the solution to this problem. Therefore, it is necessary to develop an energy-optimized virtual machine deployment method and system.

The information disclosed in this Background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.

SUMMARY OF THE INVENTION

The invention proposes an energy-optimized virtual machine deployment method and system, which can fully balance the exploration behavior and development behavior in the iterative process of the salp group through the group optimization model of salp, and ensure the overall and diversity of the salp. At the same time, an energy-optimized virtual machine deployment effect is achieved.

According to an aspect of the present invention, an energy-optimized virtual machine deployment method is proposed. The method may include: step 1: determining the virtual machine deployment target and its constraints; step 2: determining the deployment solution and calculating initialization parameters; step 3: performing population initialization through the salps algorithm, so that the deployment solution between the individual salps and the virtual machine is solved. One-to-one correspondence; Step 4: Calculate the fitness of each Salina individual in the initialization population according to the virtual machine deployment target and its constraints; Step 5: Determine the food source, leader and follower according to the fitness, and then determine the leader The position of the follower and the position of the follower are obtained, and the updated population is obtained; Step 6: Calculate the fitness of each salp individual in the updated population, determine the individual with the largest fitness value, and judge whether the fitness value of the individual salp is not. is greater than the fitness value of the food source, if so, use the individual salps as a new food source; if not, keep the original food source; Step 7: Determine whether the current iteration number is less than the maximum iteration number, if so, repeat Step 5-6, if not, output the food source position in the current population as the final optimal solution for virtual machine deployment.

Preferably, the deployment solution calculation initialization parameters include: maximum number of iterations T _max , virtual machine set V, physical host set H, host resource capability vector, virtual machine resource request vector, host full load power consumption and idle power consumption.

Preferably, the virtual machine deployment target is:

Among them, P _i,full represents the power consumption when the host hi is in full load state, P _{i ,idle} represents the power consumption when the host hi is in the idle state, P _i represents the current power consumption of the host _hi , U _{i ,CPU} is the CPU utilization of host _hi .

Preferably, the constraints are:

Load _i,CPU ≤C _i,CPU ,i=1,2,...,m (2)

Load _i,MEM ≤C _i,MEM ,i=1,2,...,m (3)

Load _i,DISK ≤C _i,DISK ,i=1,2,...,m (4)

Load _i,NETW ≤C _i,NETW ,i=1,2,...,m (5)

Among them, C _i,CPU represents the CPU capability of the physical host hi, C _{i ,MEM} represents the memory capability of the physical host hi, C _{i ,DISK} represents the storage capability of the physical host hi, and C _{i ,NETW} represents the physical host _hi is the network bandwidth capability, Load _i,CPU represents the overall CPU load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i , MEM} represents the overall memory load of the physical host _hi .

Preferably, the step (3) includes: randomizing the position of the individual salps by formula (6) to obtain an initialization population:

表示向上取整。Among them, q=1,2,...,N, N represents the size of the salps population, j=1,2,...,n, n represents the total number of virtual machines, c ₄ represents a random number in the interval [0,1] , ub _j,max represents the upper limit of the jth dimensional space, lb _j,min represents the lower limit of the jth dimensional space, x _1,j represents the position of the leader on the jth dimensional space, x _q,j represents the position of the follower on the jth dimension space,

Indicates rounded up.

Preferably, the fitness is calculated by formula (7):

Among them, fitness is the fitness.

Preferably, according to the fitness, the salp individual with the largest fitness value is used as the current food source position, and the first salp individual among the N-1 salp individuals remaining in the order outside the food source is used as the leader, Make all the remaining N-2 salps as followers.

Preferably, the leader position is determined by formula (8):

t表示当前迭代次数，T_max表示方法的最大迭代次数。Among them, x _1,j represents the position of the leader salps in the jth dimension space, F _j represents the position of the food source in the jth dimension space, ub _j,max represents the upper limit of the jth dimension space, lb _{j ,min} represents the lower limit of the jth dimension space, c ₂ and c ₃ represent random numbers in the interval [0,1], c ₁ is the convergence factor,

t represents the current number of iterations, and _Tmax represents the maximum number of iterations of the method.

Preferably, the follower position is determined by formula (9):

Among them, q'≥2, x _q',j represents the position of the follower on the jth dimension, △t represents the time, v ₀ represents the initial velocity of the follower, and the acceleration a=(v _final -v ₀ )/△ t, where v _final =x _q'-1,j -x _q',j /Δt, x _q'-1,j represents the position of the q'-1th salps on the jth dimension space.

