CN106502721B - A command unloading method, device and physical machine - Google Patents

Abstract

The invention discloses a command unloading method, device and physical machine, relates to the technical field of communication, and can effectively reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copying or clearing instructions). The method of the present invention includes: the virtual machine intercepts the operation instruction issued by the application program, and the operation instruction is used to instruct to perform a specified operation on the data to be operated; the virtual machine generates an unload command according to the operation instruction, and sends the unload command to the host, and the unload command includes copying Unload command or clear unload command; the host converts the virtual source address in the unload command into the physical source address of the data to be operated, and the virtual source address is used to indicate the location of the data to be operated in the virtual disk of the virtual machine; The virtual destination address in and the physical source address of the data to be operated perform the specified operation on the data to be operated. The present invention is applicable to command offloading process.

Description

A command unloading method, device and physical machine

technical field

The present invention relates to the field of communication technology, in particular to a command offloading method, device and physical machine.

Background technique

At present, whether it is to realize the unloading process of copying between virtual machines or the clearing command unloading process of the virtual machine itself, it is necessary for the virtual machine to send the operation command through the IO (English: Input /Output, Chinese: Input/Output) ring is sent to the back-end process, and the host computer generates the corresponding SCSI (English: Small Computer System Interface, Chinese: Small Computer System Interface) command according to the operation command, and then sends it to such as IP SAN (English: IP based SAN, Chinese: IP-based SAN device) and other physical storage devices, the physical storage device parses and executes the SCSI command, thereby completing data copy or clearing operations on the physical storage device.

It can be seen that, during the execution of the above operations, the generated SCSI command needs to be finally sent to the corresponding physical storage device, and then the physical storage device performs the corresponding operation. However, considering that the copying or clearing of the above data needs to be completed by the physical layer hardware device of the physical machine, and even requires the participation of the CPU (English: Central Processing Unit, Chinese: central processing unit) and network resources of the physical machine, High requirements on the hardware resources of the physical machine also lead to a waste of the hardware resources of the physical machine and the use of network resources.

Contents of the invention

The invention provides a command unloading method, device and physical machine, which can reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copying or clearing instructions).

To achieve the above object, the present invention adopts the following technical solutions:

In a first aspect, the present invention provides a command unloading method, the method is used for a physical machine, the physical machine runs a host, and the host runs a virtual machine. The method includes: the virtual machine intercepts the operation instruction issued by the application program, and the operation instruction is used to instruct to perform a specified operation on the data to be operated; the virtual machine generates an unload command according to the operation instruction, and sends the unload command to the host, and the unload command includes Unload command or clear unload command; the host converts the virtual source address in the unload command into the physical source address of the data to be operated, and the virtual source address is used to represent the location of the data to be operated in the virtual disk of the virtual machine; The virtual destination address in the command and the physical source address of the data to be operated perform the specified operation on the data to be operated. It can be seen that although the virtual machine and the host computer are required to complete the execution process of the specified operation together, since the host computer completes the specified operation according to the uninstall command generated by the virtual machine, no matter whether the system currently applying the present invention is a cluster file system , a distributed shared storage system, or other systems, the host can recognize the uninstall command. Moreover, both the host and the virtual machine are part of the physical machine. In the above execution process, the host performs corresponding operations after receiving the unloading command, without the need for a physical storage device to perform corresponding operations. Therefore, using the The implementation manner can effectively reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copy or clear instructions).

In a possible design, when the specified operation is a copy operation, the data to be operated includes at least one data block, and the virtual machine can generate a copy unload command for each data block in the at least one data block, and send each data block to the host Block copy unload command. It should be noted that the virtual machine can generate a copy and unload command for each data block at the same time, and package and send it to the host, or generate a copy and unload command for only one data block at a time, and send it to the host, and then when the host finishes executing the command After the copy operation indicated by the copy unload command, the next copy unload command is generated for execution by the host until all data blocks in the data to be operated are copied.

In a possible design, after the virtual machine intercepts the operation command issued by the application, the virtual machine also needs to send the creation command to the host, and the creation command carries the length of the data block in the copy unload command of each data block , the creation instruction is used to create an empty file in the disk file system of the target virtual machine. The target virtual machine runs on the host where the virtual machine is running, or runs on a host other than the host; the virtual machine receives the feedback of the virtual target from the target virtual machine address, the virtual destination address is the virtual address allocated to the destination data block of the data to be operated in the empty file, and the length of the destination data block is greater than or equal to the length of the data block in the copy unload command of each data block. In this way, it can be ensured that when the host performs the copy operation, a certain space has been reserved for the copy process in the destination virtual machine, that is, the empty file created above.

In a possible design, the method is applied to a distributed shared storage system, and the host performs specified operations on the data to be operated, which can be specifically implemented as: the host updates the virtual destination address of the data to be operated to the physical source address of the data to be operated ; According to the updated virtual destination address, the host adds one to the value of the reference count of the data block corresponding to the physical source address of the data to be operated, and the value of the reference count is used to indicate that the data block corresponding to the physical address of the data to be operated is referenced frequency. It can be seen that the copy operation of the data to be operated using the above method only establishes the mapping relationship between addresses and increases the reference count of the physical source address of the data to be operated. Therefore, for the virtual machine, the consumption of the computing network resources of the virtual machine by the storage operation of the virtual machine is reduced, and at the same time, an acceleration effect is also achieved by increasing the reference count.

In a possible design, the method is applied to the cluster file system, and the host performs specified operations on the data to be operated, which can be implemented as follows: the host establishes a link between the virtual destination address of the data to be operated and the physical source address of the data to be operated Mapping relationship: the host copies the data block corresponding to the physical source address of the data to be operated to the physical destination address of the data to be operated; the host replaces the physical source address of the data to be operated in the mapping relationship with the physical destination address of the data to be operated. It can be seen that to implement the copy operation of the data to be operated in the above manner, it is necessary to first copy the data block from the physical source address to the physical destination address, and then establish a mapping relationship between the virtual source address and the physical destination address. For the virtual machine, the consumption of virtual machine computing network resources by the virtual machine storage operation is reduced, but compared with the former possible design, the effect of effective copy acceleration cannot be achieved.

In a possible design, the process of the host converting the virtual source address in the unload command into the physical source address of the data to be operated can be implemented as follows: when the data to be operated is stored in block storage, the host will The virtual source address of the data to be operated is converted into the physical source address of the data to be operated; or, when the storage method of the data to be operated is object storage, the host converts the virtual source address of the data to be operated into an object identifier, and the object identifier is The identifier of the data block corresponding to the data to be operated. It can be seen that for different storage methods, different address conversion methods need to be adopted.

