WO2016134542A1

WO2016134542A1 - Virtual machine migration method, apparatus and device

Info

Publication number: WO2016134542A1
Application number: PCT/CN2015/073422
Authority: WO
Inventors: 徐杰; 周伟
Original assignee: 华为技术有限公司
Priority date: 2015-02-28
Filing date: 2015-02-28
Publication date: 2016-09-01
Also published as: CN106133693B; CN106133693A

Abstract

A virtual machine migration method, apparatus and device, comprising: acquiring the usage rate of a plurality of types of resources of various virtual machines in a source host, and according to the usage rate of the plurality of types of resources of the virtual machines, determining the usage rate of the plurality of types of resources of the source host (S210);according to the usage rate of the plurality of types of resources of the source host, dividing the plurality of types of resources of the source host into an overloaded resource and a non-overloaded resource, and determining a first migration weight of the overloaded resource and a second migration weight of the non-overloaded resource (S220);according to the usage rate of the plurality of types of resources of the virtual machines, the first migration weight of the overloaded resource and the second migration weight of the non-overloaded resource, selecting a virtual machine to be migrated from the virtual machines (S230); and sending a migration instruction to a monitor of the source host, so that the monitor migrates the virtual machine to be migrated to a target host (S240).Therefore, the accuracy of selecting a virtual machine to be migrated can be improved, such that the problem of the overload of a source host can be solved quickly, thereby achieving the purpose of saving computer resources.

Description

Virtual machine migration method, device and device

Technical field

The present invention relates to the field of data processing technologies, and in particular, to a method, an apparatus, and a device for migrating a virtual machine.

Background technique

With the rapid development of virtualization technology, data centers are increasingly using virtual machines to provide services. In order to load balance the data center or effectively utilize server resources, it is often necessary to migrate virtual machines. Migrating a virtual machine is the process of moving a virtual machine from one host or storage location to another. Live migration is the move of a powered-on virtual machine to a new host. With live migration, you can move a virtual machine to a new host without disrupting virtual machine availability.

In the prior art, the resource manager detects the resource utilization of each host, such as monitoring the usage rate of resources such as CPU, memory, and network bandwidth, and assigns the same weight to each of the above resources or pre-defined a fixed weight. When the host is overloaded, calculate the volume of each virtual machine running on the host (the reciprocal of the product of the various resource idle rate), and sort all the volumes in descending order, and finally select the virtual machine with the largest volume to migrate. If the above host is still overloaded at this time, the next-largest virtual machine will continue to be selected for migration until the host is no longer overloaded. For example, on a CPU overloaded host, U ₁ (CPU usage rate), M ₁ (memory usage rate), and B ₁ (bandwidth usage rate) of virtual machine VM ₁ are 0.8, 0.2, and 0.5, respectively, according to the above. VM load calculation method (three kinds of resource is free reciprocal of a sum of products), a virtual machine (VM) can be calculated the volume of ₁ L ₁ is 12.5; the virtual machine VM U _{_{_2,}} M _{2 2} a, B ₂ respectively, 0.2,0.8, 0.5, and the virtual machine (VM) can be calculated volume of ₂ L ₂ of 12.5. Therefore, if the virtual machine is selected according to the volume order of the virtual machine as described above, VM ₁ and VM ₂ have the same chance of selection when migrating, but when VM ₂ is selected for migration, the CPU usage of VM ₂ is only 0.2. Therefore, it does not solve the problem of host CPU overload, so VM ₁ needs to be migrated. In addition, since VM _{2 has} a memory usage of 0.8, this will seriously affect the migration time of the VM.

It can be seen from the above that in the prior art, different types of resources of a virtual machine are assigned the same weight or a fixed weight is predefined, which affects the correctness of the migration virtual machine selection, thereby increasing the number of virtual machine migrations and virtual machine migration. Overhead.

Summary of the invention

The embodiments of the present invention provide a method, an apparatus, and a device for migrating a virtual machine, so as to solve the problem of assigning the same weight to the different types of resources of the virtual machine or pre-defining the fixed weights in the prior art, thereby affecting the selection of the virtual machine to be migrated. Correctness, resulting in an increase in the number of migrations and a large migration overhead.

In a first aspect, a method for migrating a virtual machine is provided, the method comprising:

The resource manager obtains usage rates of multiple types of resources of each virtual machine in the source host, and determines usage rates of multiple types of resources of the source host according to usage rates of multiple types of resources of the virtual machines;

Dividing, according to the usage rate of the multiple types of resources of the source host, the multiple types of resources of the source host into an overload resource and an un-overload resource, and determining a first migration weight of the overload resource and the un-overloaded resource. Second migration weight;

Selecting a virtual machine to be migrated from the virtual machine according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the unoverloaded resource;

Sending a migration instruction to the monitor of the source host, so that the monitor migrates the virtual machine to be migrated to the target host according to the migration instruction.

With reference to the first aspect, in a first implementation manner of the first aspect, each of the virtual machines has the same type of resources, and the determining, according to the usage rates of the multiple types of resources of the virtual machines, determining The usage rates of the multiple types of resources of the source host include:

Determining the usage rate of the type resource of the source host according to the usage rate of the same type of resources of the respective virtual machines and the size of the type resources allocated to the respective virtual machines in advance; until the source is determined The usage rate of each type of resource for the host.

With reference to the first aspect, or the first implementation manner of the first aspect, in a second implementation manner of the first aspect, the source host is configured according to a usage rate of multiple types of resources of the source host The types of resources are classified into overloaded resources and unoverloaded resources including:

And determining, by the first type of resources, the first type of resources, when the usage rate of the first type of resources is greater than a preset threshold, determining the first type of resources as an overload resource;

When the usage rate of the first type of resource is not greater than a preset threshold, the first type of resource is determined as an unoverloaded resource.

With reference to the first aspect, or the first implementation manner of the first aspect, or the second implementation manner of the first aspect, in a third implementation manner of the first aspect, the determining, The second migration weight of the un-allocated resource includes:

Determining, by the overload resource, a first migration weight of the overload resource according to the usage rate of the overload resource and a preset first algorithm;

And determining, by the un-overloaded resource, a second migration weight of the un-overloaded resource according to the usage rate of the un-overloaded resource and a preset second algorithm.

With reference to the first aspect, or any one of the foregoing three implementation manners of the first aspect, in a fourth implementation manner of the first aspect, the usage rate of the multiple types of resources according to the each virtual machine, The first migration weight of the overload resource and the second migration weight of the un-overloaded resource, the virtual machine to be migrated from the virtual machine includes:

Acquiring, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource, acquiring a first migration value of each virtual machine;

Comparing the first migration value of each virtual machine, and the smallest first migration value The corresponding virtual machine is determined to be the virtual machine to be migrated.