According to another aspect of the present invention, an energy-optimized virtual machine deployment system is proposed, characterized in that the system includes: a memory storing computer-executable instructions; The computer can execute the instructions and perform the following steps: Step 1: Determine the virtual machine deployment target and its constraints; Step 2: Determine the deployment solution and calculate the initialization parameters; Step 3: Perform population initialization through the salp algorithm, so that the salp individual and the virtual machine are initialized One-to-one correspondence of the deployment solutions; Step 4: According to the virtual machine deployment target and its constraints, calculate the fitness of each Salina individual in the initialization population; Step 5: According to the fitness, determine the food source, leader and follower, Then determine the leader position and the follower position, and obtain the updated population; Step 6: Calculate the fitness of each salp individual in the updated population, determine the salp individual with the largest fitness value, and judge the adaptation of the salp individual Whether the degree value is greater than the fitness value of the food source, if so, use the individual salps as a new food source; if not, keep the original food source; Step 7: Determine whether the current number of iterations is less than the maximum number of iterations, if so , then repeat steps 5-6, if not, output the food source position in the current population as the final optimal solution for virtual machine deployment.

Preferably, the deployment solution calculation initialization parameters include: maximum number of iterations T _max , virtual machine set V, physical host set H, host resource capability vector, virtual machine resource request vector, host full load power consumption and idle power consumption.

Preferably, the virtual machine deployment target is:

Among them, P _i,full represents the power consumption when the host hi is in full load state, P _{i ,idle} represents the power consumption when the host hi is in the idle state, P _i represents the current power consumption of the host _hi , U _{i ,CPU} is the CPU utilization of host _hi .

Preferably, the constraints are:

Load _i,CPU ≤C _i,CPU ,i=1,2,...,m (2)

Load _i,MEM ≤C _i,MEM ,i=1,2,...,m (3)

Load _i,DISK ≤C _i,DISK ,i=1,2,...,m (4)

Load _i,NETW ≤C _i,NETW ,i=1,2,...,m (5)

Among them, C _i,CPU represents the CPU capability of the physical host hi, C _{i ,MEM} represents the memory capability of the physical host hi, C _{i ,DISK} represents the storage capability of the physical host hi, and C _{i ,NETW} represents the physical host _hi is the network bandwidth capability, Load _i,CPU represents the overall CPU load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i , MEM} represents the overall memory load of the physical host _hi .

Preferably, the step (3) includes: randomizing the position of the individual salps by formula (6) to obtain an initialization population:

表示向上取整。Among them, q=1,2,...,N, N represents the size of the salps population, j=1,2,...,n, n represents the total number of virtual machines, c ₄ represents a random number in the interval [0,1] , ub _j,max represents the upper limit of the jth dimensional space, lb _j,min represents the lower limit of the jth dimensional space, x _1,j represents the position of the leader on the jth dimensional space, x _q,j represents the position of the follower on the jth dimension space,

Indicates rounded up.

Preferably, the fitness is calculated by formula (7):

Among them, fitness is the fitness.

Preferably, according to the fitness, the salp individual with the largest fitness value is used as the current food source position, and the first salp individual among the N-1 salp individuals remaining in the order outside the food source is used as the leader, Make all the remaining N-2 salps as followers.

Preferably, the leader position is determined by formula (8):

t表示当前迭代次数，T_max表示方法的最大迭代次数。Among them, x _1,j represents the position of the leader salps in the jth dimension space, F _j represents the position of the food source in the jth dimension space, ub _j,max represents the upper limit of the jth dimension space, lb _{j ,min} represents the lower limit of the jth dimension space, c ₂ and c ₃ represent random numbers in the interval [0,1], c ₁ is the convergence factor,

t represents the current number of iterations, and _Tmax represents the maximum number of iterations of the method.

Preferably, the follower position is determined by formula (9):

Among them, q'≥2, x _q',j represents the position of the follower on the jth dimension, △t represents the time, v ₀ represents the initial velocity of the follower, and the acceleration a=(v _final -v ₀ )/△ t, where v _final =x _q'-1,j -x _q',j /Δt, x _q'-1,j represents the position of the q'-1th salps on the jth dimension space.

The method and apparatus of the present invention have other features and advantages that will be apparent from, or will be apparent from, the accompanying drawings and the following detailed description incorporated herein. The detailed description is set forth in the detailed description, which together with the detailed description serve to explain certain principles of the invention.

Description of drawings

The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of the exemplary embodiments of the present invention taken in conjunction with the accompanying drawings, wherein the same reference numerals generally refer to the exemplary embodiments of the present invention. same parts.

FIG. 1 shows a flowchart of steps of an energy-optimized virtual machine deployment method according to the present invention.

Detailed ways

The present invention will be described in more detail below with reference to the accompanying drawings. While preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

FIG. 1 shows a flowchart of steps of an energy-optimized virtual machine deployment method according to the present invention.

