Virtual machine live migration implementation method and system across CPU model nodes

Abstract

The invention relates to the technical field of cloud computing, in particular to a virtual machine live migration implementation method and system across CPU model nodes, and the method comprises the following steps: preparing VMM software: preparing virtual machine management software VMM which supports a built-in CPU model and a corresponding characteristic list and supports the reading of an extended CPU definition configuration file; cluster deployment: collecting and preparing models of CPUs with the same architecture in the environment and models of CPUs possibly added subsequently; the virtual machine live migration implementation method and system of the cross-CPU model node have the advantages that the CPU model supported by a simulator is defined in the mode that the CPU definition configuration file is built in and can be configured and expanded in the virtual machine management software (VMM), the CPU model of the self-defined virtual machine is configured when the virtual machine is started, the CPU characteristics of opening and closing are allowed to be configured, and the virtual machine live migration implementation method and system of the cross-CPU model node are achieved. And a compatible target CPU model list is allowed to be added, so that a compatible CPU characteristic common set is automatically calculated when the virtual machine is started, the virtual machine is started by the compatible CPU characteristics, and virtual machine thermal migration across CPU model nodes under the same rack is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud computing, and in particular to a method and system for implementing hot migration of virtual machines across CPU model nodes. Background Art

[0002] As more and more cloud vendors provide cloud services, the equipment and models of cloud centers are constantly iterating and updating. Processors with the same architecture but different models will be added to the computing resource pool of the cloud center. If the old models have not reached the replacement period or continue to be used, the computing nodes of the new and old models need to provide services to users together. Users will apply to create virtual machines on the computing nodes and perform various management operations. At this time, the virtual machines created by the new nodes may be subject to cold migration, hot migration, specification changes, etc. At this time, the scheduler will allow the virtual machines to be scheduled to nodes with common CPU models or nodes with the same architecture but different CPU models. If the specific model is not specified when the virtual machine is created or some VCPUs with incompatible features are turned on / off, the migration may fail during hot migration due to different VCPU features (which cannot be changed during runtime). In some cases, the new CPU will be compatible with the features of the old CPU, that is, the KVM virtual machine started by the old node can generally be hot migrated to the new node, otherwise it will fail.

[0003] In conventional deployments, cloud vendors will divide computing nodes of different CPU models into different resource pools, even if they have the same architecture and CPU brand. The nodes that can be scheduled during virtual machine hot migration do not include nodes from other resource pools.

[0004] Therefore, it is necessary to solve the problem that in some cases, such as when the number of machines is limited and nodes of different models can only be classified into the same resource pool, the scale of the computing node resource pool can be increased, and the virtual machine cannot be shut down, especially when different models of domestic CPUs under the same architecture are used, hot migration of virtual machines will fail. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for implementing hot migration of virtual machines across CPU model nodes to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for implementing hot migration of virtual machines across CPU model nodes, the method comprising the following steps:

[0007] VMM software preparation: prepare virtual machine management software VMM that supports built-in CPU models and corresponding feature lists and supports reading extended CPU definition configuration files;

[0008] Cluster deployment, collect and prepare CPU models of the same architecture in the environment and CPU models that may be added later, and write the extended CPU list that needs to be written in addition to the built-in CPU models of the VMM in the form of CPU definition configuration files for use by the VMM in the software preparation stage; in the VMM software preparation stage, the generation from the configuration file definition means that after the new CPU is launched, the R&D personnel need to extract the required content according to the CPU design document to generate the mentioned configuration file, which will be used when the corresponding model is added to the resource pool in the future, so that this version of VMM can support the new CPU;

[0009] Live migration of virtual machines.