With reference to the first aspect, or any one of the first three implementation manners of the first aspect, in a fifth implementation manner of the first aspect, the usage rate of the multiple types of resources according to the The first migration weight of the overload resource and the second migration weight of the un-overloaded resource, the virtual machine to be migrated from the virtual machine includes:

Determining, according to a first migration value of each of the virtual machines and a size of memory allocated to the respective virtual machines in advance, a second migration value of each of the virtual machines;

The second migration value of each virtual machine is compared, and the virtual machine corresponding to the smallest second migration value is determined as the virtual machine to be migrated.

With reference to the fourth implementation manner of the first aspect, or the fifth implementation manner of the first aspect, in the sixth implementation manner of the first aspect, before the acquiring the first migration generation value of the each virtual machine, The method further includes:

Comparing the usage rate of the type of overload resources of each virtual machine to each type of overload resource in the overload resource of the source host, and using the maximum overload resource usage rate as the target of the type overload resource Usage rate

Comparing the usage rates of the types of non-overloaded resources of the virtual machines to each of the un-overloaded resources of the source host, and using the minimum overload resource usage rate as the type is not overloaded. The target usage rate of the resource.

With reference to the sixth implementation manner of the first aspect, in a seventh implementation manner of the first aspect, the usage rate of the multiple types of resources of the each virtual machine, the first migration weight of the overload resource, and The second migration weight of the un-overloaded resource, and obtaining the first migration value of each virtual machine includes:

Calculating the first migration value of each virtual machine according to the following formula:

Wherein, D _i is the first migration value of the i-th virtual machine, n is the number of overload resource types of the source host, and A _t is the first migration weight of the t-type overload resource of the source host, A _t >1, S _it is the usage rate of the t-th type resource of the i-th virtual machine, P _t is the target usage rate of the t-type overload resource of the source host, and m is the source host The number of overload resource types, B _l is the second migration weight of the first type of un-overloaded resources of the source host, 0<B _l <1, and T _il is the use of the first type of resources of the i-th virtual machine. Rate, Q _l is the target usage rate of the first type of un-overloaded resources of the source host, and n+m is the number of resource types of the source host.

A second aspect provides a virtual machine migration apparatus, where the apparatus includes: an obtaining unit, a dividing unit, a selecting unit, and a sending unit;

The obtaining unit is configured to obtain usage rates of multiple types of resources of each virtual machine in the source host, and determine usage of multiple types of resources of the source host according to usage rates of multiple types of resources of the virtual machines. rate;

And the dividing unit is configured to divide, according to the usage rate of the multiple types of resources of the source host determined by the acquiring unit, the multiple types of resources of the source host into an overload resource and an un-overload resource, and determine the a first migration weight of the overloaded resource and a second migration weight of the unoverloaded resource;

The selecting unit is configured to select, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overload resource, and the second migration weight of the un-overloaded resource, from the virtual machine The virtual machine to be migrated;

The sending unit is configured to send a migration instruction to the monitor of the source host, so that the monitor migrates the to-be-migrated virtual machine selected by the selecting unit to the target host according to the migration instruction.

With reference to the second aspect, in a first implementation manner of the second aspect, each of the virtual machines has the same type of resources, and the acquiring unit is specifically configured to:

According to the usage rate of the same type of resources of the respective virtual machines, and in advance The size of the type resource allocated by the virtual machine determines the usage rate of the type resource of the source host; until the usage rate of each type resource of the source host is determined.

With reference to the second aspect or the first implementation manner of the second aspect, in the second implementation manner of the second aspect, the dividing unit is specifically configured to:

With reference to the second aspect, or the first implementation manner of the second aspect, or the second implementation manner of the second aspect, in the third implementation manner of the second aspect, the dividing unit is specifically configured to:

With reference to the second aspect, or any one of the foregoing three implementation manners of the second aspect, in the fourth implementation manner of the second aspect, the selecting unit is specifically configured to:

The first migration value of each virtual machine is compared, and the virtual machine corresponding to the smallest first migration value is determined as the virtual machine to be migrated.

With reference to the second aspect, or any one of the first three implementation manners of the second aspect, in the fifth implementation manner of the second aspect, the selecting unit is specifically configured to:

With reference to the fourth implementation manner of the second aspect, or the fifth implementation manner of the second aspect, in a sixth implementation manner of the second aspect, the device further includes: a comparison unit, configured to use the source host Each type of overload resource in the overload resource compares the usage rate of the type of overload resource of each virtual machine, and uses the maximum overload resource usage rate as the target usage rate of the type of overload resource;

With reference to the sixth implementation manner of the second aspect, in the seventh implementation manner of the second aspect, the selecting unit is specifically configured to:

In a third aspect, a migration device for a virtual machine is provided, where the migration device includes:

a processor, a memory, a communication interface, and a bus, wherein the processor, the memory, and The communication interface communicates via the bus;

The communication interface is configured to communicate with a switch or a control server;

The memory is used to store a program;

When the migration device is running, the processor is configured to execute the program stored by the memory to perform the migration method of the virtual machine described in the first aspect above.

The method, device, and device for migrating a virtual machine provided by the embodiment of the present invention acquires usage rates of multiple types of resources of each virtual machine in the source host, and determines various types of source hosts according to usage rates of multiple types of resources of the virtual machine. Usage of type resources; according to the usage rate of multiple types of resources of the source host, the multiple types of resources of the source host are divided into overloaded resources and un-overloaded resources, and the first migration weight of the overloaded resources and the un-overloaded resources are determined. The migration weight is selected according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the unoverloaded resource, and the virtual machine to be migrated is selected from the virtual machine; Migrate instructions so that the monitor migrates the virtual machine to be migrated to the target host. Therefore, the correctness of the selection of the virtual machine to be migrated can be improved, so that the problem of overloading the source host can be quickly solved, and the purpose of saving computer resources is achieved.

DRAWINGS

1 is an application scenario diagram of a method for migrating a virtual machine according to the present invention;

2 is a flowchart of a method for migrating a virtual machine according to Embodiment 1 of the present invention;

3 is a schematic diagram of a method for determining a weight of a resource provided by the present invention;

4 is a flowchart of a method for migrating a virtual machine according to Embodiment 2 of the present invention;

FIG. 5 is a schematic diagram of a virtual machine migration apparatus according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a migration device of a virtual machine according to an embodiment of a device according to the present invention.

detailed description

The technical solution of the present invention will be further described in detail below through the accompanying drawings and embodiments.