In this embodiment, the energy-optimized virtual machine deployment method according to the present invention may include: step 1: determining the virtual machine deployment target and its constraints; step 2: determining the deployment solution and calculating initialization parameters; step 3: using the salps algorithm Perform population initialization to make the individual salps correspond to the deployment solutions of the virtual machine one-to-one; Step 4: Calculate the fitness of each salp individual in the initialization population according to the virtual machine deployment target and its constraints; Step 5: According to the fitness , determine the food source, leader and follower, and then determine the position of the leader and the follower, and obtain the updated population; Step 6: Calculate the fitness of each individual in the updated population, and determine the individual with the largest fitness value. , determine whether the fitness value of the individual salps is greater than the fitness value of the food source, if so, use the individual salps as a new food source; if not, keep the original food source; Step 7: Determine whether the current iteration number is less than Maximum number of iterations, if yes, repeat steps 5-6, if no, output the food source position in the current population as the final optimal solution for virtual machine deployment.

In one example, the deployment decomputation initialization parameters include: the maximum number of iterations T _max , the virtual machine set V, the physical host set H, the resource capability vector of the host, the resource request vector of the virtual machine, the full load power consumption and the idle power consumption of the host .

In one example, the virtual machine deployment targets are:

Among them, P _i,full represents the power consumption when the host hi is in full load state, P _{i ,idle} represents the power consumption when the host hi is in the idle state, P _i represents the current power consumption of the host _hi , U _{i ,CPU} is the CPU utilization of host _hi .

In one example, the constraints are:

Load _i,CPU ≤C _i,CPU ,i=1,2,...,m (2)

Load _i,MEM ≤C _i,MEM ,i=1,2,...,m (3)

Load _i,DISK ≤C _i,DISK ,i=1,2,...,m (4)

Load _i,NETW ≤C _i,NETW ,i=1,2,...,m (5)

Among them, C _i,CPU represents the CPU capability of the physical host hi, C _{i ,MEM} represents the memory capability of the physical host hi, C _{i ,DISK} represents the storage capability of the physical host hi, and C _{i ,NETW} represents the physical host _hi is the network bandwidth capability, Load _i,CPU represents the overall CPU load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i , MEM} represents the overall memory load of the physical host _hi .

In one example, step (3) includes: randomizing the position of the individual salps by formula (6) to obtain an initialization population:

表示向上取整。Among them, q=1,2,...,N, N represents the size of the salps population, j=1,2,...,n, n represents the total number of virtual machines, c ₄ represents a random number in the interval [0,1] , ub _j,max represents the upper limit of the jth dimensional space, which is defined as the maximum number of available physical hosts; lb _j,min represents the lower limit of the jth dimensional space, which is defined as the minimum number of hosts used. Through the different values of ub _j,max , lb _j,min and c ₄ of each individual salp, the randomly generated salp population can be obtained. It can be seen that the deployment solution space is an N×n Euclidean space , x _1,j represents the position of the leader in the jth dimension space, that is, the jth virtual machine in the deployment solution represented by the leader salps is deployed to the host with serial number x _{1, j} , x _{q, j} (q >1) represents the position of the follower in the jth dimension space, that is, the jth virtual machine in the deployment solution represented by the follower salps q is deployed to the host with the serial number x _{q, j} ,

Indicates rounded up.

In one example, fitness is calculated by formula (7):

Among them, fitness is the fitness.

In one example, according to the fitness, the salp individual with the largest fitness value (representing the optimal deployment solution) is used as the current food source position, and the No. One salp individual acts as the leader, and all the remaining N-2 salps serve as followers.

In one example, the leader position is determined by equation (8):

t表示当前迭代次数，T_max表示方法的最大迭代次数，收敛因子在迭代过程中会从2递减至0。Among them, x _1,j represents the position of the leader salps in the j-th dimensional space, that is, the virtual machine v _j is deployed on the host with the serial number x _{1, j} ; F _j represents the position of the food source in the j-th dimensional space ; Since the food source is unknown for the salps, the position of the leader in the initial state is defined as the position of the food source; ub _j,max represents the upper limit of the j-th dimension space, which is defined as the maximum available host Quantity; lb _j,min represents the lower limit of the j-th dimension space, which is defined as the minimum number of hosts used; c ₂ and c ₃ represent random numbers in the interval [0,1], which are used to determine the lower limit of the j-th dimension space The moving direction (forward or reverse) of an updated position and the moving step size; c ₁ is called the convergence factor, which is used to balance the local development and global exploration ability of the individual salps in the iterative process of the method,

t represents the current number of iterations, _Tmax represents the maximum number of iterations of the method, and the convergence factor will decrease from 2 to 0 during the iteration.