In a possible design, when the unload command is a clear unload command, the host performs a specified operation on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated, which can be specifically implemented as: When the virtual disk is simulated by the data to be operated, the host will clear each bit in the bitmap of the physical source address of the data to be operated; or, when the virtual disk is simulated by the distributed shared storage system, the host will Release the object identifier of the data to be operated, and release the data block corresponding to the data to be operated; or, when the virtual disk is a virtual disk simulated by the cluster file system, the host will release the data block of the data to be operated; or, when there is no When operating the mapping relationship between the virtual source address of the data and the physical source address of the data to be operated, the host writes zeros to the data block corresponding to the physical source address of the data to be operated. It can be seen that, in the present invention, it is necessary to select an appropriate zeroing scheme according to the simulation situation of the virtual disk.

In a second aspect, the present invention provides a command unloading device, which includes a host Host and a virtual machine. The virtual machine is used to intercept the operation instruction issued by the application program, and the operation instruction is used to instruct to perform a specified operation on the data to be operated; the virtual machine is also used to generate an uninstall command according to the operation instruction, and send the uninstall command to the host. The uninstall command includes Copying the unloading command or clearing the unloading command; the host is used to convert the virtual source address in the unloading command into the physical source address of the data to be operated, and the virtual source address is used to indicate the position of the data to be operated in the virtual disk of the virtual machine; The host computer is also used to perform specified operations on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated. The above command offloading device can be used to complete the functions executed by the host and the virtual machine in the above method example, and the functions can be realized by hardware, or by executing corresponding software by hardware. Wherein, the hardware or software includes at least one module with corresponding functions above. It should be noted that the host computer and the virtual machine in the above-mentioned command offloading device can also be used to implement various functions described in the first aspect.

In a third aspect, the present invention provides a physical machine. The physical machine includes a hardware layer, a host running on the hardware layer, and a virtual machine. The virtual machine is used to intercept the operation instruction issued by the application program, and the operation instruction is used to instruct to perform a specified operation on the data to be operated; the virtual machine is also used to generate an uninstall command according to the operation instruction, and send the uninstall command to the host. The uninstall command includes Copying the unloading command or clearing the unloading command; the host is used to convert the virtual source address in the unloading command into the physical source address of the data to be operated, and the virtual source address is used to indicate the position of the data to be operated in the virtual disk of the virtual machine; The host computer is also used to perform specified operations on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated.

In addition, the physical machine may further include a processor, an input device, and an output device. The processor is configured to support the physical machine to execute corresponding functions in the above method. The input device and the output device can be regarded as communication interfaces for supporting communication between the physical machine and other devices. It should be noted that the physical machine may also include a memory, which is used to couple with the processor, and specifically may include a virtual machine and a host for implementing various functions described in the first aspect, and may also save the physical machine Necessary program instructions and data.

The command unloading method, device and physical machine provided by the present invention, compared with the prior art using physical storage devices to perform specified operations, in the present invention, the virtual machine can generate unloading commands according to the operating instructions issued by the application program , and then the virtual machine sends the unloading command to the host; after the host receives the unloading command sent by the virtual machine, it converts the virtual source address carried by the unloading command into the physical source address of the data to be operated, and then the host can follow the unloading command The carried virtual destination address and the physical source address of the data to be operated are used to perform specified operations on the data to be operated. Although the execution process of the specified operation needs to be completed by the virtual machine and the host computer together like this, because the host computer completes the specified operation according to the unloading command generated by the virtual machine, no matter whether the system currently applying the present invention is a cluster file system, distributed shared The storage system, or other systems, and the host can recognize the uninstall command. Moreover, both the host and the virtual machine are part of the physical machine. In the above execution process, the host performs corresponding operations after receiving the unloading command, without the need for a physical storage device to perform corresponding operations. Therefore, using the The implementation manner can effectively reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copy or clear instructions).

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For Those of ordinary skill in the art can also obtain other drawings based on these drawings without making creative efforts.

FIG. 1 is a schematic structural diagram of a data center provided by an embodiment of the present invention;

FIG. 2 is a schematic structural diagram of another data center provided by an embodiment of the present invention;

FIG. 3 is a schematic diagram of an interaction process between a virtual machine and a host provided by an embodiment of the present invention;

FIG. 4 is an interaction diagram of a command uninstallation method provided by an embodiment of the present invention;

FIG. 5 is an interaction diagram of another command uninstallation method provided by an embodiment of the present invention;

FIG. 6 is an interaction diagram of another command uninstallation method provided by an embodiment of the present invention;

FIG. 7 is a schematic structural diagram of a command unloading device provided by an embodiment of the present invention;

FIG. 8 is a schematic structural diagram of a physical machine provided by an embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

The embodiments of the present invention may be applied to different data center structures, and the embodiments of the present invention are not limited thereto. For example, the data center structure may specifically be a structure provided with a cluster file system, or a Server San structure, and the like.

As shown in FIG. 1 , a data center structure provided with a cluster file system 01 may include multiple hosts, such as a host 100 , a host 200 and a host 300 . One or more virtual machines can be run inside each host, for example, a virtual machine 10 and a virtual machine 11 are running on the host 100, wherein each virtual machine can include at least one virtual disk, for example, the virtual machine 10 includes a virtual disk 12, and there will be a back-end storage service process at the back end of each virtual disk, such as the back-end process 14, so as to provide services for the corresponding virtual machines. Among them, the backend process can be a qemu process, or a special backend process, such as blkback, vhost, etc. The back-end process will send the command transmitted through the IO ring to the cluster file system 01, and the cluster file system 01 will send it to the IP SAN03 through the network 02, and then the command will be processed by the IPSAN03.