In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the following will be combined with the present invention. The technical solutions in the embodiments of the present invention are clearly and completely described in the drawings, and the embodiments are described as a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

In order to facilitate the understanding of the embodiments of the present invention, the embodiments of the present invention are not to be construed as limiting.

1 is an application scenario diagram of a migration method of a virtual machine provided by the present invention. In FIG. 1, VM ₁ , VM ₂ ... VM _m respectively represent a virtual machine, and the virtual machine refers to a running with software function and having a complete hardware system function. A complete computer system in a completely isolated environment. Users can simulate another or multiple virtual machines on a single host, and these virtual machines work exactly like real computers. For the user, it's just an application running on a physical machine, but for an application running in a virtual machine, it's a real computer. Virtual machine resources involve multiple aspects: Central Processing Unit (CPU), Memory (MEM), Network Bandwidth (BW), and disk. The relationship between these resources should be considered when planning a virtual machine. Otherwise, the unreasonable allocation of resources will result in poor application performance in the virtual machine. Physical Machine (PM) ₁ , PM ₂ ... PM _n respectively represent n hosts, wherein each host has a virtual machine monitor (VMM), which is also called a hypervisor. An intermediate software layer running between the underlying physical machine and the operating system that allows multiple operating systems and applications to share hardware that can access all physical devices on the physical machine, including disks and memory. VMM not only coordinates access to these hardware resources, but also protects between virtual machines. When a physical machine boots and executes VMM, it loads the operating system of all virtual machine clients and allocates the appropriate amount of memory, CPU, network, and disk to each virtual machine. VMM is at the heart of all virtualization technologies, and the ability to support multi-workload migration non-disruptively is a fundamental feature of VMM.

In Figure 1, when multiple hosts are connected to the same shared storage, real-time migration of virtual machines can be supported, that is, the migration of virtual machines from one host to another does not affect the migration process. Migrate the normal communication of the virtual machine and the applications on it.

Virtual machine migration is an extremely powerful tool for data center and cluster administrators to improve system performance and increase resource allocation flexibility, and is used in many important areas:

1) Load balancing: periodically collect the resource usage rate, redistribute the virtual machine to the physical machine in the data center or cluster, and migrate some virtual machines on the overloaded physical machine to the light load physical machine to evenly distribute the workload to avoid Load imbalances are caused by frequent deployment and termination of virtual machines.

2) Online maintenance: Sometimes the physical machine needs to be maintained or upgraded. The migration can be used to migrate the virtual machine to other physical machines during maintenance without shutting down the virtual machine and improving the maintainability of the system.

3) Active Fault Tolerance: Monitor the health status of the node, such as temperature or disk error log. If a physical machine is about to fail, the virtual machine on it actively migrates to a more secure location, which reduces the probability of the user perceiving the failure and improves the reliability of the system.

4) Resource optimization and power management: adjust the resource allocation of the data center or cluster, optimize the use of resources, and integrate virtual machines from lightly loaded hosts to heavily loaded hosts. Once the migration is complete, the hosts that previously ran the migrated virtual machines can be shut down, helping to reduce costs and save energy.

In short, the migration of virtual machines improves the ability to manage and provide services in data centers and clusters, realizes flexible allocation of resources, improves system reliability, and enables more powerful computing power, larger memory, and faster communication. Ability to save energy.

The resource management algorithm in Figure 1 generally includes three phases: a pre-allocation phase, a migration planning phase, and a migration execution phase. The specific functions of each phase are described as follows:

1) Pre-allocation phase: The resource manager collects resource usage data from the running host through the VMM within a specific measurement interval. The details of the sampled data are highly dependent on the virtualization technology used and the type of data that needs to be collected. With this data, the resource manager has a general understanding of the performance level of the running host. For the previously set trigger condition, if the monitoring information shows that a conflict occurs at this time, the resource reallocation is triggered.

It should be noted that the foregoing triggering conditions may include two situations: the first case is that the virtual machine in the host is overloaded due to the competition for resources; the second case is that the host needs to be online maintained or powered. management. In the second case, the host's resources are usually not overloaded, and all virtual machines on the host need to be migrated out.

2) Migration planning phase: This phase is the most critical part of the entire process. The resource manager's responsibility is to generate appropriate migration scenarios or instructions for new virtual machine deployments to eliminate or minimize the behavior of illegal trigger conditions. The migration plan usually includes three important components: the source host, the virtual machine to be migrated, and the target host.

3) Migration execution phase: According to the migration plan and instructions generated by the resource manager, the VMM implements the specific migration work. The specific details of the phase are mainly determined by the virtualization technology adopted.

Method embodiment

FIG. 2 is a flowchart of a method for migrating a virtual machine according to Embodiment 1 of the present invention. The method is applicable to the virtual machine to be migrated involved in the migration planning phase when the trigger condition in FIG. 1 is the first case. The problem, as shown in FIG. 2, may specifically include:

S210: The resource manager obtains usage rates of multiple types of resources of each virtual machine in the source host, and determines usage rates of multiple types of resources of the source host according to usage rates of multiple types of resources of the virtual machines.

Further, each of the virtual machines has the same type of resources, and the step S210 may specifically include:

Here, the multiple types of resources of the source host may include, but are not limited to, CPU, memory, network bandwidth, and disk. It can be understood that multiple types of resources of the virtual machine running on the source host may include but are not limited to: CPU. , memory, network bandwidth, and disk. In this specification, each virtual machine includes three types of resources: CPU, memory, and network bandwidth as an example.

For example, suppose that two virtual machines are running on the source host: VM ₁ and VM ₂ , and the CPU size of the source host is assumed to be 10G, wherein the size of the CPU allocated to VM ₁ is 6G in advance, and VM _{2 is} allocated in advance. If the size of the CPU is 4G, if the resource manager obtains the CPU usage of the virtual machine VM ₁ in the host at 0.6, and the CPU usage of the virtual machine VM ₂ is 0.4, then the source host The usage rate of the CPU is 0.52 ((6×0.6+4×0.4)/10); similarly, the usage rate of other types of resources of the source host can be obtained.

It should be noted that the usage rate of the foregoing resources (CPU, memory, network bandwidth, and disk, etc.) refers to the consumption of resources by related applications carried on the virtual machine. Therefore, the resource usage rate of the virtual machine is a dynamic value. , which dynamically changes with the performance of the application; accordingly, another concept needs to be described in the present invention: the size of the resources allocated in advance for each virtual machine, which is determined when the virtual machine is created, It can be regarded as a static value, and generally does not change later. For example, the CPU size of the above source host is 10G, and the size of the CPU allocated to VM _{1 in} advance is 6G.

