WO2014110743A1

WO2014110743A1 - Method and device for resource scheduling between data centers

Info

Publication number: WO2014110743A1
Application number: PCT/CN2013/070556
Authority: WO
Inventors: 杨宇; 赵永利; 张�杰; 杨辉; 林毅; 韩建蕊
Original assignee: 华为技术有限公司
Priority date: 2013-01-16
Filing date: 2013-01-16
Publication date: 2014-07-24
Also published as: CN103443771A; CN103443771B

Abstract

The present invention relates to the field of data resources. Embodiments of the present invention provide a method and a device for resource scheduling between data centers, which can avoid resource waste and resource insufficiency in the data centers, improve the adaptability of the data centers to load changes, achieve resource adjustment without interrupting a service, and reduce a network blocking rate. The method specifically comprises: performing unified management and scheduling of computing resources by using an adaptive resource scheduling method between data centers. The present invention is applicable to scheduling of virtual resources between data centers.

Description

Data center inter-database resource scheduling method and device

Technical field

The present invention relates to the field of data resources, and in particular, to a resource scheduling method and device between data centers. Background technique

At present, huge data centers are established around the world to realize massive computing tasks and storage of massive information, to provide users with cloud computing services, and to achieve success in commercial applications. The rapid development of cloud computing technology has led to the transformation of data centers, which can effectively integrate resources, improve resource utilization, save energy, and reduce operating costs. The data center is the core of the information system. Its main function is to provide information services to users through the network. Data centers use virtualization technology to build resources into dynamic virtual resource pools. Virtual resource management technologies enable automatic deployment, dynamic expansion, and on-demand allocation of cloud computing resources.

Virtualization technology is one of the technical cornerstones of building a cloud computing environment, especially server virtualization technology, which provides a powerful means of implementation for cloud computing at the infrastructure layer. At present, in the application field of server virtualization technology, artificial static resource allocation methods are often used among data centers. The inventor found that under the demand, the number of virtual machines running in the data center is increasing, and the requirements for resource scheduling of virtual machine clusters are getting higher and higher. In a large-scale virtual machine cluster, the number of virtual machines and the load of virtual machines will change frequently with the needs of users and applications. However, artificial dynamic resource adjustments will have obvious lags, and services need to be interrupted during the adjustment process. The quality of service of the customer, in addition, the current resource scheduling method only considers the application resources or only considers the network resources, so that the virtual resource allocation on the physical machine is not balanced, the data center load is unbalanced, and resource utilization occurs in the virtual machines of some data centers. The rate is decreased or the network blocking rate is increased. Therefore, in the application of the existing data center scheduling method, the data center does not have good adaptability to load changes, and the data center may waste resources or resources. The situation of the foot. Summary of the invention

The embodiments of the present invention provide a resource scheduling method and device between data centers, which can avoid resource waste or insufficient resources in the data center, improve the adaptability of the data center to load changes, and achieve non-stop service. Adjust resources and reduce network blocking rate.

In order to achieve the above object, the embodiment of the present invention adopts the following technical solutions:

The first aspect provides a data center resource scheduling method, including:

Obtain an application resource occupancy rate of each data center, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate;

Determining whether the average application resource occupancy rate of the data center exceeds a load threshold upper limit or is lower than a load threshold lower limit;

Selecting, in the data center, a source data center for migrating load data according to the judgment result;

Selecting, in the data center, a destination data center for moving in load data according to the to-be-migrated load data of the source data center and the application resource occupancy rate and network resource occupancy rate of each data center;

The load data is migrated from the source data center, and the load data is moved into the destination data center.

In a first possible implementation, in combination with the first aspect, the selecting, by the determining, a source data center for migrating load data in the data center includes: if an average application resource occupancy rate of the data center If the upper limit of the load threshold is exceeded, the data center with the largest load of the application resource is selected in the data center, and the data center with the largest load of the application resource is used as the source data center for migrating the load data.

In a second possible implementation, in combination with the first aspect, the load data to be migrated according to the source data center and the application resource occupancy rate and network resource occupancy rate of each data center are in the Selected in the data center for moving in load data The destination data center includes:

Calculating a load of the source data center, and determining the to-be-migrated load data according to the load threshold upper limit;

Selecting, from the data center, at least one data center whose load is lower than the average application resource occupancy of the data center as a resource scheduling data center set;

Obtaining a network resource occupancy rate of each of the data in the resource scheduling data center set, and calculating, according to the application resource occupancy rate and the network resource occupancy rate of each data center in the resource scheduling data center set. The resource schedules a comprehensive weight of each data center in the data center set;

Calculating the immigration load data that can be tolerated by each data center in the resource scheduling data center set, and selecting the data center with the smallest integrated weight from the resource scheduling data center set, if the data center with the smallest integrated weight is available If the load data to be migrated is greater than the load data to be migrated from the source data center, the data center with the smallest integrated weight is used as the destination data center.

In a third possible implementation, combined with the second possible implementation, the following:

If the data load that the data center with the smallest integrated weight can bear is less than the load data to be moved out of the source data center, the data center with the smallest integrated weight is used as the destination data center;

Selecting, in the resource scheduling data center set, the data center with the second smallest weight, and using the data center with the smaller overall weight as the other destination data center until selected in the resource scheduling data center set. The immigration load data that the data center can bear is greater than or equal to the remaining load data to be migrated in the source data center.

In a fourth possible implementation, in combination with the first aspect, the selecting, in the data center, the source data center used to migrate the load data according to the determination result includes: if the average application resource occupancy rate of the data center Below the lower limit of the load threshold, the data center with the least load of the application resource is selected in the data center, and the data center with the smallest load of the application resource is used as the source data center for migrating the load data. In a fifth possible implementation, in combination with the first aspect, the load data to be migrated according to the source data center and the application resource occupancy rate and network resource occupancy rate of each data center are in the The destination data center selected for moving into the load data in the data center includes:

Calculating a load of the source data center, and using the load of the source data center as load data to be migrated;

Selecting, from the data center, at least one data center with a load higher than an average application resource occupancy of the data center as a resource scheduling data center set;

Obtaining a network resource occupancy rate of each data center in the resource scheduling data center set, and calculating, according to the application resource occupancy rate and the network resource occupancy rate of each data center in the resource scheduling data center set, Resource scheduling The overall weight of each data center in the data center collection;

Calculating the load data that can be tolerated by each data center in the data scheduling data center set, and selecting the data center with the largest comprehensive weight from the resource scheduling data center set, if the data center with the largest comprehensive weight can be If the load data to be migrated is greater than the data to be moved out of the source data center, the data center with the largest integrated weight is used as the destination data center.