As shown in Figure 2, it is a Server San architecture with a distributed shared storage system. The data center can also include multiple hosts, and each host can run one or more virtual machines. The internal structure of the virtual machine Similar to the virtual machine in the architecture shown in FIG. 1 , details are not described here. The difference from Figure 1 is that the backend process will send the command transmitted through the IO ring to the distributed shared storage system 04, and then the distributed shared storage system 04 will forward the command to the corresponding local disk, and the local disk will command is processed. It should be noted that the above local disk may be a local disk inside the host where the virtual machine corresponding to the backend process transmitting the command resides, or may be a virtual disk set on another host. For example, in the host computer 400, the virtual machine 40 finally sends the command to the distributed shared storage system 04 through the back-end process 44, and the distributed shared storage system 04 determines whether to send the command to the local disk 46 or the local disk 47 according to the meaning of the command itself. or send to a local disk located in host 500 or host 600. It should be noted that the number of local disks has nothing to do with the number of virtual machines set inside the host, and can be configured according to its own needs. For example, although the host 500 includes virtual machines 50 and 51, the host 500 only includes the local Similarly, although the host 600 only includes the virtual machine 60 , multiple local disks can be set in the host 600 , that is, a local disk 63 and a local disk 64 .

The embodiment of the present invention provides a command offloading method, which can be applied to the virtual machine 700a shown in Figure 3, and the virtual machine 700a can be located in the above two data center structures, but not limited to the above two data center structures . In the embodiment of the present invention, within the virtual machine 700a, the capture of operation instructions can be realized based on the file system 73posix interface, that is, when the application program issues operation instructions such as copying and clearing, the virtual machine 700a can use the operation instruction capture, and generate a corresponding uninstall command, and pass it to the front-end process 71. Afterwards, the front-end process 71 puts the unload command into the IO ring 72 and passes it to the back-end process 73 inside the host 700b, and the back-end process 73 parses the unload command, and finally delivers it to the distributed shared storage system 04 or the cluster file system 01 to process. The specific implementation manner will be proposed later, and will not be repeated here.

Taking the above two data center structures as an example, as shown in Figure 4, the method can be applied to a physical machine, a host (English: Host) can run on the physical machine, and a virtual machine can also run on the host, the present invention can It is realized jointly by the virtual machine and the host, and the process of the method includes:

101. The virtual machine intercepts the operation instruction issued by the application program.

Wherein, the operation instruction is used to instruct to perform a specified operation on the data to be operated.

The operation instruction issued by the application program of the virtual machine may include a clear instruction or a copy instruction. For example, the APP (English: Application, Chinese: application program) in the virtual machine issues an x length Write 0 operation, or the APP in the source virtual machine sends a copy command to copy file A from directory a to directory b of the destination virtual machine. It should be noted that the copy instruction may include at least one of a cross-virtual machine copy instruction and a virtual machine internal copy instruction that performs a copy operation on the same virtual machine.

It should be noted that, in the present invention, the virtual machine can recognize the operation instruction issued by the application program, and intercept the operation instruction.

102. The virtual machine generates an uninstallation command according to the operation instruction.

Wherein, the uninstall command includes a copy uninstall command or a clear uninstall command.

It should be noted that the virtual machine can generate an unload command according to the intercepted operation instruction, and the front-end process sends the generated unload command to the back-end process through the IO ring.

103. The virtual machine sends an uninstall command to the host.

104. The host converts the virtual source address in the unload command into a physical source address of the data to be operated.

Wherein, the virtual source address is used to indicate the location of the data to be operated in the virtual disk of the virtual machine.

105. The host performs a specified operation on the data to be operated according to the virtual destination address in the offload command and the physical source address of the data to be operated.

After the host receives the uninstall command sent by the virtual machine, it parses the uninstall command, and completes the specified operation according to the operation type indicated by the uninstall command. As shown in FIG. 3, after receiving the unloading command sent by the virtual machine 700a, the backend process 73 in the host 700b sends it to the distributed shared storage system 04 or the cluster file system 01, and uses the corresponding way to complete the specified operation.

In the present invention, the virtual machine can generate an uninstall command according to the operation instruction issued by the application program, and then the virtual machine sends the uninstall command to the host; after the host receives the uninstall command sent by the virtual machine, it sends the uninstall command The virtual source address of the data to be operated is converted into the physical source address of the data to be operated. After that, the host can perform specified operations on the data to be operated according to the virtual destination address carried in the offload command and the physical source address of the data to be operated. Although the execution process of the specified operation needs to be completed by the virtual machine and the host computer together like this, because the host computer completes the specified operation according to the unloading command generated by the virtual machine, no matter whether the system currently applying the present invention is a cluster file system, distributed shared The storage system, or other systems, and the host can recognize the uninstall command. Moreover, both the host and the virtual machine are part of the physical machine. In the above execution process, the host performs corresponding operations after receiving the unloading command, without the need for a physical storage device to perform corresponding operations. Therefore, using the The implementation manner can effectively reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copy or clear instructions).

Considering that when the specified operation is copy, the number of data blocks included in the data to be operated is at least one, in order to ensure that the data to be operated can be completely copied, in an implementation of the embodiment of the present invention, it is necessary to generate The copy offload command for each data block, thereby enabling the host to perform a copy operation for each data block. Therefore, in step 102 shown in Figure 4, the virtual machine generates an uninstall command according to the operation instruction, which can be specifically implemented as step 1021; in step 103 shown in Figure 4, the virtual machine sends an uninstall command to the host, which can be specifically implemented as step 1031:

1021. The virtual machine generates a copy and unload command for each data block in at least one data block.

1031. The virtual machine sends a copy and unload command for each data block to the host.

In the present invention, the virtual machine can send the copy unload command of each data block to the host at the same time, or only send one copy unload command to the host at a time, and after the host finishes executing the copy unload command, send the copy unload command to the host. The host sends the next copy unload command until the host completes the operations indicated by all the copy unload commands.

It should be noted that the virtual machine can generate the copy and unload command corresponding to each data block according to the order in which the data blocks are generated, and the host will sequentially execute the copy operation of each data block, but in the present invention, the virtual machine is not limited to The order in which the host generates the copy unload command and the order in which the host executes the copy operation, as long as it can ensure that all the data to be operated can be copied.

In addition, considering that during the execution of the copy operation, an empty file for storing the data to be operated needs to be created in the destination virtual machine. Therefore, before performing the above operations, it is necessary to create an empty file in the disk file system of the destination virtual machine. document. After executing step 101 shown in Figure 4, the virtual machine acquires the operating instructions issued by the application, and then steps 106 and 107 need to be performed:

106. The virtual machine sends a creation instruction to the host.

Wherein, the creation instruction carries the length of the data block in the copy and unload command of each data block, and the creation instruction is used to create an empty file in the disk file system of the target virtual machine, and the target virtual machine runs on the host machine where the virtual machine runs, Or run on a host other than the host.