S220. The multiple types of resources of the source host are divided into overload resources and non-overload resources according to the usage rate of the multiple types of resources of the source host, and the first migration weight of the overload resource and the The second migration weight of the overloaded resource.

The dividing the plurality of types of resources of the source host into overload resources and un-overloaded resources according to the usage rate of the multiple types of resources of the source host includes:

For example, assume that the usage rates of the three types of resources of the source host (in order: CPU, memory, and network bandwidth) are 0.9, 0.4, and 0.3, respectively, according to the method in the foregoing example, and the foregoing three types are preset. The thresholds of type resources are 0.85, 0.8, and 0.7, respectively, because 0.9>0.85, so the CPU of the source host is overloaded, and since 0.4<0.8, 0.3<0.7, the memory and network of the source host. The bandwidth is not overloaded, that is, the number of overload resource types of the source host is 1, and the number of un-overloaded resource types of the source host is 2.

It should be noted that, in the above example, the three different threshold values are set in advance for three different types of resources, and those skilled in the art may also set one or two thresholds in advance. The invention is not limited thereto.

Considering the different types of resources on the same host, the degree of resource strain may be different. When selecting a virtual machine to be migrated, this factor should be taken into account, that is, the method of assigning migration weight to each resource according to the host resource demand degree. For overloaded hosts, it is often desirable to migrate several virtual machines on them to quickly reduce their load.

As in the previous example, when the source host CPU is overloaded, then it is desirable to migrate the virtual machine with the highest CPU to release the load of the host as soon as possible; for the unoverloaded resources, there is no need to reduce the usage of the resources on the host. In order to make full use of resources, when migrating, the virtual machine with the lowest usage rate of resources corresponding to the un-overloaded resource is selected. Considering these two situations together, the migration weights of the overloaded resources and the unoverloaded resources are determined according to different algorithms.

Specifically, determining, by the overload resource, a first migration weight of the overload resource according to a usage rate of the overload resource and a preset first algorithm;

The present invention does not limit the foregoing preset first algorithm and the preset second algorithm, but requires the first migration weight and the un-overloaded resource of the loaded resource determined according to the preset first algorithm and the preset second algorithm. The second migration weight has the following four characteristics:

1) The first migration weight of the overloaded resource is always greater than 1, and the second migration weight of the unloaded resource is always less than 1, that is, the first migration weight of the overloaded resource is always greater than the second migration weight of the unoverloaded resource.

2) The first migration weight of the overloaded resource has a nonlinear relationship with respect to the usage rate of the overloaded resource, and the second migration weight of the unoverloaded resource has a linear relationship with respect to the usage rate of the unoverloaded resource.

3) The higher the usage rate of the overloaded resource, the larger the corresponding first migration weight, and the less overloaded resources. The higher the usage rate, the smaller the corresponding second migration weight.

4) The usage rate of multiple types of resources of the host dynamically changes with time, and the migration weight of the resources determined according to the above algorithm also dynamically changes.

To resources present invention provides a method of weight determination schematic FIG. 3, FIG. 3, 301, 302 and 303 respectively at time t _1, PM's CPU, MEM and BW usage, 307, 308 and 309 respectively at _time t, the right to migrate CPU, MEM and BW weight. Suppose at time T _1, CPU utilization rate greater than a predetermined threshold value, i.e. CPU resources belonging to the overload, and the MEM and BW usage threshold value not greater than a predetermined set, i.e. MEM and BW belongs to the un-overloaded resource. As can be seen from Figure 3, for overloaded resources, the higher resource usage rate causes it to be assigned a higher migration weight, while for the un-overloaded resource, the higher resource usage rate. As a result, the migration weight is lower. For example, the usage rate of BW is greater than the usage rate of MEM, and the migration weight of BW is smaller than the migration weight of MEM. Therefore, when the virtual machine to be migrated is selected, since the usage rate of the overloaded resource is high, it has a higher priority for resource migration to balance resource utilization. Moreover, since the migration weight of the allocated resource is always greater than the migration weight of the unoverloaded resource allocation, the overloaded resource always has a higher priority to be migrated out.

In addition, in FIG. 3, 304, 305, and 306 respectively indicate the usage rates of the CPU, MEM, and BW of the PM at times t ₁ + Δt, and 310, 311, and 312 indicate the CPU, MEM at time t ₁ + Δt, respectively. And the migration weight of the BW, that is, the migration weights of the CPU, MEM, and BW are different at different times. It can be seen that the method for determining the migration weight of the present invention can dynamically update the migration weight of the resource of the host when the usage rate of the resource of the host changes with time, that is, the method for determining the migration weight of the present invention is a dynamic weight determination method. Therefore, the method for determining the migration weight of the present invention can reflect the load condition of the host in real time, thereby quickly reducing the host load and making full use of the resources of the host.

S230. Select a virtual machine to be migrated from the virtual machine according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource.

Further, the step S230 may further include:

Here, before acquiring the first migration value of each virtual machine, the multi-dimensional decision matrix may be first established according to the usage rate of multiple types of resources of each virtual machine.

For example, suppose that five virtual machines are running on the source host: VM ₁ , VM ₂ , VM ₃ , VM _{4 ,} and VM ₅ . The three types of resources for each of the five virtual machines are: CPU, memory, and network. The bandwidth of the overloaded resource is the CPU, and the unloaded resource is the memory and the network bandwidth. The first migration weight of the overload resource CPU determined according to the method of step S220 is 5, and the second overloaded resource memory is the second. The migration weight is 0.2, and the second migration weight of the unoverloaded resource network bandwidth is 0.3. It is also assumed that the usage rates of the above three types of resources of each virtual machine are as shown in Table 1:

Table 1

	CPU的使用率CPU usage	内存的使用率Memory usage	网络带宽的使用率Network bandwidth usage
	CPU的使用率CPU usage	内存的使用率Memory usage	网络带宽的使用率Network bandwidth usage	VM₁ VM ₁	0.10.1	0.20.2	0.90.9
VM₂ VM ₂	0.20.2	0.30.3	0.60.6	VM₁ VM ₁	0.10.1	0.20.2	0.90.9
VM₂ VM ₂	0.20.2	0.30.3	0.60.6	VM₃ VM ₃	0.30.3	0.50.5	0.70.7
VM₄ VM ₄	0.80.8	0.60.6	0.20.2	VM₃ VM ₃	0.30.3	0.50.5	0.70.7
VM₄ VM ₄	0.80.8	0.60.6	0.20.2	VM₅ VM ₅	0.90.9	0.60.6	0.30.3

It should be noted that the usage rates of the three types of resources in the above table are all normative values, that is, the range is between 0-1. If the usage rate of the foregoing resource is not the normative value, the usage rate of the resource may be obtained. Normalize it.

