CN102929687A - Energy-saving virtual machine placement method for cloud computing data center - Google Patents

Abstract

The energy-saving cloud computing data center virtual machine placement method of the present invention includes: a. establishing a physical server area; b. Obtain physical server information; c. calculating virtual machine resources to be created; d. Sort the physical servers; e. Establishing a mapping between the virtual machine to be created and the physical server; f. Judging whether there is a single physical server to meet the demand; g. Judging whether there is a physical server sequence in the same partition that meets the requirements; h. Select the server sequence with the smallest air-conditioning energy consumption cost and the smallest number of physical servers. In the cloud computing data center virtual machine placement method of the present invention, a single physical server is preferred to create a virtual machine, and then a physical server in the same partition is selected to create a virtual machine; when the former two do not exist, the degree of concentration is finally selected The highest physical server sequence is used to create virtual machines to realize the energy-saving placement of virtual machines in cloud computing centers. The energy-saving effect is remarkable and it is convenient for application and promotion.

Description

An energy-saving method for placing virtual machines in cloud computing data center

technical field

The present invention relates to an energy-saving method for placing virtual machines in a cloud computing data center, and more specifically, to an energy-saving method for placing virtual machines in a cloud computing data center that maximizes the centralized placement of all virtual machines to be created.

Background technique

Cloud computing (Cloud Computing) is the product of the integration of traditional computer and network technologies such as grid computing, distributed computing, parallel computing, network storage, and virtualization. It is a commercial computing model that distributes computing tasks on a resource pool composed of a large number of physical computers, enabling various application systems to obtain computing power, storage space, and information services on demand. Its core is a service resource pool, which is usually some virtualized resources that can be maintained and managed by itself, including computing servers, storage servers, and bandwidth resources. Cloud computing includes the following levels of services:

Infrastructure as a Service (Infrastructure as a Service, referred to as IaaS), means that consumers obtain services from a complete computer infrastructure through the Internet, including processing, storage, network and other basic computing resources, on which users can freely deploy and run Arbitrary software.

Software as a Service (Software as a Service) refers to a mode of providing software services through the Internet. Users do not need to purchase software, but rent Web-based software from providers to manage business activities.

Platform as a Service (Platform as a Service, referred to as PaaS), refers to the software development platform as a service, which is submitted to users in the form of SaaS. Paas is an application of the SaaS model. The emergence of PaaS can accelerate the development of SaaS, especially the development of SaaS applications.

At present, cloud data centers are mainly constructed in the form of virtual data centers. Virtual data centers refer to using server virtualization technology to use independent and isolated virtual hosts to provide functions equivalent to physical machines, and their cost is far lower than that of physical data centers. The core of a virtual data center is a virtual machine. The so-called virtual machine is computer software that runs on physical hardware or a physical computer. It can run operating systems (called guest operating systems) and applications. It has its own virtual hardware. . Virtual machines are not emulators and simulators, they are real computers that can achieve the same or even exceed the functions of physical computers.

Users can use virtualization technology to provide large-scale virtualization services based on a small number of physical servers, thus greatly reducing the construction cost of cloud data centers. At present, power consumption has become the main operating cost of cloud computing data centers. Take Google as an example. According to the existing public information, Google has 36 data centers around the world, which consume about 79 billion kWh of electricity a year; in 2011, data centers in the United States consumed a total of 1 trillion kWh of electricity, equivalent to 7.4 billion U.S. dollars . Energy saving has become a major problem that needs to be solved urgently in the operation of cloud data centers. In the operation of the cloud computing data center, it provides users with virtual machine self-service through its huge physical server cluster. The virtual machine is mainly created on each physical server. Raise the surrounding temperature. According to estimates, the highest problem can reach more than 50 degrees. The temperature rise will cause the air conditioning and refrigeration system to work. The standard temperature of the data center is maintained through the air conditioning and refrigeration system. The resulting power consumption is about It accounts for 40% of the electricity consumption of cloud data centers. The core idea of the centralized placement of virtual machines is to concentrate the creation of virtual machines on fewer physical servers, thereby reducing the generation of waste heat of physical servers and saving power consumption. Therefore, reducing the power consumption of the cooling system through efficient centralized placement of virtual machines has become an important way to save energy in cloud data centers.