In one example, the follower position is determined by formula (9):

Among them, q'≥2, x _q',j represents the position of the follower in the jth dimension space, that is, the serial number of the host where the virtual machine v _j is deployed is x _q',j ; △t represents time, v ₀ represents the follower The initial rate of , acceleration a=(v _final -v ₀ )/Δt, where v _final =x _q'-1,j -x _q',j /Δt, x _q'-1,j represents the qth The position of '-1 salps on the jth dimension space, that is, in the deployment solution represented by the position of the q'-1 salps, the jth virtual machine is deployed to the host x _q'-1,j superior.

Specifically, due to the heterogeneity of host resource capabilities, its full-load power consumption is different, and the resources of each dimension requested by virtual machines are also different. Therefore, the deployment of n virtual machines on m physical hosts theoretically has m ⁿ cases, therefore, the virtual virtual machine deployment problem itself is NP-hard. The problem is solved by using the salps swarm algorithm, and the energy-optimized virtual machine deployment method according to the present invention may include:

Step 1: Suppose the cloud data center has m heterogeneous physical hosts, which is represented as a set H={h ₁ , h ₂ ,...,h _m }. Each physical host h _i has four types of resources: CPU, memory, storage and network bandwidth, i=1,2,...,m. Through virtualization technology, multiple virtual machines can be deployed on the same physical host at the same time, and each virtual machine can run different tasks. In order to perform application tasks, cloud users can submit load execution requirements to the cloud data center in the form of virtual machines at any time. Assuming that the number of virtual machine requests is n at this time, it is expressed as the set V={v ₁ ,v ₂ ,..., v _n }. That is, it is now necessary to deploy n virtual machines on m physical hosts.

The overall capability of the four resource types of each physical host h _i is defined as a vector C _i =(C _i,CPU ,C _i,MEM ,C _i,DISK ,C _i,NETW ), where C _i,CPU Represents the CPU capability of the physical host hi, C _{i ,MEM} represents the memory capability of the physical host hi, C _{i ,DISK} represents the storage capability of the physical host hi, and C _{i ,NETW} represents the network bandwidth capability of the physical host _hi .

The ability of a virtual machine v _j to request host resources is defined as a vector R _j =(R _j,CPU ,R _j,MEM ,R _j,DISK ,R _j,NETW ), where R _j,CPU represents a virtual machine The CPU capability requested by v _j , R _{j, MEM} represents the memory capability requested by the virtual machine v _j , R _{j, DISK} represents the storage capability requested by the virtual machine v _j , and R _{j, NETW} represents the network bandwidth capability requested by the virtual machine v _j .

Let Load _i,CPU denote the overall CPU load of the physical host h _i , Load _i,MEM denote the overall memory load of the physical host h _i , Load i,MEM denote the overall memory load of the physical host h i, Load _{i ,MEM} denote the physical host h _i The overall memory load of _hi , the CPU utilization of a physical host _hi is:

Let the virtual machine deployment π: V→H represent the deployment solution function of the virtual machine on the physical host, and define the deployment factor χ as:

That is, if the virtual machine _Vb is deployed on the host _Hg , the deployment factor χ=1; otherwise, χ=0. So,

Load _i,CPU =∑Rj _, CPUχ(V,H),i=1,2,...,m,j=1,2,...,n (12)

Load _i,MEM =∑Rj _,MEM χ(V,H),i=1,2,...,m,j=1,2,...,n (13)

Load _i,DISK =∑Rj _,DISKχ (V,H),i=1,2,...,m,j=1,2,...,n (14)

Load _i,NETW =∑Rj _,NETWχ (V,H),i=1,2,...,m,j=1,2,...,n (15)

Host energy consumption is mainly related to its resource utilization, and there is a linear relationship between the two. For the four types of resources possessed by the host, the CPU resource consumes most of the overall energy consumption, ignoring the energy consumption generated by other resource types. Therefore, in order to measure the host energy consumption, the host that has a linear relationship with the CPU utilization will be used. energy consumption model. Using Equation (12), the CPU load on the host h _i can be obtained, and using Equation (10), the CPU utilization on the host h _i can be obtained, which is the sum of the CPU requests of all virtual machines deployed on the host . The power consumption of host _hi is defined as:

Among them, P _{i ,full} represents the power consumption when the host hi is in full load state, that is, the power consumption when 100% of the CPU is occupied, in watts, P _{i ,idle} represents the power consumption when the host hi is in an idle state, Typically, the power consumption when the host is idle is 70% of the full load power consumption. P _i represents the current power consumption of the host _hi .

The goal of virtual machine deployment is to minimize the power consumption of all hosts. Therefore, the deployment goal can be expressed as formula (1), and the constraints are formulas (2)-(5), where formula (2) indicates that on the host h _i The CPU resource load cannot exceed its overall CPU capacity. Equation (3) indicates that the memory resource load on the host h _i cannot exceed its overall memory capacity. Equation (4) indicates that the storage resource load on the host h _i cannot exceed its overall storage capacity. , Equation (5) indicates that the network bandwidth resource load on the host _hi cannot exceed its overall network bandwidth capacity.