After normalizing the usage rate of the three types of resources of each virtual machine, you can root According to the content of the above table, a multi-dimensional decision matrix is established, and an ideal migration virtual machine is selected according to the established multi-dimensional decision matrix. The determination process of the ideal virtual machine is as follows:

As in the foregoing example, the overload resource of the source host includes only the CPU, and the CPU usage of the virtual machines VM ₁ , VM ₂ , VM ₃ , VM _{4 ,} and VM ₅ are compared because 0.9 (corresponding to VM ₅ )>0.8>0.3>0.2>0.1, so the CPU usage of VM ₅ is the target usage rate of the CPU of the source host; and the un-overloaded resources of the source host include memory and network bandwidth, first VM ₁ , VM ₂ , VM ₃ , The memory usage of VM ₄ and VM ₅ is compared, because 0.2 (corresponding to VM ₁ ) <0.3<0.5<0.6, so the memory usage of VM ₁ is 0.2 as the target usage of the source host's memory; The network bandwidth usage ratios of VM ₁ , VM ₂ , VM ₃ , VM _{4 ,} and VM ₅ are compared because 0.2 (corresponding to VM ₄ ) < 0.3 < 0.6 < 0.7 < 0.9, so the network bandwidth usage of VM ₄ is used. 0.2 is the target usage rate of the network bandwidth of the source host; and the CPU usage rate 0.9, the memory usage rate 0.2, and the network bandwidth usage rate 0.2 are selected as ideal virtual machines.

After the ideal virtual machine is selected, the Euclidean distance between each virtual machine and the ideal virtual machine can be calculated, that is, the first migration value of each virtual machine is obtained. Specifically, the first migration value of each virtual machine is calculated according to formula 1:

(Formula 1)

Wherein, D _i is the first migration value of the i-th virtual machine, and n is the number of overload resource types of the source host. In the foregoing example, A _t is the t-type overload resource of the source host. a migration weight, A _t >1, S _it is the usage rate of the t-type overload resource of the i-th virtual machine, and P _t is the target usage rate of the t-type overload resource of the source host, where m is The number of un-overloaded resource types of the source host, B _l is the second migration weight of the first type of un-overloaded resources of the source host, 0<B _l <1, and T _il is the _lth of the i-th virtual machine. The usage rate of the type of non-overloaded resources, Q _l is the target usage rate of the first type of un-overloaded resources of the source host, and n+m is the number of resource types of the source host.

As in the foregoing example, i=5, n=1, and A ₁ =5, P ₁ =0.9, m=2, B ₁ =0.2, Q ₁ =0.2, B ₂ =0.3, Q ₂ =0.2 in Equation 1. And when calculating the first migration value D _{1 of the} virtual machine VM ₁ , S ₁₁ =0.1, T ₁₁ =0.2, T ₁₂ =0.9, thus

Similarly, the first migration values of VM ₂ , VM ₃ , VM _{4 ,} and VM ₅ can be obtained: 3.50, 3.00, 0.506, and 0.085, respectively.

Comparing the first migration value of VM ₁ , VM ₂ , VM ₃ , VM ₄ and VM ₅ , because 0.085<0.506<3.00<3.50<4.01, and 0.085 is the first migration value of virtual machine VM ₅ , Therefore, the virtual machine VM ₅ can be determined as the virtual machine to be migrated. Here, the CPU usage of the VM ₅ is 0.9. Therefore, after the virtual machine VM _{5 is} migrated to the target host, the problem of overloading the source host CPU can be quickly solved, and the number of migrations can be reduced.

S240. Send a migration instruction to the monitor of the source host, so that the monitor migrates the to-be-migrated virtual machine to the target host according to the migration instruction.

Here, the monitor of the source host is a VMM, and the foregoing migration instruction may include source location information of the virtual machine to be migrated and target location information, where the target location information is determined by the resource manager according to the placement policy, and the determining method thereof is determined. It belongs to the prior art and will not be described here.

In the first embodiment of the method provided by the present invention, first, according to a preset threshold, multiple types of resources on the source host are divided into an overload resource and an un-overload resource, and different algorithms are used for the overload resource and the non-overload resource respectively. Determine the migration weights, and the migration weight of these resources will be taken into consideration for the subsequent selection of the virtual machine to be migrated. For an overloaded resource, the corresponding first migration weight is always greater than 1, and the higher the usage rate of the resource, the larger the corresponding first migration weight; for the non-overloaded resource, the corresponding second migration weight is always less than 1, and the resource is always The higher the usage rate, the smaller the corresponding second migration weight. Then establish a multi-dimensional decision based on the usage rate of multiple types of resources of each virtual machine on the source host. Based on the established multi-dimensional decision matrix, the ideal virtual machine is selected, and then the Euclidean distance between each virtual machine and the ideal virtual machine is calculated, and the first migration weight and the overload of the overload resource determined by the dynamic weight method are not overloaded. The second migration weight of the resource is taken into account to determine the virtual machine to be migrated.

In addition, in the first embodiment of the method provided by the present invention, when the virtual machine is migrated, the overload resource has a higher priority, so it always has a higher priority for resource balancing, and the un-allocated resource is due to its The resource usage rate does not exceed the preset threshold, so in order to make full use of resources, the migration weight is lower when migrating. Therefore, the virtual machine with a large usage rate of overload resources and a small usage rate of the non-overloaded resources will have a greater chance of being selected in the method of the first embodiment, thereby rapidly reducing the host load and reducing the virtual The number of machine migrations. Compared with the prior art, only the virtual machine with the largest virtual machine volume is selected for migration. In the first embodiment, when the virtual machine is selected, the decision matrix is used to accurately allocate resources in multiple resource dimensions such as CPU, memory, and network bandwidth. Matching, can find the most suitable virtual machine for migration, so it makes full use of the resources of each dimension and improves the utilization rate of resources.

It is described here that although the first embodiment of the method provided by the present invention is an example in which the trigger condition is the first case, when the trigger condition is the second case, since the resource of the host is not overloaded, it is easy for a person skilled in the art. I think that the resources of the host are not divided, and the migration weight of each resource is determined directly according to the resource usage rate and the preset third algorithm.

The present invention also does not limit the preset third algorithm, but requires that the migration weight of each resource determined according to the preset third algorithm has the following five characteristics:

1) Each virtual machine determines the migration weight of each type of resource according to its own demand preference for various types of resources. Therefore, different virtual machines have different migration weights of various types of resources.