107. The virtual machine receives the virtual destination address fed back by the destination virtual machine.

Wherein, the virtual destination address is the virtual address of the destination data block assigned to the data to be operated in the empty file, and the length of the destination data block is greater than or equal to the length of the data block in the copy unload command of each data block.

In this way, a certain storage space can be reserved in the disk file system of the target virtual machine for the data to be operated, so that the data to be operated can be copied from the virtual source address to the target source address.

Taking the copy operation execution process shown in Figure 5 as an example, the specific implementation steps are as follows:

201. The APP in VM (English: Virtual Machine, Chinese: virtual machine) 1 issues a copy command.

Wherein, the copy instruction is specifically to copy file A from directory a to directory b of VM3; VM1 may be regarded as a source virtual machine, and VM3 may be regarded as a destination virtual machine.

It should be noted that the copy instruction can be implemented on the same virtual machine or on different virtual machines. When the copy instruction is implemented on different virtual machines, VM1 and VM3 are different virtual machines; when the copy instruction is implemented on the same virtual machine, VM1 and VM3 are the same virtual machine.

202. VM1 intercepts the copy instruction.

203. VM1 notifies VM3 to create a destination empty file.

If VM1 and VM3 are the same virtual machine, after VM1 (that is, VM3) intercepts the copy instruction, it can directly create a target empty file in the local disk file system. It should be noted that the interaction flowchart shown in FIG. 5 is an interaction flow when VM1 and VM3 belong to different virtual machines.

204. The VM3 creates a destination empty file in the local disk file system.

205. VM1 initializes the file offset f=0.

Considering that there may be multiple data blocks in the file, and in general, the copy operation can only be performed on one data block at a time, therefore, the file offset f set here is used to determine whether to continue in the subsequent execution process Conditions for copy operations of other data blocks. That is to say, until the offset f of the file A is the same as the offset of the data block of the file A to be copied, the copy operation ends.

It should be noted that the above-mentioned discrimination method is to perform the copy operation sequentially according to the generation order of each data block in the file, but in the present invention, it is not limited to the above-mentioned method, and the copy operation can also be performed out of sequence, as long as the data blocks in the file can be guaranteed All data blocks are copied.

206. Determine whether f is greater than or equal to the offset of the end of the file.

Wherein, when f is greater than or equal to the offset of the end of the file, the copy operation ends; otherwise, continue to execute step 207 .

207. The file system in VM1 acquires offset f corresponding to data block B (vlba, vlen).

Wherein, data block B is the initial data block in file A to be copied, vlba is the initial virtual source address of data block B, and vlen is the length of data block B.

208. The VM3 allocates a corresponding data block C (vlba, vlen) in the file system.

In order to ensure that data block B can be completely copied to data block C, in the present invention, the vlen of data block B and data block C can be the same, or the length of data block C is greater than the length of data block B. It should be noted that the allocation operation shown in step 208 is only allocated in the file system of the target virtual machine, that is, VM3, rather than actually distributed to the back-end real storage device for space allocation. That is to say, the operation scheme realized by the present invention may not rely on actual physical storage devices.

209. VM1 generates a copy unload command (copy, VM1.B.vlba, VM3.C.vlba, vlen).

In the present invention, the format of the copy unload command is operation type, virtual source address, virtual destination address, and length. Wherein, the operation type is copy, the virtual source address is vlba of data block B in VM1, the virtual destination address is vlba of data block C in VM3, and the length is the length of data block B, and can also be the length of data block C at the same time, namely vlen.

210. The front-end process of VM1 sends the copy unload command to the back-end process of the host through the IO ring.

211. After receiving the copy uninstall command, the backend process of the host parses the copy uninstall command to obtain the parsed copy uninstall command (copy, plba1, VM3.C.vlba, plen).

Wherein, the format of the parsed copy unload command is an operation type, a physical source address, a virtual destination address, and a length. Wherein, the operation type is copy, the physical source address is the plba of data block B, that is, the physical address of data block B, the virtual destination address is the vlba of data block C in VM3, and the length is the corresponding length of data block B in the local disk, That is, plen.

212. The host points VM3.C.vlba to plba1, and increases the reference count of the data block corresponding to plba1.

It should be noted that step 212 only provides a specific way to perform a specified operation on the data to be operated, that is, when the above implementation is used in a distributed shared storage system, how VM1, VM3 and the host implement the above process, and For when the method provided by the present invention is applied to the cluster file system, refer to step 1053 to step 1055, which will not be repeated here. In addition, in step 212, VM3.C.vlba pointing to plba1 can be regarded as establishing a mapping relationship between VM3.C.vlba and plba1.

213. The host returns the copy result to VM1.

214. In VM1, f+=B.vlen.

It should be noted that B.vlen is the length of data block B; the above operation process can only be ended when the file A to be copied is completely copied to the specified location, that is, in the file system of VM3, otherwise the above steps 206 to Step 214, until the end of file A is traversed, that is, until all data blocks in the file are copied.

Considering that when the technical solution provided by the present invention is applied to different data center structures, there will be different copy methods, in an implementation of the embodiment of the present invention, when the method of the present invention is applied to a distributed shared storage system, The copy operation can be completed by changing the mapping relationship between addresses and increasing the reference count of the data block corresponding to the physical source address of the data to be operated. Therefore, on the basis of the above implementation, in step 105, the host performs specified operations on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated, which can be specifically implemented as steps 1051 and 1052:

1051. The host updates the virtual destination address of the data to be operated to the physical source address of the data to be operated.

1052. The host adds one to the value of the reference count of the data block corresponding to the physical source address of the data to be operated according to the updated virtual destination address.

Wherein, the value of the reference count is used to indicate the number of times the data block corresponding to the physical address of the data to be operated is referenced.

Take the copy operation execution process shown in Figure 5 as an example, when the method is applied to a distributed shared storage system, the host can establish a mapping relationship between VM3.C.vlba and plba1, and increase the reference count of the data block corresponding to plba1 , thus completing the copy operation of data block B.

In the present invention, the copy operation of the data to be operated is realized by adopting the above method, only the mapping relationship between addresses is established, and the reference count of the physical source address of the data to be operated is increased. Therefore, for the virtual machine, the consumption of the computing network resources of the virtual machine by the storage operation of the virtual machine is reduced, and at the same time, an acceleration effect is also achieved by increasing the reference count.