[0010] Preferably, the specific operations of VMM software preparation include:

[0011] Modify the built-in CPU models and feature correspondence of each VMM architecture, built-in common CPU models and feature definitions, including but not limited to CPU model name, CPU feature name, CPU feature identification bit and CPU feature identification bit value definition, and the defined correspondence;

[0012] Modify VMM to add support for reading extended CPU definition configuration files, allowing reading CPU model names, CPU feature names, CPU feature identifiers, and CPU feature identifier value definitions defined in common formats, as well as the corresponding relationships between definition types;

[0013] Modify VMM to support reading additional enabled or disabled VCPU features, support reading compatible VCPU models other than the target VCPU model, and calculate compatible public VCPU feature sets based on all VCPU lists defined in built-in and externally read CPU definition configuration files. Unsupported features are disabled, and feature enablement results are returned or allowed to be queried.

[0014] Modify the VMM to allow the feature enable bit of the VCPU including but not limited to KVM acceleration to be set according to the definition of the VCPU feature set, so that the virtual machine can be started in a form compatible with multiple VCPUs and support KVM acceleration;

[0015] Collect specification manuals of various server CPU models, and record the feature list and enabling status of the CPU in the form of an extended CPU definition configuration file, including but not limited to the CPU model name, CPU feature name, CPU feature identifier and CPU feature identifier value definition, as well as the corresponding relationship between the definitions.

[0016] Preferably, the specific operations of cluster deployment include:

[0017] Determine the computing nodes that need to be divided into the same computing resource pool, determine the node target VCPU model and compatible VCPU model; set the computing resource pool cluster configuration file, including but not limited to: virtual machine VCPU baseline model, compatible VCPU model, and enabled or disabled VCPU features.

[0018] Preferably, the specific operations of virtual machine hot migration include:

[0019] The source node receives the hot migration request, and the Hypervisor management program sends the source node virtual machine configuration file to the target node, and the target node prepares to start the virtual machine; the target node reads the virtual machine VCPU model, feature list, and compatible CPU model in the configuration file based on the virtual machine configuration file sent by the source node to start the VMM; the target node VMM calculates the compatible public VCPU feature set based on the startup parameters and the built-in and read extended CPU definition configuration files, so that the source node and target node VCPU features are the same, starts the compatible VCPU virtual machine and listens to the hot migration data transmitted by the source node; the source node sends the status data of the virtual machine to complete the hot migration of the virtual machine.

[0020] A system for implementing hot migration of virtual machines across CPU model nodes, the system comprising a software preparation module, a deployment module and a hot migration module;

[0021] The software preparation module is used for VMM software preparation, which includes preparing the virtual machine management software VMM that supports the built-in CPU models and corresponding feature lists and supports reading the extended CPU definition configuration files;

[0022] The deployment module is used for cluster deployment, collecting and preparing CPU models of the same architecture in the environment and CPU models that may be added later. In addition to the CPU models built into the VMM, the extended CPU list that needs to be written is provided in the form of a CPU definition configuration file for use by the VMM in the software preparation stage.

[0023] Hot migration module, used for hot migration of virtual machines.

[0024] Preferably, the software preparation module modifies the built-in CPU model and feature correspondence of each VMM architecture, and the built-in commonly used CPU model and feature definition, including but not limited to the CPU model name, CPU feature name, CPU feature identification bit and CPU feature identification bit value definition, and the defined correspondence;

[0025] Modify VMM to add support for reading extended CPU definition configuration files, allowing reading CPU model names, CPU feature names, CPU feature identifiers, and CPU feature identifier value definitions defined in common formats, as well as the corresponding relationships between definition types;

[0026] Modify VMM to support reading additional enabled or disabled VCPU features, support reading compatible VCPU models other than the target VCPU model, and calculate compatible public VCPU feature sets based on all VCPU lists defined in built-in and externally read CPU definition configuration files. Unsupported features are disabled, and feature enablement results are returned or allowed to be queried.

[0027] Modify the VMM to allow the feature enable bit of the VCPU including but not limited to KVM acceleration to be set according to the definition of the VCPU feature set, so that the virtual machine can be started in a form compatible with multiple VCPUs and support KVM acceleration;

[0028] Collect specification manuals of various server CPU models, and record the feature list and enabling status of the CPU in the form of an extended CPU definition configuration file, including but not limited to the CPU model name, CPU feature name, CPU feature identifier and CPU feature identifier value definition, as well as the corresponding relationship between the definitions.

