CN111399985B - Load balancing method based on storage difference iteration in cloud computing environment - Google Patents

Abstract

The invention relates to a load balancing method based on storage difference iteration in a cloud computing environment, which is characterized by comprising a difference storing mechanism and a stepping iteration mechanism, solving the problem of unbalanced load of user resource distribution in the cloud computing environment, realizing load balancing of a virtual machine for reasonably distributing storage tasks, effectively avoiding bad states of idling, overload and the like of the virtual machine, and fully playing the dynamic storage function of a server.

Description

Load balancing method based on storage difference iteration in cloud computing environment

Technical Field

The invention belongs to the technical field of cloud computing, and relates to a load balancing method based on storage difference iteration in a cloud computing environment.

Background

Cloud computing is one of distributed computing, and is to decompose a huge data computing processing program into countless small programs through a network cloud, analyze and process the small programs through a system consisting of a plurality of servers, obtain a computing result and return the computing result to a user. By this technique, the processing of tens of thousands of data can be completed in a short time, thereby providing a strong network service.

Cloud storage is a new concept provided by combining a cloud computing-based concept with big data storage, and also can be said to be a cloud computing system focusing on data storage and management. Cloud storage is a system that connects a large number of storage devices of different configurations and different types together by server clusters, distributed file systems and grid technologies, and provides data storage management as a whole. The load balancer is used for coordinating resource allocation among the servers to achieve load balancing.

At present, the load balancing strategy mainly comprises the following methods: (1) a random method, wherein the storage task is randomly allocated to the virtual machine; (2) the weighted random method is used for randomly distributing the storage tasks to the virtual machines, and the greater the weight of the virtual machines is, the greater the distribution possibility is; (3) a polling method, in which requests from users are distributed to internal virtual machines in turn; (4) the weighted polling method distributes different weights to each server according to different processing capacities of the servers, so that the servers can receive service requests with corresponding weight numbers. However, these methods still have some problems, mainly expressed in:

(1) the random strategy is unstable in the storage process, and the possibility that one or a plurality of servers are continuously stored and overloaded exists.

(2) The common polling method assumes that the processing performance of all servers is the same, does not consider the current remaining space and response speed of each server, and when the size of a storage task changes greatly, the polling algorithm easily causes load imbalance among the servers.

(3) The weights of the weighted random strategy and the weighted polling strategy in the storage need to be statically configured, and cannot be dynamically adjusted according to the current situation of the virtual machine.

Disclosure of Invention

The invention aims to provide a load balancing method based on storage difference iteration in a cloud computing environment, aiming at the problem of load imbalance of user resource distribution in the cloud computing environment.

The purpose of the invention is realized by the following technical scheme: a load balancing method based on storage difference iteration in a cloud computing environment is characterized by comprising the following steps: the content of the method comprises a differentiable value mechanism and a stepping iteration mechanism,

1) the differentiable mechanism;

in the cloud computing environment, a certain number of user demands in unit time are used as a batch processing object to be subjected to batch processing, and in the user demand batch processing process, all virtual machines are set to be M₁,M₂,...,M_nThe size of the residual space of each virtual machine is S₁,S₂,...,S_nThe performance index of the virtual machine is P₁,P₂,...,P_nPerformance index P of jth virtual machine_jThe calculation formula is as follows:

wherein S is_jRepresents the size of the residual space of the jth virtual machine, C_jDenotes the CPU utilization, V, of the jth virtual machine_jIndicating the reference calculation speed, V, of the jth virtual machine_kA performance index P of the k-th virtual machine₁,P₂,...,P_nSequencing from low to high to obtain P₁′,P₂′,...,P_n', according to the performance index P₁′,P₂′,...,P_n' ordering the virtual machines, the order of the virtual machines is M in turn₁′,M₂′,...,M_n', the residual space after the virtual machine sequencing is S in sequence₁′,S₂′,...,S_n' let the task to be stored be T₁,T₂,…,T_mThe size of the task to be stored is Z₁,Z₂,…,Z_mWhen n isWhen the number of the virtual machines is more than or equal to m, namely the number of the virtual machines is more than or equal to the number of the tasks, defining the storable difference value of each virtual machine as follows:

D_i＝S_i′-Z_i，(1≤i≤m) (2)

wherein D is_iRepresenting the storable difference value of the ith sorted virtual machine, S_i' denotes the remaining space of the i-th virtual machine after sorting, Z_iRepresenting the size of the ith task to be stored, and when n is less than m, defining the storable difference value of each virtual machine as:

D_i-(i|n)×n＝S_i-(i|n)×n′-Z_i，(1≤i≤m) (3)

wherein D is_i-(i|n)×nRepresenting the storable difference, S, of the i- (i | n) × n sorted virtual machines_i-(i|n)×n' represents the size of the remaining space of the i- (i | n) × n virtual machines after sorting, Z_iRepresenting the size of the ith task to be stored;

2) the step-back iteration mechanism;

when n is larger than or equal to m, namely the number of the virtual machines is larger than or equal to the number of the tasks, judging whether D is met or not_iNot less than 0, if D_iIf T is more than or equal to 0, then_i(i-1, 2, …, M) are assigned to M in order₁′,M₂′,...,M_n', i.e. T₁Is allocated to M₁′，T₂Is allocated to M₂′，...，T_mIs allocated to M_m', if D_i< 0, judging whether S is satisfied_i+1′-Z_iIf S is more than or equal to 0_i+1′-Z_iNot less than 0, adding T_iIs allocated to M_i+1', if S_i+1′-Z_i< 0, judging whether S is satisfied_i+2′-Z_iIf S is more than or equal to 0_i+2′-Z_iNot less than 0, adding T_iIs allocated to M_i+2', if S_i+2′-Z_i< 0, judge S_i+3′-Z_i,., following S_i′-Z_iAllocate T > 0_iIs allocated to M_i′，S_i′-Z_iStep back is judged S when the time is less than 0_i+1′-Z_iIf the rule is more than or equal to 0, the task to be stored is allocated to a restVirtual machines with a space size larger than the size of the task to be stored, in particular, when S_i′-Z_i,S_i+1′-Z_i,S_i+2′-Z_i,...,S_n′-Z_iWhen the values are all less than 0, the first round of iteration is judged to be finished at the moment, and S is returned₁′-Z_iA second iteration is performed, i.e. S is calculated one by one₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the condition is not satisfied, a third iteration is performed, …, following S_i′-Z_i,S_i+1′-Z_i,S_i+2′-Z_i,...,S_n′-Z_iAll are less than 0, return to S₁′-Z_iCarrying out a new iteration and carrying out the iteration in a circulating way until the condition of more than or equal to 0 is finished, wherein each iteration possibly meets the requirement of S according to the randomness of the stored data and the released data of the virtual machine₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the value is larger than or equal to zero, the performance of the virtual machine is arranged from low to high, and the subsequent virtual machine can store T_iThe probability of (D) becomes higher in turn, the calculation time of the virtual machine can be saved, and when n is less than m, whether D is satisfied is judged_i-(i|n)×nNot less than 0, if D_i-(i|n)×nIf the number of the distribution is more than or equal to 0, starting to distribute according to rounds, and carrying out T distribution in the first round₁Is allocated to M₁′，T₂Is allocated to M₂′，...，T_nIs allocated to M_n', the second round will T_n+1Is allocated to M₁', will T_n+2Is allocated to M₂Until T is got_mIs allocated to M_m-(m|n)×n' at this point, if D_i-(i|n)×nIf < 0, it is judged whether S is satisfied_i-(i|n)×n+1′-Z_iIf S is more than or equal to 0_i-(i|n)×n+1′-Z_iIf T is more than or equal to 0, then_iIs allocated to M_i-(i|n)×n+1If less than 0, determine whether S is satisfied_i-(i|n)×n+2′-Z_iIf S is more than or equal to 0_i-(i|n)×n+2′-Z_iIf T is more than or equal to 0, then_iIs allocated to M_i-(i|n)×n+2If it is less than0, judgment S_i-(i|n)×n+3′-Z_iIs more than or equal to 0_i-(i|n)×n′-Z_iAllocate T > 0_iIs allocated to M_i-(i|n)×n′，S_i-(i|n)×n′-Z_iStep back is judged S when the time is less than 0_i-(i|n)×n+1′-Z_iIf the rule is greater than or equal to 0, the task to be stored is allocated to a virtual machine with the residual space size greater than that of the task to be stored, in particular, when S is_i-(i|n)×n′-Z_i,S_i-(i|n)×n+1′-Z_i,S_i-(i|n)×n+2′-Z_i,...,S_n′-Z_iWhen the values are all less than 0, the first round of iteration is judged to be finished at the moment, and S is returned₁′-Z_iA new iteration is performed, i.e. S is calculated one by one₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the condition is not satisfied, a third iteration is performed, …, following S_i-(i|n)×n′-Z_i,S_i-(i|n)×n+1′-Z_i,S_i-(i|n)×n+2′-Z_i,...,S_n′-Z_iAll are less than 0, return to S₁′-Z_iAnd carrying out a new iteration and circulating until the condition that the condition is more than or equal to 0 is met.