In a sixth possible implementation, in combination with the fifth possible implementation, the method includes:

If the data load that the data center with the largest integrated weight can bear is less than the load data to be migrated from the source data center, the data center with the largest integrated weight is used as the destination data center;

Selecting, in the resource scheduling data center set, the data center with the second largest integrated weight, and using the data center with the second largest weight as the other destination data center until the data selected in the resource scheduling data center set The load data that the center can bear is greater than or equal to the remaining load data to be moved out of the source data center.

The second aspect provides a resource scheduling device, including:

a data processing unit, configured to acquire an application resource occupancy rate of each data center, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate; a management unit, configured to determine whether an average application resource occupancy rate of the data center forwarded by the data processing unit exceeds an upper limit of a load threshold or is lower than a lower limit of a load threshold;

a data center obtaining unit, configured to select, according to the judgment result forwarded by the management unit, a source data center for migrating load data;

The data center acquiring unit is further configured to select, according to the to-be-migrated load data of the source data center, the application resource occupancy rate and network resource occupancy rate of each data center, in the data center. a destination data center that is moved into the load data; the data processing unit is further configured to receive the source data center and the destination data center selected by the data center acquiring unit, and move out load data from the source data center, The load data is migrated to the destination data center.

In a first possible implementation manner, in combination with the second aspect, the data center obtaining unit includes:

a first acquiring sub-unit, configured to: if the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold, select a data center with the largest load of application resources in the data center, and load the application resource The largest data center acts as the source data center for moving out of load data.

In a second possible implementation, in combination with the second aspect, the data center obtaining unit further includes:

a first data processing sub-unit, configured to calculate a load of the source data center, and determine the to-be-migrated load data according to the upper limit of the load threshold;

The first obtaining subunit is further configured to select, from the data center, at least one data center with a load lower than an average application resource occupancy of the data center as a resource scheduling data center set;

The first data processing sub-unit is further configured to acquire a network resource occupancy rate of each data center in the resource scheduling data center set, and schedule the application resource of each data center in the data center set according to the resource scheduling Calculating the comprehensive weight of each data center in the resource scheduling data center set by using the occupancy rate and the network resource occupancy rate; the first obtaining subunit is further configured to calculate the resource scheduling data center set The data load that can be tolerated by each data center is selected, and the data center with the smallest integrated weight is selected from the resource scheduling data center set, and if the data center with the smallest comprehensive weight can bear the load data greater than When the load data of the source data center is to be moved out, the data center with the smallest integrated weight is used as the destination data center.

The first obtaining subunit is further configured to: if the data load that the data center with the smallest integrated weight can bear is less than the load data to be moved out of the source data center, the data center with the smallest integrated weight is used. As a destination data center;

The first obtaining subunit is further configured to select, in the resource scheduling data center set, the data center with a small integrated weight, and use the data center with a small integrated weight as another destination data center until The migrated load data that the data center selected in the resource scheduling data center set can bear is greater than or equal to the remaining load data to be migrated in the source data center.

In a fourth possible implementation, in combination with the second aspect, the data center obtaining unit further includes:

a second acquiring sub-unit, configured to: if the average application resource occupancy rate of the data center is lower than the lower limit of the load threshold, select a data center with a minimum load of application resources in the data center, and use the application resource The data center with the least load acts as the source data center for migrating load data.

In a fifth possible implementation, in combination with the second aspect, the data center acquisition unit further includes:

a second data processing sub-unit, configured to calculate a load of the source data center, and use a load of the source data center as load data to be migrated;

The second obtaining subunit is further configured to select, from the data center, at least one data center with a load higher than an average application resource occupancy of the data center as a resource scheduling data center set;

The second data processing sub-unit is further configured to acquire a network resource occupancy rate of each data center in the resource scheduling data center set, and perform scheduling data according to the resource Calculating the integrated resource weight of each data center in the resource scheduling data center set by using the application resource occupancy rate and the network resource occupancy rate of each data center in the central set; the second acquiring subunit is further used for calculating The resource scheduling data center can receive the load data that can be tolerated in each data center, and select the data center with the largest comprehensive weight from the resource scheduling data center set, if the data center with the largest comprehensive weight can withstand If the migration load data is greater than the load data to be migrated from the source data center, the data center with the largest integrated weight is used as the destination data center.

In a sixth possible implementation manner, in combination with the fifth possible implementation manner, the data center obtaining unit of the target includes:

The second obtaining sub-unit is further configured to: if the data load that the data center with the largest comprehensive weight can bear is less than the load data to be moved out of the source data center, the data center with the largest comprehensive weight As a destination data center;

The second obtaining subunit is further configured to select the data center with the second largest integrated weight in the resource scheduling data center set, and use the data center with the second largest integrated weight as another destination data center until The migrated load data that the data center selected in the resource scheduling data center set can bear is greater than or equal to the remaining load data to be migrated in the source data center.

In a third aspect, a resource scheduling device is provided, including: at least one processor, a memory, a communication interface, and a bus, wherein the at least one processor, the memory, and the communication interface are connected by a bus and complete communication with each other, wherein:

The processor is configured to acquire an application resource occupancy rate of each data center by using the at least one communication interface, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate;

The processor is configured to determine whether an average application resource occupancy rate of the data center that is forwarded by the processor exceeds a load threshold upper limit or is lower than a load threshold lower limit; and the processor is further configured to determine, according to the memory forwarding Resulting, by the at least one communication interface, selecting a source data center for migrating load data in the data center;

The processor is further configured to: according to the data to be moved out of the source data center, The application resource occupancy rate and network resource occupancy rate of each data center are selected by the at least one communication interface to select a destination data center for moving in load data in the data center;

The processor is further configured to receive, by the at least one communication interface, the source data center and the destination data center selected by the processor, and migrate load data from the source data center, where the load data is Move to the destination data center.

In a first possible implementation, in combination with the third aspect, the method includes:

The processor is further configured to: if the average application resource occupancy of the data center exceeds the upper limit of the load threshold, select, by using the at least one communication interface, a data center with the largest load of application resources in the data center, The data center with the largest load of the application resources is used as a source data center for migrating load data.

In a second possible implementation, in combination with the third aspect, the following:

The processor is further configured to calculate a load of the source data center, and determine the to-be-migrated load data according to the load threshold upper limit; select at least one load from the data center by using the at least one communication interface a data center that is lower than the average application resource occupancy of the data center as a resource scheduling data center set;

The processor is further configured to acquire, by using the at least one communication interface, a network resource occupancy rate of each data center in the resource scheduling data center set, and schedule each data center in the data center set according to the resource scheduling Calculating the comprehensive weight of each data center in the resource scheduling data center set by using the application resource occupancy rate and the network resource occupancy rate;

The processor is further configured to calculate, in the resource scheduling data center set, the load data that can be tolerated by each data center, and select the comprehensive weight from the resource scheduling data center set by using the at least one communication interface. The smallest data center, if the data load that the data center with the smallest integrated weight can bear is greater than the data to be moved out of the source data center, the data center with the smallest integrated weight is used as the destination data center.