Step 2: Determine the deployment solution and calculate the initialization parameters including: salps population size N, maximum number of iterations T _max , random variables c ₁ , c ₂ , c ₃ and c ₄ , upper limit ub _max and lower limit lb _min , virtual machine set V, physical host set H, host resource capability vector (C _CPU , C _MEM , C _DISK , C _NETW ), virtual machine resource request vector (R _CPU , R _MEM , R _DISK , R _NETW ), host's Full load power consumption P _full and idle power consumption P _idle ;

Step 3: Initialize the population through the salps algorithm, so that the individual salps and the deployment solution of the virtual machine correspond one-to-one. The traditional salps group optimization algorithm belongs to the continuous optimization process of the salps, and the individuals of the salps can be continuously in the search space. move at any point. For the virtual machine deployment problem, the deployment solution will be discretely constrained to different physical host serial numbers, so as to indicate which host a single virtual machine is deployed to. Therefore, in the virtual machine deployment problem, the individual positions of salps in the salp group optimization algorithm can only appear in the form of discrete numerical serial numbers between 1 and m, where m represents the maximum serial number of resources. Individual salps are encoded in the following form:

X _q = [x _q,1 ,x _q,2 ,...,x _q,n ] (17)

Among them, X _q represents the position of the salp individual q in the population, and each element x _q,j in the position matrix represents the position of the salp q in the jth dimension. At the same time, let the vector elements x _q,j be random values between [1,m], q=1,2,…,N,N represents the size of the salps population, j=1,2,…,n,n represents The total number of virtual machines to be deployed. The location of the salps, that is, the corresponding deployment solution for the virtual machine can be defined in the n-dimensional search space.

For example: if the position of the individual q of the sea squirt is X _q =[3,2,5,3,2,1,3,2,6], the specific meaning of the virtual machine deployment solution it represents is: virtual machine v _1. Virtual machine _v4 and virtual machine _v7 are deployed _on host _h3 , virtual machine v2, virtual machine _v5 and virtual machine _v8 are deployed on host h3, virtual machine _v3 is deployed _on host _h5 , Virtual machine _v6 is deployed on host _h1 , and virtual machine _v9 is deployed _on host h6.

The positions of all salps are a two-dimensional matrix, which can be defined as X:

The matrix X defines the positions of all N individuals in the iterative evolution of the salp population. Each row in the matrix represents the position of a salp individual in the n-dimensional space, and the position of a salp individual represents a possible virtual Machine deployment solution. By formula (6), the positions of the individuals of the sea squirt are randomized to obtain the initialized population, that is, the initial matrix X. During the iteration process, the N individuals continuously update their positions until the maximum number of iterations is reached, and the fitness in the matrix X is obtained. The location of the largest individual salps as the final virtual virtual machine deployment solution.

Step 4: According to the virtual machine deployment goal and its constraints, calculate the fitness of each salps individual in the initialization population by formula (7). Since the goal of virtual machine deployment is to minimize the energy consumption of the host, the adaptation of the salps is The larger the degree value, the better the deployment solution of the individual representative;

Step 5: The salp population can be divided into two groups: leaders and followers. The salp at the head of the chain is called the leader, and it has the best judgment in the process of finding food sources and guides the movement of the entire population. Except for the leader, the remaining salps are called followers. Followers follow each other and are directly or indirectly led by each other. The food source F in the search space is the search target of the salps population. According to the fitness, the salp individual with the largest fitness value is taken as the current food source position, the first salp individual among the N-1 remaining salp individuals in the order outside the food source is taken as the leader, and the remaining The N-2 salps all serve as followers.

The leader position is updated, that is, the virtual machine deployment solution represented by the salps individual as the leader is updated, and the leader position is determined by formula (8). Since the virtual machine deployment solution can only be limited to [1,m] The integer value in the interval, that is, the value of x _1,j must be within the interval [1,m]. The virtual machine deployment solution represented by the leader is:

Among them, mod represents the modulo operation, that is, taking the remainder; m represents the total number of hosts, that is, the maximum host serial number.

The follower position is determined by formula (9). Since time is the difference of the number of iterations, Δt=1. At the beginning of each iteration, the initial velocity of the follower salps is v ₀ =0, so the position update formula of the follower can be expressed as:

Equation (20) shows that the update of the deployment host serial number of the jth virtual machine in the follower salps q' is equal to the deployment solution corresponding to the salps in the previous iteration and the jth virtual machine in the follower salps q'-1. half of the sum of the deployment host sequence numbers. Based on the same considerations as the leader position, the deployment solution represented by the follower position is:

And then get the updated population.