The load balancing method based on storage difference iteration in the cloud computing environment comprises a difference storing mechanism and a stepping iteration mechanism, can solve the problem of unbalanced load of user resource distribution in the cloud computing environment, realizes load balancing of a virtual machine for reasonably distributing storage tasks, effectively avoids bad states of idling, overload and the like of the virtual machine, gives full play to the dynamic storage function of a server, and has the advantages of being scientific and reasonable, strong in applicability, good in effect and the like.

Drawings

Fig. 1 is a flowchart of a load balancing method based on storage difference iteration in a cloud computing environment according to the present invention.

Detailed Description

The invention is further illustrated by the following figures and detailed description.

Referring to fig. 1, the load balancing method based on storage difference iteration in a cloud computing environment of the present invention includes a difference storing mechanism and a back-stepping iteration mechanism:

1) the differentiable mechanism;

in the cloud computing environment, a certain number of user demands in unit time are used as a batch processing object to be subjected to batch processing, and in the user demand batch processing process, all virtual machines are set to be M₁,M₂,...,M_nThe size of the residual space of each virtual machine is S₁,S₂,...,S_nThe performance index of the virtual machine is P₁,P₂,...,P_nPerformance index P of jth virtual machine_jThe calculation formula is as follows:

wherein S is_jRepresents the size of the residual space of the jth virtual machine, C_jDenotes the CPU utilization, V, of the jth virtual machine_jIndicating the reference calculation speed, V, of the jth virtual machine_kA performance index P of the k-th virtual machine₁,P₂,...,P_nSequencing from low to high to obtain P₁′,P₂′,...,P_n', according to the performance index P₁′,P₂′,...,P_n' ordering the virtual machines, the order of the virtual machines is M in turn₁′,M₂′,...,M_n', the residual space after the virtual machine sequencing is S in sequence₁′,S₂′,...,S_n' let the task to be stored be T₁,T₂,…,T_mThe size of the task to be stored is Z₁,Z₂,…,Z_mWhen n is larger than or equal to m, namely the number of the virtual machines is larger than or equal to the number of the tasks, defining the storable difference value of each virtual machine as follows:

D_i＝S_i′-Z_i，(1≤i≤m) (2)

wherein D is_iRepresents the ith after sortingVirtual machine storable difference value, S_i' denotes the remaining space of the i-th virtual machine after sorting, Z_iRepresenting the size of the ith task to be stored, and when n is less than m, defining the storable difference value of each virtual machine as:

D_i-(i|n)×n＝S_i-(i|n)×n′-Z_i，(1≤i≤m) (3)

wherein D is_i-(i|n)×nRepresenting the storable difference, S, of the i- (i | n) × n sorted virtual machines_i-(i|n)×n' represents the size of the remaining space of the i- (i | n) × n virtual machines after sorting, Z_iRepresenting the size of the ith task to be stored;

2) the step-back iteration mechanism;

when n is larger than or equal to m, namely the number of the virtual machines is larger than or equal to the number of the tasks, judging whether D is met or not_iNot less than 0, if D_iIf T is more than or equal to 0, then_i(i-1, 2, …, M) are assigned to M in order₁′,M₂′,...,M_n', i.e. T₁Is allocated to M₁′，T₂Is allocated to M₂′，...，T_mIs allocated to M_m', if D_i< 0, judging whether S is satisfied_i+1′-Z_iIf S is more than or equal to 0_i+1′-Z_iNot less than 0, adding T_iIs allocated to M_i+1', if S_i+1′-Z_i< 0, judging whether S is satisfied_i+2′-Z_iIf S is more than or equal to 0_i+2′-Z_iNot less than 0, adding T_iIs allocated to M_i+2', if S_i+2′-Z_i< 0, judge S_i+3′-Z_i,., following S_i′-Z_iAllocate T > 0_iIs allocated to M_i′，S_i′-Z_iStep back is judged S when the time is less than 0_i+1′-Z_iIf the rule is greater than or equal to 0, the task to be stored is allocated to a virtual machine with the residual space size greater than that of the task to be stored, in particular, when S is_i′-Z_i,S_i+1′-Z_i,S_i+2′-Z_i,...,S_n′-Z_iWhen the values are all less than 0, the first round of iteration is judged to be finished at the moment, and S is returned₁′-Z_iA second iteration is performed, i.e. S is calculated one by one₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the condition is not satisfied, a third iteration is performed, …, following S_i′-Z_i,S_i+1′-Z_i,S_i+2′-Z_i,...,S_n′-Z_iAll are less than 0, return to S₁′-Z_iCarrying out a new iteration and carrying out the iteration in a circulating way until the condition of more than or equal to 0 is finished, wherein each iteration possibly meets the requirement of S according to the randomness of the stored data and the released data of the virtual machine₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the value is larger than or equal to zero, the performance of the virtual machine is arranged from low to high, and the subsequent virtual machine can store T_iThe probability of (D) becomes higher in turn, the calculation time of the virtual machine can be saved, and when n is less than m, whether D is satisfied is judged_i-(i|n)×nNot less than 0, if D_i-(i|n)×nIf the number of the distribution is more than or equal to 0, starting to distribute according to rounds, and carrying out T distribution in the first round₁Is allocated to M₁′，T₂Is allocated to M₂′，...，T_nIs allocated to M_n', the second round will T_n+1Is allocated to M₁', will T_n+2Is allocated to M₂Until T is got_mIs allocated to M_m-(m|n)×n' at this point, if D_i-(i|n)×nIf < 0, it is judged whether S is satisfied_i-(i|n)×n+1′-Z_iIf S is more than or equal to 0_i-(i|n)×n+1′-Z_iIf T is more than or equal to 0, then_iIs allocated to M_i-(i|n)×n+1If less than 0, determine whether S is satisfied_i-(i|n)×n+2′-Z_iIf S is more than or equal to 0_i-(i|n)×n+2′-Z_iIf T is more than or equal to 0, then_iIs allocated to M_i-(i|n)×n+2If it is less than 0, determine S_i-(i|n)×n+3′-Z_iIs more than or equal to 0_i-(i|n)×n′-Z_iAllocate T > 0_iIs allocated to M_i-(i|n)×n′，S_i-(i|n)×n′-Z_iStep back is judged S when the time is less than 0_i-(i|n)×n+1′-Z_iWhether or not it is greater than or equal toRule 0, assign the task to be stored to a virtual machine with a size of the remaining space larger than the size of the task to be stored, in particular, when S_i-(i|n)×n′-Z_i,S_i-(i|n)×n+1′-Z_i,S_i-(i|n)×n+2′-Z_i,...,S_n′-Z_iWhen the values are all less than 0, the first round of iteration is judged to be finished at the moment, and S is returned₁′-Z_iA new iteration is performed, i.e. S is calculated one by one₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the condition is not satisfied, a third iteration is performed, …, following S_i-(i|n)×n′-Z_i,S_i-(i|n)×n+1′-Z_i,S_i-(i|n)×n+2′-Z_i,...,S_n′-Z_iAll are less than 0, return to S₁′-Z_iAnd carrying out a new iteration and circulating until the condition that the condition is more than or equal to 0 is met.