In a third possible implementation manner, combined with the second possible implementation manner, the following: The processor is further configured to: if the data load that can be tolerated by the data center with the smallest integrated weight is less than the load data to be moved out of the source data center, use the data center with the smallest integrated weight as the destination data. center;

The processor is further configured to select, by using the at least one communication interface, the data center with a small integrated weight in the resource scheduling data center set, and use the data center with a small overall weight as another purpose The data center, until the data center selected in the resource scheduling data center set can bear the inbound load data greater than or equal to the remaining data to be migrated in the source data center.

In a fourth possible implementation, in combination with the third aspect, the method includes:

The processor is further configured to: if the average application resource occupancy of the data center is lower than the lower limit of the load threshold, select, by using the at least one communication interface, a data center with a minimum load of application resources in the data center. The data center with the least load of the application resource is used as a source data center for migrating load data.

In a fifth possible implementation, in combination with the third aspect, the method includes:

The processor is further configured to calculate a load of the source data center, and use a load of the source data center as load data to be migrated; select at least one load from the data center by using the at least one communication interface. a data center that is higher than the average application resource occupancy of the data center as a resource scheduling data center set;

The processor is further configured to calculate, in the resource scheduling data center set, the load data that can be tolerated by each data center, and select the comprehensive weight from the resource scheduling data center set by using the at least one communication interface. The largest data center, if the data load that the data center with the largest comprehensive weight can bear is greater than the data to be moved out of the source data center, the data center with the largest integrated weight is used as the destination data center. In a sixth possible implementation, the fifth possible implementation is combined, including:

The processor is further configured to: if the data load that can be tolerated by the data center with the largest comprehensive weight is smaller than the load data to be moved out of the source data center, use the data center with the largest integrated weight as the destination data. Center

The processor is further configured to select, by using the at least one communication interface, the data center with the second largest integrated weight in the resource scheduling data center set, and use the data center with the second largest weight as another destination data. The mobility load data that can be tolerated by the data center selected in the resource scheduling data center set is greater than or equal to the remaining load data to be migrated in the source data center.

The method and device for scheduling resources between data centers provided by the embodiments of the present invention can avoid resource waste and resource shortage in the data center by adopting adaptive resource scheduling between data centers, and improve the adaptation of the data center to load changes. Sexuality, resource-free adjustment can be achieved without interrupting the business, and the network blocking rate can be reduced. DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below. Obviously, the drawings in the following description are only It is a certain embodiment of the present invention, and other drawings can be obtained from those skilled in the art without any creative work.

1 is a schematic flowchart of a resource scheduling method between data centers according to an embodiment of the present invention;

2 is a schematic flowchart of a resource scheduling method between data centers according to another embodiment of the present invention;

FIG. 3 is a schematic flowchart of another resource center resource scheduling method according to another embodiment of the present invention; FIG.

4 is a schematic structural diagram of a cross-layer optimization control system according to an embodiment of the present invention; FIG. 5 is a schematic structural diagram of a resource scheduling device according to an embodiment of the present invention; FIG. 6 is a schematic structural diagram of another resource scheduling device according to an embodiment of the present invention; FIG. 7 is still another resource scheduling device according to an embodiment of the present invention; Schematic. detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

An embodiment of the present invention provides a data center resource scheduling method. Referring to FIG. 1, the method includes the following steps:

101. Obtain an application resource occupancy rate of each data center, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate.

The average application resource usage of the data center is obtained according to the Central Processing Unit (CPU) and memory utilization. If the CPU usage of the data center is n and the memory usage is m, the resource usage of each data center is =n* s+m*( l -s). Where s is a configuration parameter with a value ranging from 0 to 1. It can be adjusted, specifically, according to the result of service blocking rate and network resource utilization, select the best value to set s. Then the average application resource occupancy of the data center = the sum of each data center resource occupancy rate / the total number of data centers.

102. Determine whether the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold or is lower than the lower limit of the load threshold.

Specifically, the upper limit of the load threshold and the lower limit of the load threshold are the upper threshold and the lower threshold of the application resource occupancy rate of the data center. The upper limit of the load threshold and the lower limit of the load threshold are set in advance, and the setting of the upper limit of the load threshold or the lower limit of the load threshold comprehensively considers application resources and network resources, and is specifically acquired by a specific device operation. Among them, application resources are mainly related to CPU and memory utilization, and network resources are mainly related to bandwidth utilization. Of course, the factors affecting application resources are not limited to this, and network resources are also delayed. It is related to factors such as jitter.

103. Select a source data center for migrating load data in the data center according to the judgment result.

104. Select, according to the data to be migrated in the source data center and the application resource occupancy rate and network resource occupancy rate of each data center, select a data center for moving in the load data in the data center.

105. The load data is moved out from the source data center, and the load data is moved into the destination data center.

Specifically, since the load data is migrated from the source data center and the load data is moved to the destination data center, the process needs to consider the bandwidth occupancy rate. Therefore, it is necessary to jointly consider the application resource occupancy rate and network resource occupancy rate of each data center. Select the destination data center to migrate into the load data in the data center.

The data center inter-database resource scheduling method provided by the embodiment of the present invention can avoid resource waste and resource shortage by adopting adaptive resource scheduling between data centers, and can realize resource adjustment and reduce the non-interrupted service. Network blocking rate.

Specifically, the embodiment of the present invention provides a method for scheduling resources between data centers, and the specific steps are as follows:

201. Obtain an application resource occupancy rate of each data center, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate.

Among them, the average application resource occupancy rate of the data center is obtained according to the CPU and memory utilization rate. If the CPU usage of the data center is n and the memory usage is m, the resource usage of each data center is =n* s+m* ( l -s). Where s is a configuration parameter with a value ranging from 0 to 1. It can be adjusted, specifically based on the result of the traffic blocking rate and the utilization rate of the whole network resources, and the optimal value is selected to set s. Then, the average application resource occupancy rate of the data center=the sum of the data center resource usage ratios/the total number of data centers. In this embodiment, virtual machine software (\'Ί rtual Machine Ware (VM Ware for short) can be used to obtain each. Application resource usage in the data center.

202. Determine whether the average application resource occupancy rate of the data center exceeds a load threshold. The upper limit is below the lower limit of the load threshold.