Step 6: Calculate the fitness of each salp individual in the updated population, determine the salp individual with the largest fitness value, and judge whether the fitness value of the salp individual is greater than the fitness value of the food source. individual as a new food source; if not, retain the original food source;

Step 7: Determine whether the current number of iterations is less than the maximum number of iterations. If so, repeat steps 5-6. If not, output the food source position in the current population as the final optimal solution for virtual machine deployment.

This method fully balances the exploration behavior and development behavior in the iterative process of the salp group through the group optimization model of salp, and achieves the effect of energy-optimized virtual machine deployment while ensuring globality and diversity.

Application example

To facilitate understanding of the solutions and effects of the embodiments of the present invention, a specific application example is given below. It will be understood by those skilled in the art that this example is provided only to facilitate understanding of the invention and that any specific details thereof are not intended to limit the invention in any way.

The energy-optimized virtual machine deployment method according to the present invention includes:

Step 1: The goal of virtual machine deployment is to minimize the power consumption of all hosts. Therefore, the deployment goal can be expressed as Equation (1), and the constraints are Equations (2)-(5), where Equation (2) indicates that the host The CPU resource load on _hi cannot exceed its overall CPU capacity. Equation (3) indicates that the memory resource load on host _hi cannot exceed its overall memory capacity. Equation (4) indicates that the storage resource load on host _hi cannot exceed its overall memory capacity. Overall storage capacity, Equation (5) indicates that the network bandwidth resource load on the host _hi cannot exceed its overall network bandwidth capacity.

Step 2: Determine the deployment solution and calculate the initialization parameters including: salps population size N, maximum number of iterations T _max , random variables c ₁ , c ₂ , c ₃ and c ₄ , upper limit ub _max and lower limit lb _min , virtual The machine set V, the physical host set H, the resource capability vector of the host, the resource request vector of the virtual machine, the full load power consumption and the idle power consumption of the host;

Step 3: Initialize the population through the salps algorithm, so that the individual salps are in one-to-one correspondence with the deployment solution of the virtual machine, and randomize the position of the individual salps by formula (6) to obtain the initialized population;

Step 4: According to the virtual machine deployment target and its constraints, calculate the fitness of each salps individual in the initialization population by formula (7);

Step 5: According to the fitness, the salp individual with the largest fitness value (representing the optimal deployment solution) is used as the current food source position, and the first of the N-1 salp individuals remaining in the order outside the food source is selected. The individual salps is used as the leader, and the remaining N-2 salps are all used as followers, and then the leader position is determined by formula (8), and the follower position is determined by formula (9) to obtain the updated population;

Step 6: Calculate the fitness of each salp individual in the updated population, determine the salp individual with the largest fitness value, and judge whether the fitness value of the salp individual is greater than the fitness value of the food source. individual as a new food source; if not, retain the original food source;

Step 7: Determine whether the current number of iterations is less than the maximum number of iterations. If so, repeat steps 5-6. If not, output the food source position in the current population as the final optimal solution for virtual machine deployment.

To sum up, the present invention fully balances the exploration behavior and development behavior in the iterative process of the salp group through the group optimization model of salp, and achieves the effect of energy-optimized virtual machine deployment while ensuring globality and diversity.

It should be understood by those skilled in the art that the above description of the embodiments of the present invention is only intended to illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any examples given.

According to an embodiment of the present invention, an energy-optimized virtual machine deployment system is provided, characterized in that the system includes: a memory storing computer-executable instructions; and a processor running a computer in the memory Executable instructions, perform the following steps: Step 1: Determine the virtual machine deployment target and its constraints; Step 2: Determine the deployment solution and calculate the initialization parameters; Step 3: Perform population initialization through the salp algorithm, so that the salp individual and the virtual machine are The deployment solutions are in one-to-one correspondence; Step 4: Calculate the fitness of each salp individual in the initialization population according to the virtual machine deployment target and its constraints; Step 5: Determine the food source, leader and follower according to the fitness, and then Determine the leader position and follower position, and obtain the updated population; Step 6: Calculate the fitness of each salp individual in the updated population, determine the salp individual with the largest fitness value, and determine whether the fitness value of the salp individual is greater than the food. The fitness value of the source, if yes, use the individual salps as a new food source; if not, keep the original food source; Step 7: Determine whether the current iteration number is less than the maximum iteration number, if so, repeat steps 5-6 , if not, output the food source position in the current population as the final optimal solution for virtual machine deployment.

In one example, the deployment decomputation initialization parameters include: the maximum number of iterations T _max , the virtual machine set V, the physical host set H, the resource capability vector of the host, the resource request vector of the virtual machine, the full load power consumption and the idle power consumption of the host .