The software routines involved in the present invention are organized according to automation, networking, and computer processing techniques, and are well known to those skilled in the art.

The particular embodiments of the present invention have been shown by way of example only and not by way of limitation, and it will be understood by those skilled in the art that variations and modifications in other variations may be made in the practice of the invention, and it is not necessary to exhaustively enumerate all embodiments, but rather, obvious variations and modifications may be resorted to without departing from the scope of the invention.

Claims (1)

1. A load balancing method based on storage difference iteration in a cloud computing environment is characterized by comprising the following steps: the content of the method comprises a differentiable value mechanism and a stepping iteration mechanism,

1) the differentiable mechanism;

in the cloud computing environment, a certain number of user demands in unit time are used as a batch processing object to be subjected to batch processing, and in the user demand batch processing process, all virtual machines are set to be M₁,M₂,...,M_nThe size of the residual space of each virtual machine is S₁,S₂,...,S_nThe performance index of the virtual machine is P₁,P₂,...,P_nPerformance index P of jth virtual machine_jThe calculation formula is as follows:

wherein S is_jRepresents the size of the residual space of the jth virtual machine, C_jDenotes the CPU utilization, V, of the jth virtual machine_jIndicating the reference calculation speed, V, of the jth virtual machine_kA performance index P of the k-th virtual machine₁,P₂,...,P_nSequencing from low to high to obtain P₁′,P₂′,...,P_n', according to the performance index P₁′,P₂′,...,P_n' ordering the virtual machines, the order of the virtual machines is M in turn₁′,M₂′,...,M_n', the residual space after the virtual machine sequencing is S in sequence₁′,S₂′,...,S_n' let the task to be stored be T₁,T₂,…,T_mThe size of the task to be stored is Z₁,Z₂,…,Z_mWhen n is larger than or equal to m, namely the number of the virtual machines is larger than or equal to the number of the tasks, defining the storable difference value of each virtual machine as follows:

D_i＝S_i′-Z_i，1≤i≤m (2)

wherein D is_iRepresenting the storable difference value of the ith sorted virtual machine, S_i' denotes the remaining space of the i-th virtual machine after sorting, Z_iRepresenting the size of the ith task to be stored, and when n is less than m, defining the storable difference value of each virtual machine as:

D_i-(i|n)×n＝S_i-(i|n)×n′-Z_i，1≤i≤m (3)

wherein D is_i-(i|n)×nRepresents the i- (i | n) × n virtual items after sortingValue of the available difference of the machines, S_i-(i|n)×n' represents the size of the remaining space of the i- (i | n) × n virtual machines after sorting, Z_iRepresenting the size of the ith task to be stored;