[0029] Preferably, the deployment module determines the computing nodes that need to be divided into the same computing resource pool, determines the node target VCPU model and compatible VCPU model; and sets the computing resource pool cluster configuration file, the configuration including but not limited to: virtual machine VCPU baseline model, VCPU model to be compatible, and VCPU features turned on or off.

[0030] Preferably, in the hot migration module, the source node receives the hot migration request, the Hypervisor management program sends the source node virtual machine configuration file to the target node, and the target node prepares to start the virtual machine; the target node reads the virtual machine VCPU model, feature list, and compatible CPU model in the configuration file according to the virtual machine configuration file sent by the source node to start the VMM; the target node VMM calculates the compatible public VCPU feature set based on the startup parameters and the built-in and read extended CPU definition configuration files, so that the source node and the target node VCPU features are the same, starts the compatible VCPU virtual machine and monitors the hot migration data transmitted by the source node; the source node sends the status data of the virtual machine to complete the hot migration of the virtual machine.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention proposes a method and system for implementing hot migration of virtual machines across CPU model nodes, which defines the CPU models supported by the simulator by means of a built-in and configurable extended CPU definition configuration file in the virtual machine management software (VMM). In addition to configuring the custom virtual machine CPU model when the virtual machine is started, it allows the configuration of CPU features that are turned on and off, and allows the addition of a compatible target CPU model list, so that a common set of compatible CPU features is automatically calculated when the virtual machine is started, and the virtual machine is started with compatible CPU features, thereby implementing hot migration of virtual machines across CPU model nodes in the same frame.

[0033] It enables hot migration of virtual machines between computing nodes of the same architecture, especially hot migration between different CPU models of the same architecture and the same manufacturer / different manufacturers, breaking the ecological isolation of different CPU models of the same architecture; allowing more server models of the same architecture but different CPU models to be added to the same resource pool. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Creating a flow chart for the virtual machine of the present invention;

[0035] Figure 2 This is a flowchart of the hot migration of a virtual machine according to the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose and technical solution of the present invention clearly and completely described, and the advantages more clearly understood, the embodiments of the present invention are further described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of the embodiments, and are only used to explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention. All other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] Embodiment 1

[0038] See also Figure 1 to Figure 2 The present invention provides a technical solution: a method for implementing hot migration of virtual machines across CPU model nodes, the method comprising the following steps:

[0039] a) VMM software preparation stage. Modify the VMM software to support built-in CPU models and corresponding feature lists, and support reading extended CPU definition configuration files

[0040] i. Modify the built-in CPU model and feature correspondence of each VMM architecture, built-in common CPU model and feature definitions, including but not limited to CPU model name, CPU feature name, CPU feature identification bit and CPU feature identification bit value definition, and the correspondence between the above definitions.

[0041] ii. Modify VMM to add support for reading extended CPU definition configuration files, allowing reading of CPU model names, CPU feature names, CPU feature identifiers, and CPU feature identifier value definitions defined in common formats, as well as the correspondence between the above definition types. For example, model A CPU includes feature AA, and the feature identifiers of AA are from the Mth to the Nth bit. When the value is 1, it indicates support for basic AA functions.

[0042] iii. Modify the VMM to support reading additional enabled or disabled VCPU features, support reading compatible VCPU models other than the target VCPU model, and calculate the compatible public VCPU feature set based on all VCPU lists defined in the built-in and externally read CPU definition configuration files. Unsupported features are disabled, and the feature enabling results are returned or allowed to be queried.

[0043] iv. Modify the VMM to allow setting of feature enable bits of VCPUs including but not limited to KVM acceleration according to the definition of the VCPU feature set, so that the virtual machine can be started in a form compatible with multiple VCPUs and support KVM acceleration.