In addition, when the method of the present invention is applied to a cluster file system, the copy operation can be completed by copying the data block corresponding to the physical source address of the data to be operated, and then changing the physical source address in the mapping relationship to the physical destination address . Therefore, on the basis of the above implementation, in step 105, the host performs specified operations on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated, which can also be specifically implemented as steps 1053 to 1055:

1053. The host establishes a mapping relationship between the virtual destination address of the data to be operated and the physical source address of the data to be operated.

1054. The host copies the data block corresponding to the physical source address of the data to be operated to the physical destination address of the data to be operated.

1055. The host replaces the physical source address of the data to be operated in the mapping relationship with the physical destination address of the data to be operated.

Take the copy operation execution process shown in Figure 5 as an example, when the method is applied to the cluster file system, the host can copy the data block of plba1 to plba2, and establish the mapping relationship between VM3.C.vlba and plba2, so that Complete the copy operation of data block B.

In the present invention, the above method is used to realize the copy operation of the data to be operated. It is necessary to first copy the data block from the physical source address to the physical destination address, and then establish the mapping relationship between the virtual source address and the physical destination address. For the virtual machine, the consumption of virtual machine computing network resources by the storage operation of the virtual machine is reduced, but compared with the implementation solutions shown in steps 1051 and 1052, effective copy acceleration cannot be achieved.

Since in the data structure center, the storage method of the data to be operated may be block storage or object storage, and before the copy operation, it is necessary to complete the conversion operation of the virtual source address of the data to be operated according to different storage methods, in the present invention In an implementation of the embodiment, specific solutions for converting the virtual source address of the data to be operated are provided for different storage methods. Therefore, on the basis of the above implementation, the host in step 104 converts the virtual source address in the unload command into the physical source address of the data to be operated, which can be specifically implemented as step 1041 or step 1042:

1041. When the storage mode of the data to be operated is block storage, the host converts the virtual source address of the data to be operated into the physical source address of the data to be operated.

1042. When the storage mode of the data to be operated is object storage, the host converts the virtual source address of the data to be operated into an object identifier.

Wherein, the object identifier is an identifier of a data block corresponding to the data to be operated.

It should be noted that when the backend process of the host parses the unload command, vlba and vlen need to be converted, and the backend process can specifically convert the virtual source address vlba of the data to be operated in block storage into a physical source address plba, or convert the virtual source address vlba of the data to be operated in object storage into an object identifier oid.

In the present invention, in the process of unloading command parsing by the backend process of the host, the virtual source address of the data to be operated is converted, and then handed over to the distributed shared storage system or the cluster file system for processing.

When the operation instruction issued by the application program in the virtual machine indicates that the operation is clearing, in an implementation manner of the embodiment of the present invention, an appropriate zeroing method may be selected according to the simulation situation of the virtual disk. Therefore, on the basis of the implementation shown in FIG. 4 , in step 105, the host performs specified operations on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated. One item in step 1059:

1056. When the virtual disk is simulated by the data to be operated, the host clears each bit in the bitmap of the physical source address of the data to be operated to zero.

1057. When the virtual disk is simulated by the distributed shared storage system, the host releases the object identifier of the data to be operated, and releases the data block corresponding to the data to be operated.

1058. When the virtual disk is a virtual disk simulated by the cluster file system, the host releases the data block of the data to be operated.

1059. When there is no mapping relationship between the virtual source address of the data to be operated and the physical source address of the data to be operated, the host writes zero to the data block corresponding to the physical source address of the data to be operated.

For example: if the virtual disk is simulated by a vhd (English: Microsoft Virtual Hard Disk format, Chinese: virtual disk format) file, clear the bitmap corresponding to the plba address in the vhd file to 0; if the virtual disk is provided by Server San , then release the object identifier of the corresponding block in Server San, and release the corresponding data block; if the virtual disk is provided by the cluster file system, release the data block of the file corresponding to plba in the cluster file system; if there is no metadata, Then write 0 to the data block corresponding to plba, and this method can only reduce the consumption of virtual machine computing network resources by virtual machine storage operations, but cannot achieve the acceleration effect like the above three implementation methods.

Taking the execution process of the clearing operation as shown in Figure 6 as an example, the specific implementation steps are as follows:

301. The APP in VM1 issues a clearing instruction.

Wherein, the clearing command may specifically be an operation of writing 0 to the f offset in file A.

302. VM1 intercepts the clear instruction, and converts the clear instruction into a clear unload command (clear, vlba, vlen).

In the present invention, the format of the clear unload command is operation type, virtual source address, and length. Wherein, the operation type is clear, the virtual source address is vlba of data block B in VM1, and the length is the length of data block B, namely vlen.

303. The front-end process of VM1 sends a clearing and unloading command to the back-end process via the IO ring.

304. After receiving the clearing and uninstalling command, the backend process of the host parses the clearing and uninstalling command, and obtains the parsed clearing and uninstalling command (clearing, plba1, plen).

Wherein, the format of the parsed clearing and unloading command is an operation type, a physical source address, and a length. Wherein, the operation type is clear, the physical source address is the plba of the data block B, and the length is the corresponding length of the data block B in the local disk, that is, plen.

305. The host performs a clearing operation on metadata corresponding to plba and plen.

Wherein, the metadata is used to represent the mapping relationship between the virtual source address and the physical source address. It should be noted that, in the present invention, the specific clearing operation process includes, but is not limited to, different clearing methods for different situations shown in steps 1056 to 1059 above, which will not be repeated here.

306. The host returns the zeroing result to VM1.

In the present invention, when the virtual machine acquires that the operation instruction issued by the application program is an instruction indicating a clearing operation, the host can select an appropriate clearing method according to the simulation situation of the virtual disk, and then perform a clearing operation on the data to be operated .

The foregoing mainly introduces the solution provided by the embodiment of the present invention from the perspective of interaction between the virtual machine and the host. It can be understood that, in order to realize the above functions, the present invention also provides a command unloading device 80, as shown in FIG. 7, the device 80 includes corresponding software modules for performing various functions. Those skilled in the art should easily realize that the present invention can be realized in the form of hardware or a combination of hardware and computer software in combination with the units and algorithm steps of each example described in the embodiments disclosed herein. Whether a certain function is executed by hardware or computer software drives hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.

In the embodiment of the present invention, the device 80 may be divided into functional modules according to the above method examples. For example, each functional module may be divided corresponding to each function, or two or more functions may be integrated into one processing module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software function modules. It should be noted that the division of modules in the embodiment of the present invention is schematic, and is only a logical function division, and there may be another division manner in actual implementation.