Specifically, the upper limit of the load threshold and the lower limit of the load threshold are the upper threshold and the lower threshold of the application resource occupancy rate of the data center. The load threshold upper limit and the load threshold lower limit are preset, and the load threshold upper limit or the load threshold lower limit is set in consideration of application resources and network resources, and is specifically acquired by a specific device operation. Among them, application resources are mainly related to CPU and memory utilization, and network resources are mainly related to bandwidth utilization. Of course, the factors affecting application resources are not limited to this. Network resources are also related to factors such as delay and jitter.

If the result of the determination in step 202 is that the average application resource occupancy of the data center exceeds the upper limit of the load threshold, steps 203 a to 212a are performed, as specifically shown in FIG. 2;

203 a. Select the data center with the largest load of application resources in the data center, and use the data center with the largest load of application resources as the source data center for moving out the load data.

204a. Calculate the load of the source data center, and determine the load data to be migrated according to the upper limit of the load threshold.

205a. Select, from the data center, at least one data center with a load lower than an average application resource occupancy of the data center as a resource scheduling data center set.

206a. Obtain a network resource occupancy rate of each data center in the resource scheduling data center set, and calculate each of the resource scheduling data center sets according to the application resource occupancy rate and the network resource occupancy rate of each data center in the resource scheduling data center set. The overall weight of the data center.

Among them, the comprehensive weight of each data center = « * application resource occupancy rate + network resource occupancy rate. Where “the sum is the weight coefficient, the specific value is determined according to the specific operating environment.

207a: Calculate the load data that each data center in the resource scheduling data center set can bear, and select the data center with the smallest integrated weight from the resource scheduling data center set.

Specifically, each data center can withstand the immigration load data = the data center's load threshold upper limit - the current load of the data center. Among them, the current load of the data center is mainly The CPU usage and memory usage of the data center are determined.

208a, determining whether the immigration load data that the data center with the smallest integrated weight can bear is less than the load data to be moved out of the source data center, and if the data center with the smallest integrated weight can bear the load data that is greater than or equal to the source data center. After the load data is moved out, step 209a is performed;

Specifically, the program logic structure that can be implemented by the computer processing system is taken as an example for description. A migrating data center set can be set, and by default, all destination data centers in the migpable data center set can independently carry the source data center. The load data is to be moved out.

209a. The data center with the smallest integrated weight is used as the destination data center, and the load data is moved out from the source data center, and the load data is moved into the destination data center.

Specifically, if the data center with the smallest integrated weight can independently carry the load data to be migrated from the source data center, the data center with the smallest integrated weight can be considered as a member of the migratable data center set. In this case, the data center can be migrated. If the set is not empty, the data center with the smallest integrated weight is selected as the member of the default destination data center set, and the load data to be moved out of the source data center is moved in the data center with the smallest integrated weight.

Alternatively, the result of the determination in step 208 is that the data load that the data center with the smallest integrated weight can bear is less than the load data to be moved out of the source data center, and steps 210a to 212a are performed.

210a. The data center with the smallest integrated weight is used as the destination data center.

In this case, the data center with the smallest integrated weight cannot independently carry the data to be migrated from the source data center, and the data center set can be migrated. Therefore, the destination data center required for the resource scheduling process is multiple. The data center with the smallest combined weight is a member of the default destination data center collection.

21 1 a , selecting a data center with a small integrated weight in the resource scheduling data center set, and using the data center with the smaller integrated weight as the other destination data center until the data center selected in the resource scheduling data center set can withstand The migration load data is greater than or equal to the remaining load data to be migrated in the source data center. In this process, the data center with the second smallest weight is also used as a member of the default destination data center set, and the process is repeated until the data load that the selected data center in the resource scheduling data center set can withstand is greater than or The data of the to-be-migrated load data of the source data center is equal to or greater than the data center of the data center to be migrated.

212a, migrating load data from the source data center, and migrating the load data to all selected destination data centers.

If the migrating data center collection is not empty, the data center with the load data to be moved out of the load data that is greater than or equal to the remaining data in the source data center is used as a member of the default destination data center set, and the source data is used. The load data to be moved out of the center is moved to the destination data center in the destination data center set.

If the result of the determination in step 202 is that the data center average load is lower than the load threshold lower limit, steps 203b to 212b are performed, as shown in FIG.

203b: Select a data center with the least load of the application resource in the data center, and use a data center with the smallest load of the application resource as the source data center for migrating the load data.

204b. Calculate the load of the source data center and use the load of the source data center as the data to be moved out.

205b. Select, from the data center, at least one data center with a load higher than an average application resource occupancy of the data center as a resource scheduling data center set.

206b. Obtain a network resource occupancy rate of each data center in the resource scheduling data center set, and calculate each of the resource scheduling data center sets according to the application resource occupancy rate and the network resource occupancy rate of each data center in the resource scheduling data center set. The overall weight of the data center.

207b, computing resource scheduling data center collection in each data center can withstand the migration Into the load data, select the data center with the largest comprehensive weight from the resource scheduling data center set.

Specifically, specifically, each data center can withstand the load data of the load = the upper limit of the load threshold of the data center - the current load of the data center. The current load of the data center is mainly determined by the CPU usage and memory usage of the data center.

208b, determining whether the data load that the data center with the largest comprehensive weight can bear is less than the load data to be moved out of the source data center, and if the data load that the data center with the largest comprehensive weight can bear is greater than or equal to the data center of the source data center After the load data to be migrated, step 209b is performed;

209b. The data center with the largest integrated weight is used as the destination data center, and the load data is moved out from the source data center, and the load data is moved into the destination data center.

Specifically, if the data center with the largest integrated weight can independently carry the load data to be migrated from the source data center, the data center with the largest integrated weight can be considered as a member of the migrated data center set. In this case, the data center can be migrated. If the set is not empty, the data center with the largest integrated weight is selected as the member of the default destination data center set, and the load data to be moved out of the source data center is moved in the data center with the largest integrated weight.

Alternatively, the result of the determination in step 208b is that the data load that the data center with the largest integrated weight can bear is less than the load data to be migrated from the source data center, and steps 210b to 212b are performed.

210b. The data center with the largest integrated weight is used as the destination data center.

In this case, the data center with the largest integrated weight cannot independently carry the load data to be migrated from the source data center, and the data center set can be migrated. Therefore, the destination data center required for the resource scheduling process is multiple. The data center with the largest combined weight is a member of the default destination data center collection. 21 1b , selecting the data center with the second largest comprehensive weight in the resource scheduling data center set, and using the data center with the second largest weight as the data center of another destination, until the data center selected in the resource scheduling data center set can withstand the migration The inbound load data is greater than or equal to the remaining load data to be migrated in the source data center.