2) the step-back iteration mechanism;

when n is larger than or equal to m, namely the number of the virtual machines is larger than or equal to the number of the tasks, judging whether D is met or not_iNot less than 0, if D_iIf T is more than or equal to 0, then_iI is 1,2, …, M is assigned to M in sequence₁′,M₂′,...,M_n', i.e. T₁Is allocated to M₁′，T₂Is allocated to M₂′，...，T_mIs allocated to M_m', if D_i< 0, judging whether S is satisfied_i+1′-Z_iIf S is more than or equal to 0_i+1′-Z_iNot less than 0, adding T_iIs allocated to M_i+1', if S_i+1′-Z_i< 0, judging whether S is satisfied_i+2′-Z_iIf S is more than or equal to 0_i+2′-Z_iNot less than 0, adding T_iIs allocated to M_i+2', if S_i+2′-Z_i< 0, judge S_i+3′-Z_i,., following S_i′-Z_iAllocate T > 0_iIs allocated to M_i′，S_i′-Z_iStep back is judged S when the time is less than 0_i+1′-Z_iIf the rule is greater than or equal to 0, the task to be stored is allocated to a virtual machine with the residual space size greater than that of the task to be stored, in particular, when S is_i′-Z_i,S_i+1′-Z_i,S_i+2′-Z_i,...,S_n′-Z_iWhen the values are all less than 0, the first round of iteration is judged to be finished at the moment, and S is returned₁′-Z_iA second iteration is performed, i.e. S is calculated one by one₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the condition is not satisfied, a third iteration is performed, …, following S_i′-Z_i,S_i+1′-Z_i,S_i+2′-Z_i,...,S_n′-Z_iAll are less than 0, return to S₁′-Z_iCarrying out a new iteration and carrying out the iteration in a circulating way until the condition of more than or equal to 0 is finished, wherein each iteration possibly meets the requirement of S according to the randomness of the stored data and the released data of the virtual machine₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the value is larger than or equal to zero, the performance of the virtual machine is arranged from low to high, and the subsequent virtual machine can store T_iThe probability of (D) becomes higher in turn, the calculation time of the virtual machine can be saved, and when n is less than m, whether D is satisfied is judged_i-(i|n)×nNot less than 0, if D_i-(i|n)×nIf the number of the distribution is more than or equal to 0, starting to distribute according to rounds, and carrying out T distribution in the first round₁Is allocated to M₁′，T₂Is allocated to M₂′，...，T_nIs allocated to M_n', the second round will T_n+1Is allocated to M₁', will T_n+2Is allocated to M₂Until T is got_mIs allocated to M_m-(m|n)×n' at this point, if D_i-(i|n)×nIf < 0, it is judged whether S is satisfied_i-(i|n)×n+1′-Z_iIf S is more than or equal to 0_i-(i|n)×n+1′-Z_iIf T is more than or equal to 0, then_iIs allocated to M_i-(i|n)×n+1If less than 0, determine whether S is satisfied_i-(i|n)×n+2′-Z_iIf S is more than or equal to 0_i-(i|n)×n+2′-Z_iIf T is more than or equal to 0, then_iIs allocated to M_i-(i|n)×n+2If it is less than 0, determine S_i-(i|n)×n+3′-Z_iIs more than or equal to 0_i-(i|n)×n′-Z_iAllocate T > 0_iIs allocated to M_i-(i|n)×n′，S_i-(i|n)×n′-Z_iStep back is judged S when the time is less than 0_i-(i|n)×n+1′-Z_iIf the rule is greater than or equal to 0, the task to be stored is allocated to a virtual machine with the residual space size greater than that of the task to be stored, in particular, when S is_i-(i|n)×n′-Z_i,S_i-(i|n)×n+1′-Z_i,S_i-(i|n)×n+2′-Z_i,...,S_n′-Z_iWhen the values are all less than 0, the first round of iteration is judged to be finished at the moment, and S is returned₁′-Z_iA new iteration is performed, i.e. S is calculated one by one₁′-Z_i,S₂′-Z_i,S₃′-Z_i,...,S_n′-Z_iIf the condition is not satisfied, a third iteration is performed, …, following S_i-(i|n)×n′-Z_i,S_i-(i|n)×n+1′-Z_i,S_i-(i|n)×n+2′-Z_i,...,S_n′-Z_iAll are less than 0, return to S₁′-Z_iAnd carrying out a new iteration and circulating until the condition that the condition is more than or equal to 0 is met.

Images