Today's mainstream virtual machine placement methods mainly include packaging method, striping method, load-aware method and memory affinity method.

The basic idea of the packaging method is to use as few nodes as possible, and concentrate the virtual machines to run on some nodes in the cloud computing. In terms of implementation, the principle of the largest number of running virtual machines is adopted, that is, when it is necessary to select a host for a new virtual machine For the host, choose the host with the largest number of virtual machines running. The advantage of the packaging method is that this method can make a large number of virtual machines run on a small number of physical nodes, which can reduce the cost of physical servers. The disadvantage of the packaging method is that the virtual machines are too large. Concentration leads to a high probability of virtual machine resource preemption. In order to ensure the quality of virtual servers, it is necessary to migrate a large number of virtual machines and adjust resources, which will generate a lot of overhead.

The basic idea of the striping method is to maximize the available resources of a single server node, and distribute the virtual machines to run on all nodes. In terms of implementation, the principle of the least number of virtual machines running is adopted, that is, when it is necessary to select a host for a new virtual machine Yes, the basic idea of choosing a host with the least number of virtual machines running is inspired by cluster load balancing. The number of virtual machines is evenly distributed according to the number of virtual machines to reduce the probability of virtual machine resource preemption. The disadvantage is that the virtual machines are placed according to the number of The number of machines is only abstracted as the load value on the node, which is too single, and it is impossible to distinguish the actual load caused by the virtual machine requested by different resources (such as CPU, memory, hard disk, etc.), which is not detailed enough to achieve finer granularity and precision resource allocation needs.

The goal of load-aware approach is the same as that of striping, which seeks to maximize the available resources on a single node. The basic design idea is inspired by the minimum node load, and the newly created virtual machine is placed on the node with the minimum load to run. In terms of implementation, the principle of giving priority to the maximum CPU idle rate is adopted, that is, when a host machine needs to be selected for a new virtual machine, the host machine with the highest CPU idle rate is selected. Its advantages are the same as those of the striping method. Inspired by load balancing, it can achieve CPU resource load balancing within the scope of the distributed computing system by considering the usage of CPU resources on nodes. Its disadvantage is that this method only considers CPU resources, and lacks consideration of the important components responsible for cluster nodes such as node memory, network and disk usage.

The memory proximity method is a virtual machine placement system based on memory sharing awareness proposed by Dr. Timothy Wood, including a memory identification system that can effectively judge the memory sharing potential among a group of virtual machines and calculate a more effective placement method . In addition, as the load changes, the system will also use online migration to optimize virtual machine placement. The advantage is to find virtual machines with the same virtual memory pages from the cluster range, so that they can migrate to the same cluster node and share virtual memory, which can improve physical memory. Utilization, save memory, and increase the number of virtual machines in the cluster. Its disadvantage is that because virtual machine migration is required to realize virtual machine placement, the method process is relatively complicated, and if there are too many virtual machines sharing virtual memory, the service quality of the virtual server cannot be guaranteed.

For the virtual machine placement method proposed above, the problem is only considered from the virtual machine placement itself, and the design is not combined with the energy consumption of the air conditioning and refrigeration system. For the packaging method, the idea is to place multiple virtual machines in fewer places as much as possible. However, this method only considers the logical concentration of physical servers. In practice, the deployed servers may be located in different physical cabinet areas. Deploying in this way will increase the temperature of different physical server areas. , causing the cooling system to work, thereby consuming electric energy; for the method of striping, load sensing, and memory closeness, the idea is to distribute the virtual machine to multiple physical nodes, so as to achieve the function of load balancing, this type of method will cause The rise in temperature on a large scale causes large-scale work of refrigeration and air conditioning. Therefore, the above method cannot realize the energy saving of the air conditioning and refrigeration system during the implementation process.

Contents of the invention

In order to overcome the shortcomings of the above-mentioned technical problems, the present invention proposes an energy-saving cloud computing data center virtual machine placement method that maximizes the centralized placement of all virtual machines to be created.