In one example, the virtual machine deployment targets are:

Among them, P _i,full represents the power consumption when the host hi is in full load state, P _{i ,idle} represents the power consumption when the host hi is in the idle state, P _i represents the current power consumption of the host _hi , U _{i ,CPU} is the CPU utilization of host _hi .

In one example, the constraints are:

Load _i,CPU ≤C _i,CPU ,i=1,2,...,m (2)

Load _i,MEM ≤C _i,MEM ,i=1,2,...,m (3)

Load _i,DISK ≤C _i,DISK ,i=1,2,...,m (4)

Load _i,NETW ≤C _i,NETW ,i=1,2,...,m (5)

Among them, C _i,CPU represents the CPU capability of the physical host hi, C _{i ,MEM} represents the memory capability of the physical host hi, C _{i ,DISK} represents the storage capability of the physical host hi, and C _{i ,NETW} represents the physical host _hi is the network bandwidth capability, Load _i,CPU represents the overall CPU load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i ,MEM} represents the overall memory load of the physical host hi, Load _{i , MEM} represents the overall memory load of the physical host _hi .

In one example, step (3) includes: randomizing the position of the individual salps by formula (6) to obtain an initialization population:

表示向上取整。Among them, q=1,2,...,N, N represents the size of the salps population, j=1,2,...,n, n represents the total number of virtual machines, c ₄ represents a random number in the interval [0,1] , ub _j,max represents the upper limit of the jth dimensional space, lb _j,min represents the lower limit of the jth dimensional space, x _1,j represents the position of the leader on the jth dimensional space, x _q,j represents the position of the follower on the jth dimension space,

Indicates rounded up.

In one example, fitness is calculated by formula (7):

Among them, fitness is the fitness.

In one example, according to the fitness, the salp individual with the largest fitness value is taken as the current food source position, and the first salp individual among the N-1 remaining salp individuals ranked outside the food source is taken as the leader , take all the remaining N-2 salps as followers.

In one example, the leader position is determined by equation (8):

t表示当前迭代次数，T_max表示方法的最大迭代次数。Among them, x _1,j represents the position of the leader salps in the jth dimension space, F _j represents the position of the food source in the jth dimension space, ub _j,max represents the upper limit of the jth dimension space, lb _{j ,min} represents the lower limit of the jth dimension space, c ₂ and c ₃ represent random numbers in the interval [0,1], c ₁ is the convergence factor,

t represents the current number of iterations, and _Tmax represents the maximum number of iterations of the method.

In one example, the follower position is determined by formula (9):

Among them, q'≥2, x _q',j represents the position of the follower on the jth dimension, △t represents the time, v ₀ represents the initial velocity of the follower, and the acceleration a=(v _final -v ₀ )/△ t, where v _final =x _q'-1,j -x _q',j /Δt, x _q'-1,j represents the position of the q'-1th salps on the jth dimension space.

This system fully balances the exploration behavior and development behavior in the iterative process of the salps through the group optimization model of salps, and achieves the effect of energy-optimized virtual machine deployment while ensuring globality and diversity.

It should be understood by those skilled in the art that the above description of the embodiments of the present invention is only intended to illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any examples given.

Various embodiments of the present invention have been described above, and the foregoing descriptions are exemplary, not exhaustive, and not limiting of the disclosed embodiments. Numerous modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.

Claims (10)

1. A method for energy-optimized deployment of virtual machines, comprising:

step 1: determining a virtual machine deployment target and constraint conditions thereof;

step 2: determining a deployment solution calculation initialization parameter;

and step 3: performing population initialization through a goblet sea squirt algorithm to enable the goblet sea squirt individuals to correspond to the deployment solution of the virtual machine one by one;

and 4, step 4: calculating the individual fitness of each bottle sea squirt in the initialized population according to the deployment target of the virtual machine and the constraint conditions thereof;

and 5: determining a food source, a leader and a follower according to the fitness, and further determining the position of the leader and the position of the follower to obtain an updated population;

step 6: calculating the fitness of each individual goblet ascidian in the updated population, determining the individual goblet ascidian with the largest fitness value, judging whether the fitness value of the individual goblet ascidian is greater than the fitness value of the food source, and if so, taking the individual goblet ascidian as a new food source; if not, the original food source is reserved;

and 7: and judging whether the current iteration times are smaller than the maximum iteration times, if so, repeating the steps 5-6, and if not, outputting the position of the food source in the current population as a final optimal solution for deployment of the virtual machine.

2. The energy-optimized virtual machine deployment method according to claim 1, wherein deploying solution computation initialization parameters comprises: maximum number of iterations T_maxThe method comprises the following steps of A, a virtual machine set V, a physical host set H, a resource capacity vector of a host, a resource request vector of a virtual machine, and full power consumption and idle power consumption of the host.