In the case of dividing each functional module corresponding to each function, FIG. 7 shows a possible structural diagram of an apparatus 80 involved in the above embodiment. The apparatus 80 includes: a virtual machine 81 and a host 82 . The virtual machine 81 is used to intercept the operation instruction issued by the application program, and the operation instruction is used to instruct to perform a specified operation on the data to be operated; the virtual machine 81 is also used to generate an uninstall command according to the operation instruction, and send the uninstall command to the host 82, The unloading command includes a copy unloading command or a clear unloading command; the host 82 is used to convert the virtual source address in the unloading command into the physical source address of the data to be operated, and the virtual source address is used to represent the virtual source address of the data to be operated in the virtual machine 81. The location in the disk; the host 82 is also used to perform specified operations on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated.

In the present invention, the virtual machine 81 is not only used to execute the process 101 to the process 103 in FIG. 4, but also used to execute all the processes performed by VM1 in FIG. 5 and FIG. 6; the host computer 82 is not only used to execute the process in FIG. 4 104 and process 105 are also used to execute all processes executed by the host in FIG. 5 and FIG. 6 . Wherein, all relevant content of each step involved in the above-mentioned method embodiment can be referred to the function description of the corresponding function module, and will not be repeated here. In addition, the host 82 and the virtual machine 81 in the device 80 can also be used to execute other processes of the techniques described herein. In the present invention, the device 80 may further include a communication module 83 for communicating with other external devices, and a storage module 84 for storing program codes and data. Wherein, the communication module 83 may specifically be a transceiver, a transceiver circuit, or a communication interface, etc., the storage module 84 may specifically be a memory, and the host 82 and the virtual machine 81 may be integrated in the storage module 84, or the storage module 84 may be There are at least two, and they are set inside the host machine 82 and the virtual machine 81 respectively. In addition, the device 80 may further include a processing module 85 . Wherein, the processing module 85 can be a processor or a controller, such as a central processing unit, a general-purpose processor, a digital signal processor (English: Digital Signal Processor, referred to as: DSP), an application-specific integrated circuit (English: Application-Specific Integrated Circuit, referred to as: ASIC), field programmable gate array (English: Field Programmable Gate Array, referred to as: FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute the various illustrative logical blocks, modules and circuits described in connection with the present disclosure. The processor may also be a combination of computing functions, for example, a combination of one or more microprocessors, a combination of DSP and a microprocessor, and so on.

Compared with the command unloading device provided by the present invention, which uses a physical storage device to perform specified operations in the prior art, in the present invention, the virtual machine can generate an unloading command according to the operation instructions issued by the application program, and then the virtual machine Send the unloading command to the host; after the host receives the unloading command sent by the virtual machine, it converts the virtual source address carried by the unloading command into the physical source address of the data to be operated, and then the host can The address and the physical source address of the data to be operated are used to perform the specified operation on the data to be operated. Although the execution process of the specified operation needs to be completed by the virtual machine and the host computer together like this, because the host computer completes the specified operation according to the unloading command generated by the virtual machine, no matter whether the system currently applying the present invention is a cluster file system, distributed shared The storage system, or other systems, and the host can recognize the uninstall command. Moreover, both the host and the virtual machine are part of the physical machine. In the above execution process, the host performs corresponding operations after receiving the unloading command, without the need for a physical storage device to perform corresponding operations. Therefore, using the The implementation manner can effectively reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copy or clear instructions).

When the communication module 83 is an input device and an output device, and the virtual machine 81 and the host 82 are integrated on the same storage module 84, the storage module can be specifically a processor, and when the storage module 84 is a memory, the physical The machine may be the physical machine 90 shown in FIG. 8 .

Referring to FIG. 8 , the physical machine 90 includes: a processor 91 , an input device 92 , an output device 93 , a memory 94 and a bus 95 . Wherein, the processor 91, the input device 92, the output device 93, and the memory 94 are connected to each other through a bus 95, and the memory 94 may specifically include a host computer 941 and a virtual machine 942; the bus 95 may be a peripheral component interconnection standard (English: Peripheral Component Interconnect (PCI for short) bus or Extended Industry Standard Architecture (English: Extended Industry Standard Architecture, EISA for short) bus, etc. The bus can be divided into address bus, data bus, control bus and so on. For ease of representation, only one thick line is used in FIG. 8 , but it does not mean that there is only one bus or one type of bus.

The steps of the methods or algorithms described in conjunction with the disclosure of the present invention may be implemented in the form of hardware, or may be implemented in the form of a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (English: Random Access Memory, abbreviated as RAM), flash memory, read-only memory (English: Read Only Memory, abbreviated: ROM), and Erasable Programmable Read-Only Memory (English: Erasable Programmable ROM, referred to as: EPROM), Electrically Erasable Programmable Read-Only Memory (English: Electrically EPROM, referred to as: EEPROM), registers, hard disk, mobile hard disk, CD-ROM (abbreviated: CD-ROM) or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. Of course, the storage medium may also be a component of the processor. The processor and storage medium can be located in the ASIC. In addition, the ASIC may be located in the core network interface device. Certainly, the processor and the storage medium may also exist in the core network interface device as discrete components.

It should be noted that, whether the input device 92 and the output device 93 need to be set can be determined according to whether the physical machine 90 needs to communicate with other devices. The host 941 shown in FIG. 8 acts as a management layer to complete the management and allocation of hardware resources; present a virtual hardware platform for the virtual machine 942; and realize the scheduling and isolation of virtual machines. Wherein, the host 941 may be a virtual machine monitor (VMM for short); in addition, sometimes the VMM cooperates with a privileged virtual machine, and the combination of the two forms the host 941 . Wherein, the virtual hardware platform provides various hardware resources for each virtual machine running on it, such as virtual processor, memory, virtual disk, virtual network card and so on. Wherein, the virtual disk may correspond to a file or a logical block device of the host 941 . The virtual machine 942 runs on the virtual hardware platform prepared for it by the host 941, and one or more virtual machines such as 942 run on the host 941. Among them, the virtual machine 942 can simulate one or more virtual computers on a physical machine through the virtual machine software, and these virtual machines work like real computers, and the operating system and applications can be installed on the virtual machine 942. program, the virtual machine 942 can also access network resources. For applications running in the virtual machine 942, the virtual machine is like working on a real computer.