和

；d．对物理服务器进行排序，对所有物理服务器依据可用资源的大小进行排序，形成物理服务器序列集合，该序列集合记为；e．建立待创建虚拟机与物理服务器之间的映射，在可用资源满足待创建虚拟机请求资源的基础上，搜索满足需求的物理服务器资源序列；并按照序列中包含物理服务器数量的多少对该资源序列进行分类，形成物理服务器序列集合，分别记为

，，…，，其中

表示包含n个物理服务器且可用资源满足需求的所有序列的集合，n为正整数；f．判断是否有单个物理服务器满足需求，如果有集合

存在，则中任何一台物理服务器的可用资源均可创建出所有虚拟机，选择其中任一物理服务器作为待创建虚拟机的节点；如不存在集合

，则执行步骤g；g．判断是否存在同一分区内的满足要求的物理服务器序列，遍历物理服务器序列集合

，…，

，寻找出位于同一物理服务器分区内的物理服务器序列，在该物理服务器序列上创建所有虚拟机；如不存在处于同一分区内的满足要求的物理服务器序列，则执行步骤h；h．选取空调能耗成本最小、且物理服务器数量最少的服务器序列，来创建所有虚拟机；空调能耗成本大小与物理服务器集中程度大小成反比。 The energy-saving cloud computing data center virtual machine placement method of the present invention is special in that: a. Establish a physical server area, divide all physical servers covered by an air outlet of the air-conditioning and refrigeration system into the same physical server partition, and form a cloud-based The calculated physical server area; and the number of physical server partitions included in the horizontal and vertical directions of the physical server area is respectively x and y, and x and y are both positive integers; b. Obtaining physical server information, obtaining resource usage information and partition information of all physical servers in step a; c. Calculate the virtual machine resources to be created, obtain the CPU, memory and hard disk size of each virtual machine to be created, and calculate the requested resources required to create all virtual machines; set the required CPU, memory and hard disk requested resources as ,

and

; d. Sort the physical servers, and sort all the physical servers according to the size of the available resources to form a sequence set of physical servers, which is denoted as ; e. Establish a mapping between the virtual machine to be created and the physical server, and search for a sequence of physical server resources that meet the requirements on the basis that the available resources meet the resources requested by the virtual machine to be created; and sequence the resources according to the number of physical servers included in the sequence Classify to form a set of physical server sequences, which are respectively denoted as

,,..., ,in

Indicates the collection of all sequences that contain n physical servers and available resources meet the requirements, n is a positive integer; f. Determine whether there is a single physical server to meet the demand, if there is a collection

exist, then All virtual machines can be created from the available resources of any one of the physical servers, and any one of the physical servers can be selected as the node of the virtual machine to be created; if there is no set

, execute step g; g. Determine whether there is a physical server sequence that meets the requirements in the same partition, and traverse the physical server sequence set

,...,

, find out the physical server sequence located in the same physical server partition, and create all virtual machines on the physical server sequence; if there is no physical server sequence meeting the requirements in the same partition, then perform step h; h. Select the server sequence with the smallest air-conditioning energy consumption cost and the least number of physical servers to create all virtual machines; the air-conditioning energy consumption cost is inversely proportional to the concentration of physical servers.

为所有待创建虚拟机的CPU之和，为所有待创建虚拟机的内存之和，为所有待创建虚拟机的硬盘之和。步骤d中，对物理服务器进行排序，是为了步骤e中便于与待创建虚拟机形成映射。步骤e中，形成了满足资源需求的所有物理服务器序列集合，以便于选择出能耗最低的虚拟机放置方案。步骤f中，如果有集合

存在，则表明有一台物理服务器的可用资源即可满足创建所有虚拟机的要求，则在一台物理服务器上进行创建，以便达到最佳的节能目的。步骤g中，在同一分区内的物理服务器上创建所有虚拟机，也有利于服务器的集中散热，达到较佳的节能目的。步骤h中，如果不存在一个或同一分区的物理服务器满足创建虚拟机的要求，则选取两个或两个以上分区内的物理服务器来创建虚拟机，选取物理服务器集中程度最大的服务器序列来放置虚拟机，以降低能耗成本。 In step a, each physical server partition should be provided with its own temperature sensor, so that the air conditioning and refrigeration system can control the opening and closing of the corresponding air outlet according to the detected temperature; the physical server partition is referred to as partition. In step b, the resource usage information of the physical server is the usage information of CPU, memory and hard disk, and the partition information is the coordinate information of the partition where the physical server is located. In step c,

is the sum of CPUs of all virtual machines to be created, is the sum of memory of all virtual machines to be created, It is the sum of the hard disks of all virtual machines to be created. In step d, sorting the physical servers is for the convenience of forming a mapping with the virtual machine to be created in step e. In step e, a sequence set of all physical servers meeting resource requirements is formed, so as to select a virtual machine placement scheme with the lowest energy consumption. In step f, if there is a set