In this process, the data center with the second largest weight is also used as a member of the default destination data center set, and the process is repeated until the data load that the selected data center in the resource scheduling data center set can withstand is greater than or The data of the to-be-migrated load data of the source data center is equal to or greater than the data center of the data center to be migrated.

212b, migrating load data from the source data center and migrating the load data to the destination data center.

The data center in this embodiment is divided into two states, namely, an operating state and a dormant state. After the source data center in the running state enters the destination data center, the source data center automatically enters a sleep state, so that the standby state consumes various resources. At the lowest level, avoid wasting resources. The data centers in the resource scheduling process are all in the running data center. When the average load of the data center is lower than the lower limit of the load threshold, the load data is migrated from the source data center, and the load data is moved to the destination data center. Specifically, the resource balance of the data center is embodied on the physical machine, that is, when When there are few virtual machines on the physical machine and the application resources used on the physical machine are low, the virtual machines on the physical machine are collectively migrated to other physical machines, and the idle physical machines are shut down; when there are many virtual machines on the physical machine When the application resources used on the physical machine are too high, some virtual machines are migrated to other relatively idle physical machines, or more standby physical machines are started and some virtual machines are migrated to the standby physical machines, thereby realizing the physical machines. Scheduling of data center resources, because the virtual machine directly interacts with the client, so The load changes with time, responds to the change of the virtual machine load in time, and allocates more abundant resources to the high-loaded virtual machine through the scheduling of resources between the data centers to meet the virtual machine's demand for resources, thus achieving energy saving and improving energy. The purpose of the ratio.

The data center resource scheduling method provided by the embodiment of the present invention can avoid resource waste and resource shortage in the data center by adopting adaptive resource scheduling between data centers, and improve the adaptability of the data center to load changes. Reaching resources without interruption can achieve resource adjustment and reduce network blocking rate. At the same time, the quality of service of users can be improved.

The application of the inter-data resource scheduling method in this embodiment is based on the cross-layer optimization control system. Referring to FIG. 4, the cross-layer optimization control system includes: the resource scheduling device 3 1 provided by the embodiment of the present invention, and the user processes The device 32 and the database 33, wherein: the user processing device 32 is configured to provide an interface between the data center and the user plane provided by the embodiment of the present invention, process the service request of the user, and authenticate the service request authority of the user.

The resource scheduling device 3 1 is configured to implement the method according to the embodiment of the present invention to perform equalization and regulation on resources of the virtual data center in the database, including evaluating resource load status of the virtual machine for network characteristics of different types of resources, and according to the resource scheduling method. Select the virtual machine that needs to perform resources to realize the joint optimization of resources. This module is the core module of the resource scheduling device. It senses the CPU and memory of the virtual resource and monitors the load of all virtual machines in real time. At the same time, the services that need resource scheduling are cached to ensure that the user services are continuously running during the migration process; the network resources are perceived, the network resource information is provided to the cross-layer optimization control system, and the resources are jointly optimized. At the same time, the calculation and resource reservation of the network layer path based on conditional constraints are completed. And the application resources processed in the data center to realize the transmission of application resources.

The resource scheduling device 4 provided by the embodiment of the present invention is applied to the cross-layer optimization control system to implement the foregoing data center resource scheduling method. Referring to FIG. 5, the device includes: a data processing unit 41, a management unit 42 and The data center obtaining unit 43 has a towel: The data processing unit 41 is configured to acquire an application resource occupancy rate of each data center, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate.

The management unit 42 is configured to determine whether the average application resource occupancy of the data center forwarded by the data processing unit 41 exceeds a load threshold upper limit or is lower than a load threshold lower limit.

The data center obtaining unit 43 is configured to select a source data center for migrating load data according to the judgment result forwarded by the management unit 42.

The data center obtaining unit 43 is further configured to select, in the data center, a destination data center for moving in the load data according to the to-be-migrated load data of the source data center and the application resource occupancy rate and the network resource occupancy rate of each data center.

The data processing unit 41 is further configured to receive the source data center and the destination data center selected by the data center acquiring unit 43 and migrate the load data from the source data center to move the load data into the destination data center.

The resource scheduling device provided by the embodiment of the present invention can avoid resource waste and resource shortage in the data center by adopting adaptive resource scheduling between data centers, thereby improving the adaptability of the data center to load changes, and achieving non-interruption. The business can adjust the resources and reduce the network blocking rate.

Further, as shown in FIG. 6, the data center obtaining unit 43 further includes: a first obtaining subunit 43 l a and a first data processing subunit 432a, where:

The first obtaining sub-unit 43 1 a is configured to: if the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold, select a data center with the largest load of application resources in the data center, and maximize the load of the application resource. The data center acts as the source data center for migrating load data.

The first data processing sub-unit 432a is configured to calculate a load of the source data center, and determine the to-be-migrated load data according to the upper limit of the load threshold.

The first obtaining sub-unit 43 1 a is configured to select, from the data center, at least one data center that has an average application resource occupancy lower than the data center as the resource scheduling data center set.

The first data processing sub-unit 432a is configured to acquire a network resource occupancy rate of each data center in the resource scheduling data center set, and schedule the data center collection according to the resource. The application resource occupancy rate and network resource occupancy rate of each data center calculate the comprehensive weight of each data center in the resource scheduling data center set.

The first obtaining subunit 431a is further configured to calculate the inbound load data that each data center in the resource scheduling data center set can bear, and select the data center with the smallest integrated weight from the resource scheduling data center set, if the data with the smallest integrated weight is If the load data that the center can bear is greater than the data to be moved out of the source data center, the data center with the smallest integrated weight is used as the destination data center.

The first obtaining sub-unit 431a is further configured to use the data center with the smallest integrated weight as the destination data center if the load data that can be accepted by the data center with the smallest integrated weight is less than the load data to be moved out of the source data center.

The first obtaining sub-unit 431a is further configured to select a data center with a small integrated weight in the resource scheduling data center set, and use the data center with the smaller weight as the other destination data center until the resource scheduling data center set The selected load data that the selected data center can bear is greater than or equal to the remaining load data to be moved out of the source data center. Optionally, as shown in FIG. 6, the device further includes:

The data center obtaining unit further includes: a second obtaining subunit 431b and a second data processing subunit 432b, wherein:

The second obtaining sub-unit 431b is configured to: if the average application resource occupancy rate of the data center is lower than the lower limit of the load threshold, select a data center with the least load of the application resource in the data center, and minimize the load of the application resource. The data center acts as a source data center for migrating load data.

The second data processing sub-unit 432b is configured to calculate the load of the source data center, and the load of the source data center is used as the data to be migrated out.

The second obtaining sub-unit 431b is further configured to select, from the data center, at least one data center with a load higher than the average application resource occupancy of the data center as the resource scheduling data center set.