2) The migration weight of various types of resources is always greater than 1.

3) The migration weight of various types of resources has a nonlinear relationship with respect to the utilization rate of resources.

4) The higher the usage rate of various types of resources, the greater the corresponding migration weight.

5) The usage rate of various types of resources changes dynamically with time. According to the dynamic weight determination method, the migration weight of the corresponding resources also dynamically changes.

For the case where the virtual machine migration is triggered due to online maintenance and power management, etc., the virtual machine pair can be quantitatively determined by determining the migration weight of the resource according to the performance preference of the virtual machine itself for various types of resources. The demand situation of type resources provides effective help for later redistribution of data center resources and redeployment of the virtual machine.

In addition, since all the virtual machines on the host need to be migrated when the trigger condition is the second case, in this case, it is not necessary to consider the selection of the virtual machine to be migrated.

The migration method of the virtual machine provided in the first embodiment of the method of the present invention is applicable to the case where the network bandwidth required for migrating the virtual machine is sufficiently large, and the memory of the virtual machine is not a main consideration; when the network bandwidth required for migrating the virtual machine is relatively tight In this case, the time required to migrate one virtual machine may be longer. The long migration time means that the performance of the application running on the virtual machine may be affected. Therefore, the present invention also provides a virtual machine migration method. The method of the example takes into account the performance of the application on the virtual machine and the utilization of resources.

FIG. 4 is a flowchart of a method for migrating a virtual machine according to Embodiment 2 of the present invention. The method is applicable to the solution in the migration planning phase when the trigger condition in FIG. 1 is the first case and the migration performance is considered at the same time. The problem of the virtual machine to be migrated. During virtual machine migration, migration time and downtime can affect migration performance, and virtual machine memory size and network bandwidth are important factors affecting migration time and downtime. Therefore, on the same host, if the virtual machine migration bandwidth allocation is insufficient and all virtual machine dynamic migration uses the same network bandwidth, selecting a virtual machine that occupies less memory for migration can improve migration performance. As shown in FIG. 4, the method may specifically include:

S410. The resource manager obtains usage rates of multiple types of resources of each virtual machine in the source host, where each virtual machine has the same type of resources.

S420, determining, according to the usage rate of the same type of resources of the respective virtual machines and the size of the type resources allocated to the respective virtual machines in advance, determining the usage rate of the type resources of the source host; The usage rate of each type of resource of the source host.

S430. According to the usage rate of multiple types of resources of the source host, multiple types of the source host are used. The type resource is divided into an overload resource and an un-overload resource, and determines a first migration weight of the overload resource and a second migration weight of the un-overload resource.

S440. Acquire a first migration value of each virtual machine according to a usage rate of the multiple types of resources of the virtual machine, a first migration weight of the overloaded resource, and a second migration weight of the un-overloaded resource. .

S450. Determine a second mobility value of each virtual machine according to a first migration value of each virtual machine and a size of a memory allocated in advance for each virtual machine.

Suppose the source host memory to 10G, in advance of the source host each virtual machine _{_{_{VM 1, VM 2, VM 3}}} , VM 4 and memory size VM ₅ assigned respectively 1G, 2G, 3G, 2G, and 2G; preferably The size of the memory of each virtual machine is normalized separately, and the relative values obtained after the normalization process are 0.1, 0.2, 0.3, 0.2, and 0.2. For example, assume that the first migration values of VM ₁ , VM ₂ , VM ₃ , VM _{4 ,} and VM ₅ obtained by the method according to the first embodiment of the present invention are: 4.01, 3.50, 3.00, 0.506, and 0.085, respectively; The relative values obtained by normalizing the size of the memory allocated in advance for each of the virtual machines VM ₁ , VM ₂ , VM ₃ , VM _{4 ,} and VM _{5 in the} source host are: 0.1, 0.2, 0.3, 0.2, and 0.2, respectively. , it can be determined that the second migration value of VM ₁ , VM ₂ , VM ₃ , VM ₄ and VM ₅ are: 4.01×0.1, 3.50×0.2, 3.00×0.3, 0.506×0.2, 0.085×0.2, ie VM ₁ The second migration values of VM ₂ , VM ₃ , VM _{4 ,} and VM ₅ are: 0.401, 0.7, 0.9, 0.1012, and 0.017, respectively.

Through the above steps, you can preferentially select a virtual machine with a small pre-allocated memory for migration. The pre-allocated memory is small, which means that it takes less time to migrate the virtual machine (because the memory of the virtual machine needs to be copied repeatedly during the migration process). Pages, small memory means a small amount of data to be copied, which reduces migration overhead and reduces the impact of migration operations on the performance of related applications hosted on virtual machines.

S460. Compare the second migration value of each virtual machine, and determine the virtual machine corresponding to the smallest second migration value as the virtual machine to be migrated.

The _{_{_{VM 1, VM 2, VM 3}}} , VM 4 and the value of second-generation VM migration ₅ for comparison, since 0.017 <0.1012 <0.401 <0.7 <0.9, and 0.017 for the second virtual machine (VM) migration of value generation _5, Therefore, the virtual machine VM ₅ can be determined as the virtual machine to be migrated. As can be seen from Table 1, the virtual machine VM ₅ has the smallest network bandwidth and is pre-allocated with smaller memory, and the network bandwidth and memory are affecting the migration performance. Factor, so virtual machine VM ₅ as a virtual machine to be migrated, can improve the performance of virtual machine migration.

S470. Send a migration instruction to the monitor of the source host, so that the monitor migrates the to-be-migrated virtual machine to the target host according to the migration instruction.

In the second embodiment of the method provided by the present invention, when the virtual machine to be migrated is selected, the migration cost of each virtual machine is taken into consideration, and the migration performance can be further improved on the basis of reducing the number of virtual machine migrations and increasing the resource utilization rate. (ie, reduce migration overhead), reducing the possibility of migration causing user service level agreement (SLA) violations.

Device embodiment

FIG. 5 is a schematic diagram of a device for migrating a virtual machine according to an embodiment of the present invention. The device may be used to perform the method described in FIG. 2 or FIG. 4. In FIG. 5, the device includes: an obtaining unit 501, a dividing unit 502, The unit 503 and the transmitting unit 504 are selected.

The obtaining unit 501 is configured to obtain usage rates of multiple types of resources of each virtual machine in the source host, and determine usage rates of multiple types of resources of the source host according to usage rates of multiple types of resources of the virtual machines. .