If it exists, it means that the available resources of one physical server can meet the requirements for creating all virtual machines, and then create them on one physical server to achieve the best energy saving purpose. In step g, all the virtual machines are created on the physical servers in the same partition, which is also conducive to the centralized heat dissipation of the servers and better energy saving. In step h, if there is no physical server in one or the same partition that meets the requirements for creating a virtual machine, select physical servers in two or more partitions to create a virtual machine, and select the server sequence with the largest concentration of physical servers to place virtual machines to reduce energy costs.

In the energy-saving cloud computing data center virtual machine placement method of the present invention, the location of the physical server partition is represented by two-dimensional coordinates P=(i, j); the energy consumption cost of the air conditioner is marked by cost, and the definition is:

标示满足条件的物理服务器所在两分区之间的拓扑距离；，标示满足条件的物理服务器所在分区两两分区之间的拓扑距离的总和；其特征在于：所述步骤g中，如果计算出物理服务器序列集合的空调能耗成本cost的值为1，则表明该物理服务器序列集合位于同一物理服务器分区内；如果物理服务器分区

、

的坐标分别为（i1，j1）、（i2，j2），则=

；步骤h中，按照

，…，

的顺序求取所有物理服务器序列的空调能耗成本cost的值，选取第一个最小cost值所对应的物理服务器序列来创建所有虚拟机。 in The number of partitions included in the physical server resource sequence to meet the demand,

Indicates the topological distance between the two partitions where the physical server meets the conditions; , indicating the sum of the topological distances between two partitions where the physical servers meet the conditions; it is characterized in that: in the step g, if the calculated value of the air conditioning energy consumption cost of the physical server sequence set is 1, it means The collection of physical server sequences is within the same physical server partition; if the physical server partition

,

The coordinates of are respectively (i1, j1), (i2, j2), then =

; In step h, according to

,...,

Calculate the air-conditioning energy cost cost values of all physical server sequences in the order of , and select the physical server sequence corresponding to the first minimum cost value to create all virtual machines.

采用公式

进行计算，是物理服务器分区在横向、纵向间距均相等的条件下得出的。 For the calculation formula of air-conditioning energy cost cost, it is marked by the sum of the number of partitions and the topological distance between two partitions, which can better reflect the concentration of physical servers in the server sequence, and can better reflect the actual energy consumption. for

use the formula

The calculation is obtained under the condition that the physical server partitions are equal in horizontal and vertical spacing.

、

和标示，分别采用以下公式进行计算： In the energy-saving cloud computing data center virtual machine placement method of the present invention, in step d, the available resources of the physical server include available resources of CPU, memory and hard disk, which are respectively used

,

and marked and calculated using the following formulas:

、

和

，则认为该序列为满足需求的物理服务器资源序列。 Among them, the bottom right marked as total indicates the total amount of resources, used indicates the amount of used resources, and threshold indicates the set reserved threshold; in step e, if one or more physical computers have CPU, memory and The available resources of the hard disk are respectively greater than or equal to

,

and

, the sequence is considered to be a physical server resource sequence that meets the requirements.

In a physical server, setting corresponding reserved thresholds for the use of CPU, memory, and hard disk is necessary to ensure the normal operation of the server. Therefore, the available amount of resources is not simply the difference between the total amount of known resources and the currently used amount. The reservation threshold defined by each resource should be subtracted from this difference.