In the present invention, the virtual machine 942 is used to intercept the operation instruction issued by the application program, and the operation instruction is used to instruct to perform a specified operation on the data to be operated; the virtual machine 942 is also used to generate an uninstall command according to the operation instruction, and send the command to the host 941 sends an unload command, and the unload command includes a copy unload command or a clear unload command; the host 941 is used to convert the virtual source address in the unload command into the physical source address of the data to be operated, and the virtual source address is used to indicate that the data to be operated is in The location in the virtual disk of the virtual machine 942; the host 941 is also used to perform a specified operation on the data to be operated according to the virtual destination address in the unload command and the physical source address of the data to be operated.

In addition, the virtual machine 942 can also be used to perform the functions implemented by the virtual machine in Figure 5 and Figure 6, and the host 941 can also be used to perform the functions implemented by the host in Figure 5 and Figure 6, for details, please refer to the above method implementation Each step in the example will not be repeated here.

Compared with the physical machine provided by the present invention, which uses a physical storage device to perform specified operations in the prior art, in the present invention, the virtual machine can generate an uninstall command according to the operation instructions issued by the application program, and then the virtual machine will The unloading command is sent to the host; after the host receives the unloading command sent by the virtual machine, it converts the virtual source address carried by the unloading command into the physical source address of the data to be operated, and then the host can use the virtual destination address carried by the unloading command and the physical source address of the above data to be operated, perform the specified operation on the data to be operated. Although the execution process of the specified operation needs to be completed by the virtual machine and the host computer together like this, because the host computer completes the specified operation according to the unloading command generated by the virtual machine, no matter whether the system currently applying the present invention is a cluster file system, distributed shared The storage system, or other systems, and the host can recognize the uninstall command. Moreover, both the host and the virtual machine are part of the physical machine. In the above execution process, the host performs corresponding operations after receiving the unloading command, without the need for a physical storage device to perform corresponding operations. Therefore, using the The implementation manner can effectively reduce the waste of hardware resources of the physical machine when processing operation instructions (such as copy or clear instructions).

Those skilled in the art should be aware that, in the above one or more examples, the functions described in the present invention may be implemented by hardware, software, firmware or any combination thereof. When implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a general purpose or special purpose computer.

The specific embodiments described above have further described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above descriptions are only specific embodiments of the present invention and are not intended to limit the scope of the present invention. Protection scope, any modification, equivalent replacement, improvement, etc. made on the basis of the technical solution of the present invention shall be included in the protection scope of the present invention.

Claims (21)

1. A method for command offload, the method being used for a physical machine running a Host on which a virtual machine runs, the method comprising:

the virtual machine intercepts an operation instruction issued by an application program, and the operation instruction is used for indicating to-be-operated data to execute specified operation;

the virtual machine generates an unloading command according to the operation instruction, and sends the unloading command to the host, wherein the unloading command comprises a copy unloading command or a zero clearing unloading command;

the host converts a virtual source address in the unloading command into a physical source address of data to be operated, wherein the virtual source address is used for representing the position of the data to be operated in a virtual disk of the virtual machine;

and the host executes the specified operation on the data to be operated according to the virtual destination address in the unloading command and the physical source address of the data to be operated.

2. The method according to claim 1, wherein when the designated operation is a copy operation, the data to be operated on includes at least one data block, and the virtual machine generates an offload command according to the operation instruction and sends the offload command to the host, including:

the virtual machine generates a copy offload command for each of the at least one data block and sends the copy offload command for each data block to the host.

3. The method of claim 2, wherein after the virtual machine intercepts an operation instruction issued by an application program, the method further comprises:

the virtual machine sends a creation instruction to the host, wherein the creation instruction carries the length of the data block in the copy unloading command of each data block, the creation instruction is used for creating an empty file in a disk file system of the target virtual machine, and the target virtual machine runs on the host where the virtual machine runs or runs on a host except the host;

and the virtual machine receives a virtual destination address fed back by the destination virtual machine, wherein the virtual destination address is a virtual address of a destination data block allocated to the data to be operated in the empty file, and the length of the destination data block is greater than or equal to the length of a data block in the copy unloading command of each data block.

4. The method according to claim 1, wherein the method is applied to a distributed shared storage system, and the host performs the specified operation on the data to be operated according to a virtual destination address in the offload command and a physical source address of the data to be operated, and includes:

the host updates the virtual destination address of the data to be operated into the physical source address of the data to be operated;

and the host adds one to the numerical value of the reference count of the data block corresponding to the physical source address of the data to be operated according to the updated virtual destination address, wherein the value of the reference count is used for representing the number of times that the data block corresponding to the physical address of the data to be operated is referred.

5. The method according to claim 1, wherein the method is applied to a cluster file system, and the host performs the specified operation on the data to be operated according to a virtual destination address in the offload command and a physical source address of the data to be operated, and includes:

the host establishes a mapping relation between the virtual destination address of the data to be operated and the physical source address of the data to be operated;

the host copies the data block corresponding to the physical source address of the data to be operated to the physical destination address of the data to be operated;

and the host replaces the physical source address of the data to be operated in the mapping relation with the physical destination address of the data to be operated.

6. The method of claim 4 or 5, wherein the host converting the virtual source address in the offload command to a physical source address of the data to be operated on comprises:

under the condition that the storage mode of the data to be operated is block storage, the host converts a virtual source address of the data to be operated into a physical source address of the data to be operated;

or,

and under the condition that the storage mode of the data to be operated is object storage, the host converts the virtual source address of the data to be operated into an object identifier, wherein the object identifier is an identifier of a data block corresponding to the data to be operated.

7. The method according to claim 1, wherein when the offload command is the clear offload command, the host performs a specified operation on the data to be operated according to a virtual destination address in the offload command and a physical source address of the data to be operated, and the method includes:

when the virtual disk is a virtual disk simulated by the data to be operated, the host machine clears each bit in the bitmap of the physical source address of the data to be operated;

or,

when the virtual disk is a virtual disk simulated by a distributed shared storage system, the host releases the object identifier of the data to be operated and releases the data block corresponding to the data to be operated;

or,

when the virtual disk is a virtual disk simulated by a cluster file system, the host releases the data block of the data to be operated;

or,

and when the mapping relation between the virtual source address of the data to be operated and the physical source address of the data to be operated does not exist, the host writes zero in the data block corresponding to the physical source address of the data to be operated.