The second data processing sub-unit 432b is further configured to acquire a resource scheduling data center set. The network resource occupancy rate of each data center is calculated, and the comprehensive weight of each data center in the resource scheduling data center set is calculated according to the application resource occupancy rate and the network resource occupancy rate of each data center in the resource scheduling data center set.

The second obtaining sub-unit 43 1 b is further configured to calculate the immigration load data that each data center in the resource scheduling data center set can bear, and select the data center with the largest comprehensive weight from the resource scheduling data center set, if the comprehensive weight is the largest The data load that the data center can withstand is greater than the data to be moved out of the source data center, and the data center with the largest integrated weight is used as the destination data center.

The second obtaining sub-unit 43 1 b is further configured to use the data center with the largest comprehensive weight as the destination data center if the load data that can be accepted by the data center with the largest integrated weight is less than the load data to be moved out of the source data center.

The second obtaining sub-unit 43 1 b is further configured to select a data center with the second largest integrated weight in the resource scheduling data center set, and use the data center with the second largest weight as the other destination data center until the data center is collected in the resource scheduling data center. The migrated load data that the selected data center can withstand is greater than or equal to the remaining load data to be moved out of the source data center.

The data center in this embodiment is divided into two states, namely, an operating state and a dormant state. After the source data center in the running state enters the destination data center, the source data center automatically enters a sleep state, so that the standby state consumes various resources. At the lowest level, avoid wasting resources. Among them, the data center in the process of resource scheduling is the data center in the running state. When the average load of the data center is lower than the lower limit of the load threshold, the load data is migrated from the source data center, and the load data is moved to the destination data center to reduce waste of resources.

The resource scheduling device provided by the embodiment of the present invention can avoid resource waste and resource shortage in the data center by adopting adaptive resource scheduling between data centers, thereby improving the adaptability of the data center to load changes, and achieving non-interruption. The business can adjust the resources and reduce the network blocking rate. At the same time, the quality of service of users can be improved.

An embodiment of the present invention provides a resource scheduling device 5. Referring to FIG. 7, the device includes: at least one processor 51, a memory 52, a communication interface 53, and a bus 54, At least one processor 51, memory 52 and communication interface 53 are connected by bus 54 and communicate with each other, wherein:

The bus 54 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Wait. The bus 54 can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is shown in Figure 7, but it does not mean that there is only one bus or one type of bus. Its towel:

Memory 52 is for storing executable program code, the program code including computer operating instructions. Memory 52 may contain high speed RAM memory and may also include non-volatile memory, for example: 3⁄4 port at least one disk memory.

The processor 51 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more configured to implement the embodiments of the present invention. integrated circuit.

The communication interface 53 is mainly used to implement communication between devices in this embodiment. The processor 51 is configured to obtain an application resource occupancy rate of each data center by using at least one communication interface 53, and calculate an average application resource occupancy rate of the data center according to the application resource occupancy rate.

The memory 52 is configured to determine whether the average application resource occupancy of the data center forwarded by the processor 51 exceeds a load threshold upper limit or is lower than a load threshold lower limit.

The processor 51 is further configured to select, in the data center, the source data center for migrating the load data through the at least one communication interface 53 according to the judgment result of the forwarding by the memory 52.

The processor 51 is further configured to select, in the data center, the data to be migrated into the data center according to the to-be-migrated load data of the source data center and the application resource occupancy rate and the network resource occupancy rate of each data center by using at least one communication interface 53. Destination data center.

The processor 51 is further configured to receive, by using the at least one communication interface 53, the source data center and the destination data center selected by the processor 51, and migrate the load data from the source data center, and The load data is moved into the destination data center.

Further, the processor 5 1 is further configured to: if the average application resource occupancy of the data center exceeds the upper limit of the load threshold, select, by using at least one communication interface 53, a data center with the largest load of application resources in the data center, The data center with the largest load of application resources is used as the source data center for migrating load data.

The processor 5 1 is further configured to calculate a load of the source data center, and determine the to-be-migrated load data according to the load threshold upper limit; select at least one load from the data center by using the at least one communication interface 53 to be lower than the average application resource occupation of the data center. The rate of the data center serves as a collection of resource scheduling data centers.

The processor 5 1 is further configured to acquire, by using at least one communication interface, a network resource occupancy rate of each data center in the resource scheduling data center set, and schedule application resource occupancy rate and network resources of each data center in the data center set according to the resource scheduling. The occupancy calculation calculates the overall weight of each data center in the data center collection.

The processor 5 1 is further configured to calculate the immigration load data that each data center in the resource scheduling data center set can bear, and select the data center with the smallest integrated weight from the resource scheduling data center set through the at least one communication interface 53. The data load that can be tolerated by the data center with the smallest weight is greater than the load data to be migrated from the source data center, and the data center with the smallest integrated weight is used as the destination data center.

The processor 5 1 is further configured to use the data center with the smallest integrated weight as the destination data center if the data load that the data center with the smallest integrated weight can bear is less than the load data to be moved out of the source data center.

The processor 5 1 is further configured to select, by using at least one communication interface 53, a data center with a small integrated weight in the resource scheduling data center set, and use the data center with the smaller weight as the other destination data center until the resource scheduling The data load that the data center selected in the data center collection can withstand is greater than or equal to the remaining load data to be migrated in the source data center.

Optionally, the processor 51 is further configured to: if the average application resource occupancy of the data center is lower than the lower limit of the load threshold, select, by using at least one communication interface 53, a data center with the least load of application resources in the data center. , minimizing the load of application resources The data center acts as a source data center for migrating load data.

The processor 5 1 is further configured to calculate a load of the source data center, and use the load of the source data center as the data to be migrated out; select at least one load from the data center by using at least one communication interface 53 to be higher than the average application of the data center. The data center of resource occupancy is used as a collection of resource scheduling data centers.

The processor 5 1 is further configured to acquire, by using the at least one communication interface 53, the network resource occupancy rate of each data center in the resource scheduling data center set, and schedule application resource occupancy rate and network of each data center in the data center set according to the resource scheduling. Resource occupancy calculation The overall weight of each data center in the resource scheduling data center collection.

The processor 5 1 is further configured to calculate the immigration load data that each data center in the resource scheduling data center set can bear, and select the data center with the largest integrated weight from the resource scheduling data center set through at least one communication interface, if the comprehensive weight The data load that the largest data center can withstand is greater than the data to be moved out of the source data center, and the data center with the largest integrated weight is used as the destination data center.

The processor 5 1 is further configured to use the data center with the largest integrated weight as the destination data center if the data load that the data center with the largest integrated weight can bear is less than the data to be moved out of the source data center.