The beneficial effects of the present invention are: in the cloud computing data center virtual machine placement method of the present invention, a physical server that meets the requirements is preferentially selected to create a virtual machine; If the former two do not exist, the most concentrated physical server sequence is finally selected to create the virtual machine, realizing the most energy-saving virtual machine placement method. In the cloud computing data center virtual machine placement method of the present invention, by establishing the calculation formula of air-conditioning energy consumption cost, the energy consumption cost of air-conditioning can be effectively and accurately represented by the concentration degree of physical servers, so as to select the most energy-saving virtual machine. Machine placement plan.

Description of drawings

Fig. 1 is the flow chart of the virtual machine placing method of the present invention;

Fig. 2 is a schematic structural diagram of a physical server partition in the present invention;

FIG. 3 is a schematic diagram of calculating topological distances between different physical server partitions in the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

As shown in Fig. 1, Fig. 2 and Fig. 3, respectively provide the flowchart of virtual machine placement method, the structural diagram of physical server partition and the topological distance calculation diagram between different partitions, the cloud computing data center virtual machine of the present invention The placement method includes the following steps:

a. Establish a physical server area, divide all physical servers covered by an air outlet of the air conditioning and refrigeration system into the same physical server partition, and form a physical server area based on cloud computing; and set the physical server area contained in the horizontal and vertical directions. The number of server partitions is x and y respectively, and both x and y are positive integers;

Suppose the location of the physical server partition is represented by two-dimensional coordinates P=(i, j); the air-conditioning energy consumption cost is marked by cost, defined as:

为满足需求的物理服务器资源序列中所包含的分区的个数，

标示满足条件的物理服务器所在两分区之间的拓扑距离；

，标示满足条件的物理服务器所在分区两两分区之间的拓扑距离的总和； in

The number of partitions included in the physical server resource sequence to meet the demand,

Indicates the topological distance between the two partitions where the physical server meets the conditions;

, indicating the sum of the topological distances between two partitions of the partitions where the physical servers meet the conditions;

b. Obtain the physical server information, obtain the resource usage information and partition information of all physical servers in step a; the resource usage information of the physical server is CPU, memory and hard disk information, and the partition information is the coordinate information of the physical server;

、和

； c. Calculate the virtual machine resources to be created, obtain the CPU, memory and hard disk size of each virtual machine to be created, and calculate the requested resources required to create all virtual machines; set the required CPU, memory and hard disk requested resources as

, and

;

d. Sort the physical servers, and sort all the physical servers according to the size of the available resources to form a sequence set of physical servers, which is denoted as ;

、和

标示，分别采用以下公式进行计算： The available resources of the physical server include the available resources of CPU, memory and hard disk, which are respectively used

, and

marked and calculated using the following formulas:

Among them, the bottom right marked as total indicates the total amount of resources, used indicates the amount of used resources, and threshold indicates the set reserved threshold;

，

，…，

，其中

表示包含n个物理服务器且可用资源满足需求的所有序列的集合，n为正整数； e. Establish a mapping between the virtual machine to be created and the physical server, and search for a sequence of physical server resources that meet the requirements on the basis that the available resources meet the resources requested by the virtual machine to be created; and sequence the resources according to the number of physical servers included in the sequence Classify to form a set of physical server sequences, which are respectively denoted as

,

,...,

,in

Indicates the collection of all sequences that contain n physical servers and available resources meet the requirements, n is a positive integer;

、和

，则认为该序列为满足需求的物理服务器资源序列； In this step, if the available resources of CPU, memory and hard disk of one or more physical computers are greater than or equal to

, and

, the sequence is considered to be a physical server resource sequence that meets the requirements;

存在，则

中任何一台物理服务器的可用资源均可创建出所有虚拟机，选择其中任一物理服务器作为待创建虚拟机的节点；如不存在集合

，则执行步骤g； f. Determine whether there is a single physical server to meet the demand, if there is a collection

exist, then

All virtual machines can be created from the available resources of any one of the physical servers, and any one of the physical servers can be selected as the node of the virtual machine to be created; if there is no set

, then execute step g;

，…，

，寻找出位于同一物理服务器分区内的物理服务器序列，在该物理服务器序列上创建所有虚拟机；如不存在处于同一分区内的满足要求的物理服务器序列，则执行步骤h； g. Determine whether there is a physical server sequence that meets the requirements in the same partition, and traverse the physical server sequence set