Optionally, each of the virtual machines has the same type of resources, and the obtaining unit 501 is specifically configured to:

The dividing unit 502 is configured to divide the multiple types of resources of the source host into overload resources and non-overload resources according to the usage rate of the multiple types of resources of the source host determined by the obtaining unit 501, and determine the overload resources. The first migration weight and the second migration weight of the unoverloaded resource.

The dividing unit 502 is specifically configured to: when the usage rate of the first type resource is greater than a preset threshold, the first type is used. The resource is determined to be an overload resource;

The dividing unit 502 is further configured to determine, according to the usage rate of the overload resource and a preset first algorithm, the first migration weight of the overload resource, where the overload resource is used;

The selecting unit 503 is configured to select, according to the usage rate of the multiple types of resources of the respective virtual machines, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource, from the virtual machine. Migrate virtual machines.

The selecting unit 503 is specifically configured to: acquire the virtual machines according to the usage rate of the multiple types of resources of the virtual machines, the first migration weight of the overloaded resources, and the second migration weight of the un-overloaded resources. First migration value;

The sending unit 504 is configured to send a migration instruction to the monitor of the source host, so that the monitor migrates the to-be-migrated virtual machine selected by the selecting unit 503 to the target host according to the migration instruction.

Optionally, the selecting unit 503 is further specifically configured to:

Optionally, the device further includes: a comparing unit 505, configured to compare, for each type of overload resource in the overload resource of the source host, a usage rate of the type of overload resource of each virtual machine The maximum utilization rate of the overload resource is used as the target usage rate of the overload resource of the type;

Optionally, the selecting unit 503 is specifically configured to: calculate a first migration value of each virtual machine according to the following formula:

The device provided in the embodiment of the present invention is embedded in the method provided by the method embodiment of the present invention. Therefore, the specific working process of the device provided by the present invention is not described herein.

The device for acquiring the virtual machine provided by the embodiment of the present invention, the obtaining unit 501 acquires the usage rate of the multiple types of resources of each virtual machine in the source host, and determines the usage rate according to the usage rate of the multiple types of resources of the virtual machines. The usage rate of the multiple types of resources of the source host; the dividing unit 502 divides the multiple types of resources of the source host into overload resources and non-overload resources according to the usage rate of the multiple types of resources of the source host, and determines a first migration weight of the overload resource and a second migration weight of the overloaded resource; the selecting unit 503, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overload resource, and the Second migration right of unoverloaded resources Selecting a virtual machine to be migrated from the virtual machine; the sending unit 504 sends a migration instruction to the monitor of the source host, so that the monitor migrates the virtual machine to be migrated to the migration instruction according to the migration instruction Target host. Therefore, the correctness of the selection of the virtual machine to be migrated can be improved, so that the problem of overloading the source host can be quickly solved, and the purpose of saving computer resources is achieved.

Equipment example

FIG. 6 is a schematic diagram of a migration device of a virtual machine according to an embodiment of a device according to the present invention. As shown in FIG. 6, the migration device 600 includes a processor 602, a memory 604, a communication interface 606, and a bus 608. Among them, the processor 602, the memory 604, and the communication interface 606 implement a communication connection with each other through the bus 608.

The processor 602 can be a general-purpose central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), or one or more integrated circuits for executing related programs. The technical solution provided by the foregoing method embodiments of the present invention is implemented.

The memory 604 may be a read only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). Memory 604 can store operating systems and other applications. When the technical solution provided by the embodiment of the present invention is implemented by software or firmware, the program code for implementing any of the optional technical solutions provided by the foregoing method embodiments of the present invention is stored in the memory 604 and executed by the processor 602. .

Communication interface 606 is used to communicate with other switches or control servers in the SDN network.

Bus 608 can include a path for communicating information between various components of migration device 600, such as processor 602, memory 604, and communication interface 606.

The migration device of the virtual machine provided by the device embodiment of the present invention can improve the correctness of the selection of the virtual machine to be migrated, thereby quickly solving the problem of overloading the source host and achieving the purpose of saving computer resources.

A person skilled in the art will further appreciate that the elements and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. Implementations, in order to clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the solution. A person skilled in the art can use different methods for implementing the described functions for each particular application, but such implementation should not be considered to be beyond the scope of the present invention.

The steps of a method or algorithm described in connection with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be placed in random access memory (RAM), memory, read only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or technical field. Any other form of storage medium known.

The specific embodiments of the present invention have been described in detail with reference to the preferred embodiments of the present invention. All modifications, equivalent substitutions, improvements, etc., made within the spirit and scope of the invention are intended to be included within the scope of the invention.

Claims

A method for migrating a virtual machine, the method comprising:

The resource manager obtains usage rates of multiple types of resources of each virtual machine in the source host, and determines usage rates of multiple types of resources of the source host according to usage rates of multiple types of resources of the virtual machines;

Dividing, according to the usage rate of the multiple types of resources of the source host, the multiple types of resources of the source host into an overload resource and an un-overload resource, and determining a first migration weight of the overload resource and the un-overloaded resource. Second migration weight;

Selecting a virtual machine to be migrated from the virtual machine according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the unoverloaded resource;

Sending a migration instruction to the monitor of the source host, so that the monitor migrates the virtual machine to be migrated to the target host according to the migration instruction.
The method according to claim 1, wherein each of the virtual machines has the same type of resources, and the plurality of types of the source hosts are determined according to usage rates of the plurality of types of resources of the respective virtual machines. Resource usage includes:

Determining the usage rate of the type resource of the source host according to the usage rate of the same type of resources of the respective virtual machines and the size of the type resources allocated to the respective virtual machines in advance; until the source is determined The usage rate of each type of resource for the host.
The method according to claim 1 or 2, wherein the dividing the plurality of types of resources of the source host into overload resources and un-overloaded resources according to the usage rate of the plurality of types of resources of the source host includes :

And determining, by the first type of resources, the first type of resources, when the usage rate of the first type of resources is greater than a preset threshold, determining the first type of resources as an overload resource;

When the usage rate of the first type of resource is not greater than a preset threshold, the first type of resource is determined as an unoverloaded resource.
The method according to any one of claims 1-3, wherein the determining the first migration weight of the overload resource and the second migration weight of the un-overloaded resource comprises:

Determining, by the overload resource, a first migration weight of the overload resource according to the usage rate of the overload resource and a preset first algorithm;

And determining, by the un-overloaded resource, a second migration weight of the un-overloaded resource according to the usage rate of the un-overloaded resource and a preset second algorithm.
The method according to any one of claims 1-4, wherein the usage rate of the plurality of types of resources of the respective virtual machines, the first migration weight of the overloaded resources, and the un-overloaded resources The second migration weight, the virtual machine to be migrated from the virtual machine includes:

Acquiring, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource, acquiring a first migration value of each virtual machine;

The first migration value of each virtual machine is compared, and the virtual machine corresponding to the smallest first migration value is determined as the virtual machine to be migrated.
The method according to any one of claims 1-4, wherein the usage rate of the plurality of types of resources of the respective virtual machines, the first migration weight of the overloaded resources, and the un-overloaded resources The second migration weight, the virtual machine to be migrated from the virtual machine includes:

Acquiring, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource, acquiring a first migration value of each virtual machine;

Determining, according to a first migration value of each of the virtual machines and a size of memory allocated to the respective virtual machines in advance, a second migration value of each of the virtual machines;

The second migration value of each virtual machine is compared, and the virtual machine corresponding to the smallest second migration value is determined as the virtual machine to be migrated.
The method according to claim 5 or 6, wherein before the obtaining the first migration value of the respective virtual machines, the method further comprises:

Comparing the usage rate of the type of overload resources of each virtual machine to each type of overload resource in the overload resource of the source host, and using the maximum overload resource usage rate as the target of the type overload resource Usage rate

Comparing the usage rates of the types of non-overloaded resources of the virtual machines to each of the un-overloaded resources of the source host, and using the minimum overload resource usage rate as the type is not overloaded. The target usage rate of the resource.
The method according to claim 7, wherein the usage rate of the plurality of types of resources of the respective virtual machines, the first migration weight of the overloaded resources, and the second migration weight of the un-overloaded resources Obtaining the first migration value of each of the virtual machines includes:

Calculating the first migration value of each virtual machine according to the following formula:

Wherein, D i is the first migration value of the i-th virtual machine, n is the number of overload resource types of the source host, and A t is the first migration weight of the t-type overload resource of the source host, A t >1, S it is the usage rate of the t-th type resource of the i-th virtual machine, P t is the target usage rate of the t-type overload resource of the source host, and m is the source host The number of overload resource types, B l is the second migration weight of the first type of un-overloaded resources of the source host, 0<B l <1, and T il is the use of the first type of resources of the i-th virtual machine. Rate, Q l is the target usage rate of the first type of un-overloaded resources of the source host, and n+m is the number of resource types of the source host.
A device for migrating a virtual machine, the device comprising: an obtaining unit, a dividing unit, a selecting unit, and a sending unit;

The obtaining unit is configured to obtain usage rates of multiple types of resources of each virtual machine in the source host, and determine usage of multiple types of resources of the source host according to usage rates of multiple types of resources of the virtual machines. rate;

The dividing unit is configured to divide the multiple types of resources of the source host into overload resources and non-overload resources according to the usage rate of the multiple types of resources of the source host determined by the acquiring unit, And determining a first migration weight of the overload resource and a second migration weight of the unoverloaded resource;

The selecting unit is configured to select, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overload resource, and the second migration weight of the un-overloaded resource, from the virtual machine The virtual machine to be migrated;

The sending unit is configured to send a migration instruction to the monitor of the source host, so that the monitor migrates the to-be-migrated virtual machine selected by the selecting unit to the target host according to the migration instruction.
The device according to claim 9, wherein each of the virtual machines has the same type of resources, and the obtaining unit is specifically configured to:

Determining the usage rate of the type resource of the source host according to the usage rate of the same type of resources of the respective virtual machines and the size of the type resources allocated to the respective virtual machines in advance; until the source is determined The usage rate of each type of resource for the host.
The device according to claim 9 or 10, wherein the dividing unit is specifically configured to:

And determining, by the first type of resources, the first type of resources, when the usage rate of the first type of resources is greater than a preset threshold, determining the first type of resources as an overload resource;

When the usage rate of the first type of resource is not greater than a preset threshold, the first type of resource is determined as an unoverloaded resource.
The device according to any one of claims 9 to 11, wherein the dividing unit is specifically configured to:

Determining, by the overload resource, a first migration weight of the overload resource according to the usage rate of the overload resource and a preset first algorithm;

And determining, by the un-overloaded resource, a second migration weight of the un-overloaded resource according to the usage rate of the un-overloaded resource and a preset second algorithm.
The device according to any one of claims 9 to 12, wherein the selecting unit is specifically configured to:

Acquiring, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource, acquiring a first migration value of each virtual machine;

The first migration value of each virtual machine is compared, and the virtual machine corresponding to the smallest first migration value is determined as the virtual machine to be migrated.
The device according to any one of claims 9 to 12, wherein the selecting unit is specifically configured to:

Acquiring, according to the usage rate of the multiple types of resources of the virtual machine, the first migration weight of the overloaded resource, and the second migration weight of the un-overloaded resource, acquiring a first migration value of each virtual machine;

Determining, according to a first migration value of each of the virtual machines and a size of memory allocated to the respective virtual machines in advance, a second migration value of each of the virtual machines;

The second migration value of each virtual machine is compared, and the virtual machine corresponding to the smallest second migration value is determined as the virtual machine to be migrated.
The device according to claim 13 or 14, wherein the device further comprises: a comparison unit, configured to load each type of overload resource in the overload resource of the source host, where the virtual machine is located The usage rate of the type overload resource is compared, and the usage rate of the largest overload resource is used as the target usage rate of the overload resource of the type;

Comparing the usage rates of the types of non-overloaded resources of the virtual machines to each of the un-overloaded resources of the source host, and using the minimum overload resource usage rate as the type is not overloaded. The target usage rate of the resource.
The device according to claim 15, wherein the selecting unit is specifically configured to:

Calculating the first migration value of each virtual machine according to the following formula:

Wherein, D i is the first migration value of the i-th virtual machine, n is the number of overload resource types of the source host, and A t is the first migration weight of the t-type overload resource of the source host, A t >1, S it is the usage rate of the t-th type resource of the i-th virtual machine, P t is the target usage rate of the t-type overload resource of the source host, and m is the source host The number of overload resource types, B l is the second migration weight of the first type of un-overloaded resources of the source host, 0<B l <1, and T il is the use of the first type of resources of the i-th virtual machine. Rate, Q l is the target usage rate of the first type of un-overloaded resources of the source host, and n+m is the number of resource types of the source host.
A migration device of a virtual machine, wherein the migration device includes:

a processor, a memory, a communication interface, and a bus, wherein the processor, the memory, and the communication interface communicate via the bus;

The communication interface is configured to communicate with a switch or a control server;

The memory is used to store a program;

The processor is configured to execute the program stored by the memory to perform the method of any one of claims 1 to 8 when the migration device is in operation.