The processor 5 1 is further configured to select, by using the at least one communication interface 53, the data center with the second largest integrated weight in the resource scheduling data center set, and use the data center with the second largest weight as the other destination data center until the resource scheduling data The data load that the selected data center can bear in the central set is greater than or equal to the remaining load data to be moved out of the source data center.

The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto, and any skilled person skilled in the art may disclose the technology disclosed in the present invention. Variations or substitutions are readily conceivable within the scope of the invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims

claims

1. A resource scheduling method between data centers, characterized in that it includes: obtaining the application resource occupancy rate of each data center, and calculating the average application resource occupancy rate of the data center according to the application resource occupancy rate;

Determine whether the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold or is lower than the lower limit of the load threshold;

Select the source data center for migrating the load data from the data center according to the judgment result;

Select the destination data center for migrating the load data in the data center according to the load data to be migrated from the source data center and the application resource occupancy rate and network resource occupancy rate of each data center;

Migrate load data from the source data center and move the load data into the destination data center.

2. The method according to claim 1, characterized in that, selecting the source data center for migrating load data in the data center according to the judgment result includes:

If the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold, then the data center with the largest load of application resources is selected from the data center, and the data center with the largest load of the application resources is used for migration. The source data center from which the load data is exported.

3. The method according to claim 1, characterized in that, the load data to be migrated according to the source data center and the application resource occupancy rate and network resource occupancy rate of each data center are located at the location of the data center. Select the destination data center for migrating load data from the above data centers, including:

Calculate the load of the source data center, and determine the load data to be migrated based on the upper limit of the load threshold;

Select at least one data center whose load is lower than the average application resource occupancy rate of the data center from the data center as a resource scheduling data center set;

Obtain the network resource occupancy rate of each data center in the resource scheduling data center set, and calculate the application resource occupancy rate and the network resource occupancy rate of each data center in the resource scheduling data center set. Resource Scheduling Data Center Collection The comprehensive weight of each data center;

Calculate the in-migration load data that each data center in the resource scheduling data center set can withstand, and select the data center with the smallest comprehensive weight from the resource scheduling data center set. If the data center with the smallest comprehensive weight can If the load data to be moved in is greater than the load data to be moved out of the source data center, the data center with the smallest comprehensive weight will be used as the destination data center.

4. The method according to claim 3, characterized in that, the method further includes: if the migrating load data that the data center with the smallest comprehensive weight can withstand is less than the to-be-migrated load data of the source data center, Then the data center with the smallest comprehensive weight will be used as the destination data center;

The data center with the second smallest comprehensive weight is selected from the resource scheduling data center set, and the data center with the second smallest comprehensive weight is used as another destination data center until the data center selected from the resource scheduling data center set is The load data that the data center can bear to move in is greater than or equal to the load data that remains to be moved out of the source data center.

5. The method according to claim 1, wherein the selecting a source data center for migrating load data in the data center according to the judgment result includes:

If the average application resource occupancy rate of the data center is lower than the lower limit of the load threshold, select the data center with the smallest load of application resources among the data centers, and use the data center with the smallest load of the application resources as the data center. Migrate the source data center of the load data.

6. The method according to claim 1, characterized in that: the load data to be migrated according to the source data center and the application resource occupancy rate and network resource occupancy rate of each data center are located at the location of the data center. The destination data centers selected for migrating load data from the above data centers include:

Calculate the load of the source data center and use the load of the source data center as the load data to be migrated;

Select at least one data center from the data center whose load is higher than the average application resource occupancy rate of the data center as a set of resource scheduling data centers;

Obtain the network resource occupancy rate of each data center in the resource scheduling data center set, and schedule the application according to the application of each data center in the resource scheduling data center set. The resource occupancy rate and the network resource occupancy rate calculate the comprehensive weight of each data center in the resource scheduling data center set;

Calculate the in-migration load data that each data center in the resource scheduling data center set can withstand, and select the data center with the largest comprehensive weight from the resource scheduling data center set. If the data center with the largest comprehensive weight can If the in-migration load data is greater than the out-migration load data of the source data center, the data center with the highest comprehensive weight will be used as the destination data center.

7. The method according to claim 6, characterized in that, the method further includes: if the relocation load data that the data center with the largest comprehensive weight can withstand is less than the to-be-relocation load data of the source data center, Then the data center with the largest comprehensive weight will be used as the destination data center;

Select the data center with the second largest comprehensive weight in the resource scheduling data center set, and use the data center with the second largest comprehensive weight as another destination data center until the data selected in the resource scheduling data center set The in-migration load data that the center can bear is greater than or equal to the remaining to-be-migration load data in the source data.

8. A resource scheduling device, characterized by including:

A data processing unit, configured to obtain the application resource occupancy rate of each data center, and calculate the average application resource occupancy rate of the data center based on the application resource occupancy rate;

A management unit, used to determine whether the average application resource occupancy rate of the data center forwarded by the data processing unit exceeds the upper limit of the load threshold or is lower than the lower limit of the load threshold; a data center acquisition unit, used to determine based on the judgment forwarded by the management unit The result is to select the source data center used to migrate the load data;

The data center acquisition unit is also configured to select the application resource in the data center based on the load data to be moved out of the source data center and the application resource occupancy rate and network resource occupancy rate of each data center. Move the load data to the destination data center;

The data processing unit is also configured to receive the source data center and the destination data center selected by the data center acquisition unit and migrate the load data from the source data center, and migrate the load data into the destination data center.

9. The device according to claim 8, characterized in that the data center obtains Retrieval units include:

The first acquisition subunit is used to, if the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold, select the data center with the largest load of application resources among the data centers, and calculate the load of the application resources. The largest data center serves as the source data center for migrating workload data.

10. The device according to claim 8, characterized in that the data center acquisition unit further includes:

The first data processing subunit is used to calculate the load of the source data center and determine the load data to be migrated based on the upper limit of the load threshold;

The first acquisition subunit is also used to select at least one data center from the data center whose load is lower than the average application resource occupancy rate of the data center as a resource scheduling data center set;

The first data processing sub-unit is also used to obtain the network resource occupancy rate of each data center in the resource scheduling data center set, and based on the application resources of each data center in the resource scheduling data center set The occupancy rate and the network resource occupancy rate calculate the comprehensive weight of each data in the resource scheduling data center set;

The first acquisition subunit is also used to calculate the in-migration load data that each data center in the resource scheduling data center set can withstand, and select the data center with the smallest comprehensive weight from the resource scheduling data center set. , if the migrating load data that the data center with the smallest comprehensive weight can withstand is greater than the load data to be moved out of the source data center, then the data center with the smallest comprehensive weight will be used as the destination data center.