,...,

, find out the physical server sequence in the same physical server partition, and create all virtual machines on the physical server sequence; if there is no physical server sequence in the same partition that meets the requirements, go to step h;

In this step, if the calculated value of the air-conditioning energy consumption cost of the physical server sequence set is 1, it indicates that the physical server sequence set is located in the same physical server partition;

h. Select the server sequence with the smallest air-conditioning energy consumption cost and the least number of physical servers to create all virtual machines; the air-conditioning energy consumption cost is inversely proportional to the concentration of physical servers;

、

的坐标分别为（i1，j1）、（i2，j2），则

=

；步骤h中，按照

，…，

的顺序求取所有物理服务器序列的空调能耗成本cost的值，选取第一个最小cost值所对应的物理服务器序列来创建所有虚拟机。 If the physical server is partitioned

,

The coordinates of are respectively (i1, j1), (i2, j2), then

=

; In step h, according to

,...,

Calculate the air-conditioning energy cost cost values of all physical server sequences in the order of , and select the physical server sequence corresponding to the first minimum cost value to create all virtual machines.

Wherein, the reserved threshold refers to the minimum remaining capacity of the CPU, memory, and hard disk set in order to ensure the normal operation of the physical server and reserve a certain expansion space for the resources running on it. The topological distance between partitions refers to the distance between two physical server partitions, which is used to measure the degree of concentration between partitions.

As shown in Figure 2, the physical servers in the same air outlet area of the air-conditioning and refrigeration system are divided into the same partition. During the process of placing the virtual machine, first determine whether there is a physical server with available resources that meet the creation requirements. If so, Then create a virtual machine on the physical server; thus, the virtual machines can be centralized, which facilitates uniform heat dissipation and achieves effective energy saving. If there is no single physical server that meets the requirements, it is judged whether there is a server sequence in the same partition that meets the creation requirements. If so, the server sequence in the same partition is used for creation, which is also conducive to the concentration of physical servers. It is convenient for the air-conditioning and refrigeration system to carry out temperature control uniformly, and is also conducive to energy saving.

来说，=2，=

，则cost=5。还存在3台物理服务器满足创建所有虚拟机的要求，其中1台位于分区Region9中，余下的两台位于分区Region12中。由于第一种情形下为2台服务器，第二种情形为3台服务器，则选取物理服务器数目最少的序列来放置虚拟机，以实现最佳的节能效果。 If the above-mentioned single physical server or server sequence in the same partition does not exist, then use the calculation of the cost of air conditioning energy consumption to select the most energy-efficient physical server sequence for placing virtual machines. As shown in Figure 3, it is assumed that there are two physical servers that meet the requirements for creating all virtual machines, which are located in Region9 and Region12 respectively. For the formula

say, =2, =

, then cost=5. There are also 3 physical servers that meet the requirements for creating all virtual machines, one of which is located in the partition Region9, and the remaining two are located in the partition Region12. Since there are 2 servers in the first case and 3 servers in the second case, the sequence with the least number of physical servers is selected to place the virtual machines to achieve the best energy saving effect.

、

、

），其分别处于分区Region4、Region9和Region12中，则在计算空调能耗成本cost的过程中，

=3，

=

+

+

=+

+

=5+

，这样既可计算出空调能耗成本cost的大小。 As shown in Figure 3, if the server sequence that meets the creation requirements contains 3 physical servers (set their coordinate points as

,

,

), which are respectively located in Region4, Region9 and Region12, then in the process of calculating the energy cost of the air conditioner,

=3,

=

+

+

= +

+

=5+

, so that the cost of air-conditioning energy consumption can be calculated.

As a specific experimental application, the topology of the cloud computing data center is shown in Figure 2. There are 6 server partitions, each partition is equipped with 4 physical servers, and a single physical server is equipped with an Intel Xeon E5620 2.4G 4-core 8-thread processor , 32GB RAM, SAS 2.0T RAID5 hard disk, the physical server uses the Ubuntu Server 12.04 LTS operating system, and Xen Server5.5 is used as the virtualization software. The virtual machine creation adopts a unified specification, and allocates Intel Xeon E5620*1 CPU, 2G memory, 30G The hard disk runs Windows Server 2003 SP1 operating system. The cooling adopts HIRES precision air conditioner Super Precision HFDC/HFUC 0070, the power of the host is 6.5KW, and the power of the fan is 0.25KW.