[0044] v. Collect various server CPU model specifications, and record the CPU feature list and enabled status in the form of an extended CPU definition configuration file, including but not limited to the CPU model name, CPU feature name, CPU feature bit and CPU feature bit value definition, as well as the corresponding relationship between the above definitions

[0045] b) Cluster deployment phase

[0046] i. Collect and prepare CPU models of the same architecture in the environment and CPU models that may be added later. In addition to the CPU models built into the VMM, write a list of extended CPUs that need to be written in the form of CPU definition configuration files for use by the VMM in the software preparation phase, to make up for the extended VCPU models that were not fully defined in the preparation phase. If the extended VCPUs in the preparation phase have covered the VPU models included in the computing resource pool, only the VMM software and its corresponding configuration need to be deployed.

[0047] ii. Determine the computing nodes that need to be divided into the same computing resource pool, and determine the node target VCPU model and compatible VCPU model

[0048] iii. Set up the computing resource pool cluster configuration file, including but not limited to: virtual machine VCPU baseline model, compatible VCPU model, and enabled or disabled VCPU features.

[0049] c) Hot migration stage

[0050] i. The source node receives the hot migration request, and the Hypervisor management program (such as libvirt or other management applications) sends the source node virtual machine configuration file to the target node, and the target node prepares to start the virtual machine.

[0051] ii. The target node reads the virtual machine VCPU model, feature list, and compatible CPU model in the configuration file according to the virtual machine configuration file sent by the source node and starts the VMM.

[0052] iii. The target node VMM calculates the compatible public VCPU feature set based on the startup parameters and the built-in and read extended CPU definition configuration files, so that the source node and target node VCPU features are the same, starts the compatible VCPU virtual machine and listens to the hot migration data transmitted by the source node.

[0053] iv. The source node sends the status data of the virtual machine, such as VCPU register values, memory pages, etc., to complete the hot migration of the virtual machine.

[0054] After the configuration, the computing resource pool can pass the target CPU model, CPU feature enable / disable list, and compatible CPU model list of the virtual machine to be started to the VMM when the computing node virtual machine is created, so that the VMM can automatically calculate the compatible CPU feature common set, enable or disable the VCPU feature flag, and start the virtual machine that meets the feature requirements, so as to realize the hot migration of virtual machines between nodes with different CPU models in the same rack.

[0055] Embodiment 2

[0056] Based on the first embodiment, a system for implementing hot migration of virtual machines across CPU model nodes is proposed, the system comprising a software preparation module, a deployment module and a hot migration module;

[0057] The software preparation module is used for VMM software preparation, preparing virtual machine management software VMM that supports built-in CPU models and corresponding feature lists and supports reading extended CPU definition configuration files; modifying the built-in CPU models and feature correspondence of each VMM architecture, built-in common CPU models and feature definitions, including but not limited to CPU model name, CPU feature name, CPU feature identification bit and CPU feature identification bit value definition, and the defined correspondence;

[0058] Modify VMM to add support for reading extended CPU definition configuration files, allowing reading CPU model names, CPU feature names, CPU feature identifiers, and CPU feature identifier value definitions defined in common formats, as well as the corresponding relationships between definition types;

[0059] Modify VMM to support reading additional enabled or disabled VCPU features, support reading compatible VCPU models other than the target VCPU model, and calculate compatible public VCPU feature sets based on all VCPU lists defined in built-in and externally read CPU definition configuration files. Unsupported features are disabled, and feature enablement results are returned or allowed to be queried.

[0060] Modify the VMM to allow the feature enable bit of the VCPU including but not limited to KVM acceleration to be set according to the definition of the VCPU feature set, so that the virtual machine can be started in a form compatible with multiple VCPUs and support KVM acceleration;

[0061] Collect specification manuals of various server CPU models, and record the feature list and enabling status of the CPU in the form of an extended CPU definition configuration file, including but not limited to the CPU model name, CPU feature name, CPU feature identifier and CPU feature identifier value definition, as well as the corresponding relationship between the definitions.