3. The energy-optimized virtual machine deployment method of claim 1, wherein the virtual machine deployment goal is:

wherein, P_i,fullIndicates the host h_iPower consumption in the fully loaded state, P_i,idleIndicates the host h_iPower consumption in idle state, P_iIndicates the host h_iCurrent power consumption of U_i,CPUIs a main machine h_iCPU utilization of.

4. The energy-optimized virtual machine deployment method according to claim 1, wherein the constraints are:

Load_i,CPU≤C_i,CPU,i＝1,2,...,m (2)

Load_i,MEM≤C_i,MEM,i＝1,2,...,m (3)

Load_i,DISK≤C_i,DISK,i＝1,2,...,m (4)

Load_i,NETW≤C_i,NETW,i＝1,2,...,m (5)

wherein, C_i,CPURepresents a physical host h_iCPU capability of C_i,MEMRepresents a physical host h_iMemory capacity of C_i,DISKRepresents a physical host h_iStorage capacity of C_i,NETWRepresents a physical host h_iNetwork bandwidth capability, Load_i,CPURepresents a physical host h_iLoad, Load_i,MEMRepresents a physical host h_iLoad, Load_i,MEMRepresents a physical host h_iLoad, Load_i,MEMRepresents a physical host h_iThe overall memory load.

5. The energy-optimized virtual machine deployment method of claim 1, wherein said step (3) comprises: randomizing the position of the individual goblet sea squirt by formula (6) to obtain an initialization population:

wherein q is 1,2, …, N, N represents the size of the ascidian goblet population, j is 1,2, …, N, N represents the total amount of virtual machines, c₄Represents the interval [0,1]Inner random number, ub_j,maxDenotes an upper limit value, lb, of a j-th space_j,minRepresenting a lower bound, x, of a j-space_1,jRepresenting the leader's position in the j-dimensional space, x_q,jRepresenting the position of the follower in the j-th dimension space,

indicating rounding up.

6. The energy-optimized virtual machine deployment method according to claim 1, wherein fitness is calculated by equation (7):

wherein, the fitness is the fitness.

7. The energy-optimized virtual machine deployment method according to claim 1, wherein, according to the fitness, the ascidian goblet with the highest fitness value is used as the current food source position, the first ascidian goblet among the N-1 ascidian goblet individuals in the sequence except the food source is used as the leader, and all the remaining N-2 ascidians goblet are used as followers.

8. The energy-optimized virtual machine deployment method according to claim 1, wherein leader location is determined by equation (8):

wherein x is_1,jRepresenting the position of the sea squirt of the leader goblet in dimension j, F_jRepresenting the position, ub, of the food source in the j-space_j,maxDenotes an upper limit value, lb, of a j-th space_j,minA lower limit value, c, representing the j-th dimension₂And c₃Represents the interval [0,1]Random number of cells, c₁In order to be a factor of convergence, the method comprises the following steps,

t denotes the current number of iterations, T_maxRepresenting the maximum number of iterations of the method.

9. The energy-optimized virtual machine deployment method according to claim 1, wherein follower location is determined by equation (9):

wherein q' is not less than 2, x_q’,jRepresenting the position of the follower in the j-space, △ t representing time, v₀Representing the initial velocity of the follower, acceleration a ═ v_final-v₀) /△ t, wherein v_final＝x_q’-1,j-x_q’,j/△t，x_q’-1,jRepresents the location of the q' -1 st ascidian in the j-dimension space.

10. An energy-optimized virtual machine deployment system, comprising:

a memory storing computer-executable instructions;

a processor executing computer executable instructions in the memory to perform the steps of:

step 1: determining a virtual machine deployment target and constraint conditions thereof;

step 2: determining a deployment solution calculation initialization parameter;

and step 3: performing population initialization through a goblet sea squirt algorithm to enable the goblet sea squirt individuals to correspond to the deployment solution of the virtual machine one by one;

and 4, step 4: calculating the individual fitness of each bottle sea squirt in the initialized population according to the deployment target of the virtual machine and the constraint conditions thereof;

and 5: determining a food source, a leader and a follower according to the fitness, and further determining the position of the leader and the position of the follower to obtain an updated population;

step 6: calculating the fitness of each individual goblet ascidian in the updated population, determining the individual goblet ascidian with the largest fitness value, judging whether the fitness value of the individual goblet ascidian is greater than the fitness value of the food source, and if so, taking the individual goblet ascidian as a new food source; if not, the original food source is reserved;

and 7: and judging whether the current iteration times are smaller than the maximum iteration times, if so, repeating the steps 5-6, and if not, outputting the position of the food source in the current population as a final optimal solution for deployment of the virtual machine.

Images