The test adopts the random virtual machine placement method commonly used in the current data center and the virtual machine placement method proposed by the present invention. After 10 times of 50 virtual machine placement experiments each time, the method of the present invention can be used to centrally place the virtual machines in 1-2 physical In the partition, the random placement method generally places the virtual machine in 3 or more physical partitions. Since the placement of the virtual machine will cause the temperature of the partition to rise and trigger the work of the air outlet of the air conditioner, the method of the present invention can effectively reduce the temperature rise. The number of areas is high, thereby reducing the energy consumption of air conditioners. It is estimated that in this test environment, more than 10% of air conditioner power can be saved.

Claims (3)

1. an energy-saving cloud computing data center virtual machine placement method, is characterized in that, comprises the following steps: a. Establish a physical server area, divide all physical servers covered by an air outlet of the air conditioning and refrigeration system into the same physical server partition, and form a physical server area based on cloud computing; and set the physical server area contained in the horizontal and vertical directions. The number of server partitions is x and y respectively, and both x and y are positive integers; b. Obtain physical server information, obtain resource usage information and partition information of all physical servers in step a; c. Calculate the virtual machine resources to be created, obtain the CPU, memory and hard disk size of each virtual machine to be created, and calculate the requested resources required to create all virtual machines; set the required CPU, memory and hard disk requested resources as

,

and

; d. Sort the physical servers, and sort all the physical servers according to the size of the available resources to form a sequence set of physical servers, which is denoted as ; e. Establish a mapping between the virtual machine to be created and the physical server, and search for a sequence of physical server resources that meet the requirements on the basis that the available resources meet the resources requested by the virtual machine to be created; and sequence the resources according to the number of physical servers included in the sequence Classify to form a set of physical server sequences, which are respectively denoted as

,

,...,

,in

Indicates the collection of all sequences that contain n physical servers and available resources meet the requirements, n is a positive integer; f. Determine whether there is a single physical server to meet the demand, if there is a collection

exist, then All virtual machines can be created from the available resources of any one of the physical servers, and any one of the physical servers can be selected as the node of the virtual machine to be created; if there is no set

, then execute step g; g. Determine whether there is a physical server sequence that meets the requirements in the same partition, and traverse the physical server sequence set

,...,

, find out the physical server sequence in the same physical server partition, and create all virtual machines on the physical server sequence; if there is no physical server sequence in the same partition that meets the requirements, go to step h; h. Select the server sequence with the smallest air-conditioning energy consumption cost and the least number of physical servers to create all virtual machines; the air-conditioning energy consumption cost is inversely proportional to the concentration of physical servers. 2. The energy-saving cloud computing data center virtual machine placement method according to claim 1, assuming that the location of the physical server partition is represented by two-dimensional coordinates P=(i, j); the energy consumption cost of the air conditioner is marked by cost, defined as: in

The number of partitions included in the physical server resource sequence to meet the demand,

Indicates the topological distance between the two partitions where the physical server meets the conditions;

, indicating the sum of the topological distances between two partitions where the physical servers meet the conditions; it is characterized in that: in the step g, if the calculated value of the air conditioning energy consumption cost of the physical server sequence set is 1, it means The collection of physical server sequences is within the same physical server partition; if the physical server partition

,

The coordinates of are respectively (i1, j1), (i2, j2), then

=

; In step h, according to

,...,

Calculate the air-conditioning energy cost cost values of all physical server sequences in the order of , and select the physical server sequence corresponding to the first minimum cost value to create all virtual machines. 3. The energy-saving cloud computing data center virtual machine placement method according to claim 1 or 2, characterized in that: in step d, the available resources of the physical server include available resources of CPU, internal memory and hard disk, which are respectively used

,

and

marked and calculated using the following formulas:

Among them, the bottom right marked as total indicates the total amount of resources, used indicates the amount of used resources, and threshold indicates the set reserved threshold; In step e, if the available resources of CPU, memory and hard disk of one or more physical computers are respectively greater than or equal to

,

and

, the sequence is considered to be a physical server resource sequence that meets the requirements.

Images