[0062] The deployment module is used for cluster deployment, collecting and preparing CPU models of the same architecture in the environment and CPU models that may be added later. In addition to the CPU models built into the VMM, the extended CPU list that needs to be written is provided in the form of a CPU definition configuration file for use by the VMM in the software preparation stage; determining the computing nodes that need to be divided into the same computing resource pool, determining the target VCPU model of the node and the compatible VCPU model; setting the computing resource pool cluster configuration file, including but not limited to: virtual machine VCPU baseline model, VCPU model to be compatible, and VCPU features that are enabled or disabled.

[0063] The hot migration module is used for hot migration of virtual machines. The source node receives the hot migration request, and the Hypervisor management program sends the source node virtual machine configuration file to the target node, and the target node prepares to start the virtual machine; the target node reads the virtual machine VCPU model, feature list, and compatible CPU model in the configuration file according to the virtual machine configuration file sent by the source node to start the VMM; the target node VMM calculates the compatible public VCPU feature set based on the startup parameters and the built-in and read extended CPU definition configuration files, so that the source node and target node VCPU features are the same, starts the compatible VCPU virtual machine and monitors the hot migration data transmitted by the source node; the source node sends the status data of the virtual machine to complete the hot migration of the virtual machine.

[0064] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for implementing hot migration of virtual machines across CPU model nodes, characterized in that: The method comprises the following steps: VMM software preparation: prepare virtual machine management software VMM that supports built-in CPU models and corresponding feature lists and supports reading extended CPU definition configuration files; Cluster deployment, collect and prepare CPU models of the same architecture in the environment and CPU models that may be added later, and write the extended CPU list that needs to be written in addition to the built-in CPU models of the VMM in the form of CPU definition configuration files for use by the VMM in the software preparation stage; Live migration of virtual machines.

2. According to a method for implementing hot migration of virtual machines across CPU model nodes as described in claim 1, it is characterized in that: The specific operations of VMM software preparation include: Modify the built-in CPU models and feature correspondence of each VMM architecture, built-in common CPU models and feature definitions, including but not limited to CPU model name, CPU feature name, CPU feature identification bit and CPU feature identification bit value definition, and the defined correspondence; Modify VMM to add support for reading extended CPU definition configuration files, allowing reading CPU model names, CPU feature names, CPU feature identifiers, and CPU feature identifier value definitions defined in common formats, as well as the corresponding relationships between definition types; Modify VMM to support reading additional enabled or disabled VCPU features, support reading compatible VCPU models other than the target VCPU model, and calculate compatible public VCPU feature sets based on all VCPU lists defined in built-in and externally read CPU definition configuration files. Unsupported features are disabled, and feature enablement results are returned or allowed to be queried. Modify the VMM to allow the feature enable bit of the VCPU including but not limited to KVM acceleration to be set according to the definition of the VCPU feature set, so that the virtual machine can be started in a form compatible with multiple VCPUs and support KVM acceleration; Collect specification manuals of various server CPU models, and record the feature list and enabling status of the CPU in the form of an extended CPU definition configuration file, including but not limited to the CPU model name, CPU feature name, CPU feature identifier and CPU feature identifier value definition, as well as the corresponding relationship between the definitions.

3. According to a method for implementing hot migration of virtual machines across CPU model nodes as described in claim 1, it is characterized in that: The specific operations of cluster deployment include: Determine the computing nodes that need to be divided into the same computing resource pool, determine the node target VCPU model and compatible VCPU model; set the computing resource pool cluster configuration file, including but not limited to: virtual machine VCPU baseline model, compatible VCPU model, and enabled or disabled VCPU features.