1 1. The device according to claim 10, characterized in that it includes: the first acquisition subunit, which is also configured to: if the data center with the smallest comprehensive weight can withstand the in-migration load data is less than the source data For the load data to be migrated from the center, the data center with the smallest comprehensive weight will be used as the destination data center;

The first acquisition subunit is also used to select the data center with the second smallest comprehensive weight from the resource scheduling data center set, and use the data center with the second smallest comprehensive weight as another destination data center, until The in-migration load data that the data center selected from the resource scheduling data center set can withstand is greater than or equal to the remaining load data of the source data center. The load data is to be migrated out.

12. The device according to claim 8, characterized in that the data center acquisition unit further includes:

The second acquisition subunit is used to select the data center with the smallest load of application resources among the data centers if the average application resource occupancy rate of the data center is lower than the lower limit of the load threshold, and obtain the data center with the smallest load of the application resources. The data center with the least load serves as the source data center for migrating the load data.

13. The device according to claim 8, characterized in that the data center acquisition unit further includes:

The second data processing subunit is used to calculate the load of the source data center, and use the load of the source data center as the load data to be migrated;

The second acquisition subunit is also used to select at least one data center from the data center whose load is higher than the average application resource occupancy rate of the data center as a resource scheduling data center set;

The second data processing sub-unit is also used to obtain the network resource occupancy rate of each data center in the resource scheduling data center set, and based on the application resources of each data center in the resource scheduling data center set The occupancy rate and the network resource occupancy rate calculate the comprehensive weight of each data in the resource scheduling data center set;

The second acquisition subunit is also used to calculate the in-migration load data that each data center in the resource scheduling data center set can withstand, and select the data center with the largest comprehensive weight from the resource scheduling data center set. , if the data center with the largest comprehensive weight can withstand the in-migration load data is greater than the load data to be moved out of the source data center, then the data center with the largest comprehensive weight will be used as the destination data center.

14. The device according to claim 13, characterized in that,

The second acquisition subunit is also used to obtain the data center with the highest comprehensive weight if the in-migration load data that the data center with the largest comprehensive weight can bear is less than the to-be-migrated load data of the source data center. As a destination data center;

The second acquisition subunit is also configured to select the data center with the second largest comprehensive weight in the resource scheduling data center set, and use the data center with the second largest comprehensive weight as the data center. For another destination data center, until the migrating load data that the data center selected in the resource scheduling data center set can bear is greater than or equal to the remaining to-be-migrated load data of the source data center.

15. A resource scheduling device, characterized in that it includes: at least one processor, memory, communication interface and bus, the at least one processor, memory and communication interface are connected through the bus and communicate with each other, wherein:

The processor is configured to obtain the application resource occupancy rate of each data center through the at least one communication interface, and calculate the average application resource occupancy rate of the data center based on the application resource occupancy rate;

The memory is used to determine whether the average application resource occupancy rate of the data center forwarded by the processor exceeds the upper limit of the load threshold or is lower than the lower limit of the load threshold;

The processor is further configured to select a source data center for migrating load data in the data center through the at least one communication interface according to the judgment result of the memory forwarding;

The processor is also configured to process the data in the data through the at least one communication interface according to the load data to be migrated from the source data center and the application resource occupancy rate and network resource occupancy rate of each data center. Select the destination data center for migrating load data from the center;

The processor is further configured to receive the source data center and the destination data center selected by the processor through the at least one communication interface and migrate the load data from the source data center, and transfer the load data Move into the destination data center.

16. The device according to claim 15, characterized in that: the processor is further configured to: if the average application resource occupancy rate of the data center exceeds the upper limit of the load threshold, through the at least one The communication interface selects the data center with the largest load of application resources among the data centers, and uses the data center with the largest load of application resources as the source data center for migrating the load data.

17. The device according to claim 15, comprising: the processor, further configured to calculate the load of the source data center, and determine the load data to be migrated based on the upper limit of the load threshold; via said at least one communication interface from Select at least one data center in the data center whose load is lower than the average application resource occupancy rate of the data center as a set of resource scheduling data centers;

The processor is also configured to obtain the network resource occupancy rate of each data center in the resource scheduling data center set through the at least one communication interface, and calculate the network resource occupancy rate of each data center in the resource scheduling data center set according to the at least one communication interface. The application resource occupancy rate and the network resource occupancy rate calculate the comprehensive weight of each data center in the resource scheduling data center set;

The processor is also configured to calculate the in-migration load data that each data center in the resource scheduling data center set can withstand, and select the comprehensive weight from the resource scheduling data center set through the at least one communication interface. The smallest data center, if the data center with the smallest comprehensive weight can withstand the in-migration load data is greater than the to-be-migrated load data of the source data center, then the data center with the smallest comprehensive weight will be used as the destination data center.

18. The device according to claim 17, characterized in that: the processor is further configured to: If the load data is output, the data center with the smallest comprehensive weight will be used as the destination data center;

The processor is further configured to select the data center with the second smallest comprehensive weight from the resource scheduling data center set through the at least one communication interface, and use the data center with the second smallest comprehensive weight as another purpose. data center until the relocation load data that the data center selected in the resource scheduling data center set can withstand is greater than or equal to the remaining to-be-relocation load data of the source data center.

19. The device according to claim 15, comprising: the processor, further configured to: if the average application resource occupancy rate of the data center is lower than the lower load threshold, through the at least one The communication interface selects the data center with the smallest load of application resources among the data centers, and uses the data center with the smallest load of the application resources as the source data center for migrating the load data.

20. The device according to claim 15, characterized by comprising: the processor, further configured to calculate the load of the source data center and convert the source data into The load of the data center is used as load data to be migrated; selecting at least one data center from the data center with a load higher than the average application resource occupancy rate of the data center through the at least one communication interface as a set of resource scheduling data centers;

The processor is also configured to calculate the in-migration load data that each data center in the resource scheduling data center set can withstand, and select the comprehensive weight from the resource scheduling data center set through the at least one communication interface. The largest data center. If the data center with the largest comprehensive weight can withstand the in-migration load data is greater than the to-be-migrated load data in the source data, then the data center with the largest comprehensive weight will be used as the destination data center. .

21. The device according to claim 20, characterized in that: the processor is further configured to: if the data center with the largest comprehensive weight can withstand the migrating load data is less than the to-be-migrated data center of the source data center, If the load data is output, the data center with the largest comprehensive weight will be used as the destination data center;

The processor is further configured to select the data center with the second largest comprehensive weight from the resource scheduling data center set through the at least one communication interface, and use the data center with the second largest comprehensive weight as another destination data. center until the relocation load data that the data center selected in the resource scheduling data center set can withstand is greater than or equal to the remaining load data to be migrated out of the source data center.