4. According to a method for implementing hot migration of virtual machines across CPU model nodes as described in claim 1, it is characterized in that: The specific operations of virtual machine hot migration include: The source node receives the hot migration request, and the Hypervisor management program sends the source node virtual machine configuration file to the target node, and the target node prepares to start the virtual machine; the target node reads the virtual machine VCPU model, feature list, and compatible CPU model in the configuration file based on the virtual machine configuration file sent by the source node to start the VMM; the target node VMM calculates the compatible public VCPU feature set based on the startup parameters and the built-in and read extended CPU definition configuration files, so that the source node and target node VCPU features are the same, starts the compatible VCPU virtual machine and listens to the hot migration data transmitted by the source node; the source node sends the status data of the virtual machine to complete the hot migration of the virtual machine.

5. A system for implementing hot migration of virtual machines across CPU model nodes according to the method for implementing hot migration of virtual machines across CPU model nodes according to any one of claims 1 to 4, characterized in that: The system consists of a software preparation module, a deployment module and a hot migration module; The software preparation module is used for VMM software preparation, which includes preparing the virtual machine management software VMM that supports the built-in CPU models and corresponding feature lists and supports reading the extended CPU definition configuration files; The deployment module is used for cluster deployment, collecting and preparing CPU models of the same architecture in the environment and CPU models that may be added later. In addition to the CPU models built into the VMM, the extended CPU list that needs to be written is provided in the form of a CPU definition configuration file for use by the VMM in the software preparation stage. Hot migration module, used for hot migration of virtual machines.

6. According to the system for implementing hot migration of virtual machines across CPU model nodes as described in claim 5, it is characterized in that: The software preparation module modifies the built-in CPU model and feature correspondence of each VMM architecture, and the built-in common CPU model and feature definition, including but not limited to the CPU model name, CPU feature name, CPU feature identification bit and CPU feature identification bit value definition, and the defined correspondence; Modify VMM to add support for reading extended CPU definition configuration files, allowing reading CPU model names, CPU feature names, CPU feature identifiers, and CPU feature identifier value definitions defined in common formats, as well as the corresponding relationships between definition types; Modify VMM to support reading additional enabled or disabled VCPU features, support reading compatible VCPU models other than the target VCPU model, and calculate compatible public VCPU feature sets based on all VCPU lists defined in built-in and externally read CPU definition configuration files. Unsupported features are disabled, and feature enablement results are returned or allowed to be queried. Modify the VMM to allow the feature enable bit of the VCPU including but not limited to KVM acceleration to be set according to the definition of the VCPU feature set, so that the virtual machine can be started in a form compatible with multiple VCPUs and support KVM acceleration; Collect specification manuals of various server CPU models, and record the feature list and enabling status of the CPU in the form of an extended CPU definition configuration file, including but not limited to the CPU model name, CPU feature name, CPU feature identifier and CPU feature identifier value definition, as well as the corresponding relationship between the definitions.

7. According to a system for implementing hot migration of virtual machines across CPU model nodes as described in claim 5, it is characterized in that: The deployment module determines the computing nodes that need to be divided into the same computing resource pool, determines the node target VCPU model and compatible VCPU model; and sets the computing resource pool cluster configuration file, the configuration includes but is not limited to: virtual machine VCPU baseline model, VCPU model to be compatible, and VCPU features turned on or off.

8. The system for implementing hot migration of virtual machines across CPU model nodes according to claim 5 is characterized in that: In the hot migration module, the source node receives the hot migration request, the Hypervisor management program sends the source node virtual machine configuration file to the target node, and the target node prepares to start the virtual machine; the target node reads the virtual machine VCPU model, feature list, and compatible CPU model in the configuration file according to the virtual machine configuration file sent by the source node to start the VMM; the target node VMM calculates the compatible public VCPU feature set based on the startup parameters and the built-in and read extended CPU definition configuration files, so that the source node and the target node VCPU features are the same, starts the compatible VCPU virtual machine and monitors the hot migration data transmitted by the source node; the source node sends the status data of the virtual machine to complete the hot migration of the virtual machine.

Images