8. A command unloading device comprises a Host and a virtual machine, and is characterized in that the virtual machine is used for intercepting an operation instruction sent by an application program, and the operation instruction is used for indicating to execute specified operation on data to be operated;

the virtual machine is further configured to generate an unload command according to the operation instruction, and send the unload command to the host, where the unload command includes a copy unload command or a clear unload command;

the host is used for converting a virtual source address in the unloading command into a physical source address of data to be operated, wherein the virtual source address is used for representing the position of the data to be operated in a virtual disk of the virtual machine;

the host is further configured to execute the specified operation on the data to be operated according to the virtual destination address in the offload command and the physical source address of the data to be operated.

9. The apparatus according to claim 8, wherein when the designated operation is a copy operation, the data to be operated on includes at least one data block, and in terms of generating an offload command according to the operation instruction and sending the offload command to the host, the virtual machine is specifically configured to:

and generating a copy offload command for each of the at least one data block, and sending the copy offload command for each data block to the host.

10. The apparatus according to claim 9, wherein the virtual machine is further configured to send a create instruction to the host, where the create instruction carries a length of a data block in the copy offload command for each data block, and the create instruction is used to create an empty file in a disk file system of the destination virtual machine, where the destination virtual machine runs on a host on which the virtual machine runs or runs on a host other than the host;

the virtual machine is further configured to receive a virtual destination address fed back by the destination virtual machine, where the virtual destination address is a virtual address of a destination data block allocated to the data to be operated in the empty file, and the length of the destination data block is greater than or equal to the length of a data block in the copy offload command of each data block.

11. The apparatus according to claim 8, wherein the apparatus is applied to a distributed shared storage system, and in terms of performing the specified operation on the data to be operated according to a virtual destination address in the offload command and a physical source address of the data to be operated, the host is specifically configured to:

updating the virtual destination address of the data to be operated into the physical source address of the data to be operated;

and adding one to the numerical value of the reference count of the data block corresponding to the physical source address of the data to be operated according to the updated virtual destination address, wherein the value of the reference count is used for representing the number of times that the data block corresponding to the physical address of the data to be operated is referred.

12. The apparatus according to claim 8, wherein the apparatus is applied to a cluster file system, and in terms of performing the specified operation on the data to be operated according to the virtual destination address in the offload command and the physical source address of the data to be operated, the host is specifically configured to:

establishing a mapping relation between a virtual destination address of the data to be operated and a physical source address of the data to be operated;

copying a data block corresponding to the physical source address of the data to be operated to the physical destination address of the data to be operated;

and replacing the physical source address of the data to be operated in the mapping relation with the physical destination address of the data to be operated.

13. The apparatus according to claim 11 or 12, wherein in terms of converting a virtual source address in the offload command to a physical source address of the data to be operated on, the host is specifically configured to:

under the condition that the storage mode of the data to be operated is block storage, converting a virtual source address of the data to be operated into a physical source address of the data to be operated;

or,

and under the condition that the storage mode of the data to be operated is object storage, converting the virtual source address of the data to be operated into an object identifier, wherein the object identifier is an identifier of a data block corresponding to the data to be operated.

14. The apparatus according to claim 8, wherein when the offload command is the clear offload command, in an aspect that a specified operation is performed on the data to be operated according to a virtual destination address in the offload command and a physical source address of the data to be operated, the host is specifically configured to:

when the virtual disk is a virtual disk simulated by the data to be operated, clearing each bit in a bitmap of a physical source address of the data to be operated;

or,

when the virtual disk is a virtual disk simulated by a distributed shared storage system, releasing the object identifier of the data to be operated, and releasing the data block corresponding to the data to be operated;

or,

when the virtual disk is a virtual disk simulated by a cluster file system, releasing the data block of the data to be operated;

or,

and writing zero into a data block corresponding to the physical source address of the data to be operated when the mapping relation between the virtual source address of the data to be operated and the physical source address of the data to be operated does not exist.

15. A physical machine comprises a hardware layer, a Host computer Host running on the hardware layer and a virtual machine, and is characterized in that the virtual machine is used for intercepting an operation instruction sent by an application program, and the operation instruction is used for indicating to execute specified operation on data to be operated;

the virtual machine is further configured to generate an unload command according to the operation instruction, and send the unload command to the host, where the unload command includes a copy unload command or a clear unload command;

the host is used for converting a virtual source address in the unloading command into a physical source address of data to be operated, wherein the virtual source address is used for representing the position of the data to be operated in a virtual disk of the virtual machine;

the host is further configured to execute the specified operation on the data to be operated according to the virtual destination address in the offload command and the physical source address of the data to be operated.

16. The physical machine according to claim 15, wherein when the designated operation is a copy operation, the data to be operated includes at least one data block, and in terms of generating an offload command according to the operation instruction and sending the offload command to the host, the virtual machine is specifically configured to:

and generating a copy offload command for each of the at least one data block, and sending the copy offload command for each data block to the host.

17. The physical machine according to claim 16, wherein the virtual machine is further configured to send a create instruction to the host, where the create instruction carries a length of a data block in the copy offload command for each data block, and the create instruction is used to create an empty file in a disk file system of the destination virtual machine, where the destination virtual machine runs on a host on which the virtual machine runs or runs on a host other than the host;

the virtual machine is further configured to receive a virtual destination address fed back by the destination virtual machine, where the virtual destination address is a virtual address of a destination data block allocated to the data to be operated in the empty file, and the length of the destination data block is greater than or equal to the length of a data block in the copy offload command of each data block.

18. The physical machine according to claim 15, wherein the physical machine is applied to a distributed shared storage system, and in terms of performing the specified operation on the data to be operated according to the virtual destination address in the offload command and the physical source address of the data to be operated, the host is specifically configured to:

updating the virtual destination address of the data to be operated into the physical source address of the data to be operated;

and adding one to the numerical value of the reference count of the data block corresponding to the physical source address of the data to be operated according to the updated virtual destination address, wherein the value of the reference count is used for representing the number of times that the data block corresponding to the physical address of the data to be operated is referred.

19. The physical machine according to claim 15, wherein the physical machine is applied to a cluster file system, and in terms of performing the specified operation on the data to be operated according to the virtual destination address in the offload command and the physical source address of the data to be operated, the host is specifically configured to:

establishing a mapping relation between a virtual destination address of the data to be operated and a physical source address of the data to be operated;

copying a data block corresponding to the physical source address of the data to be operated to the physical destination address of the data to be operated;

and replacing the physical source address of the data to be operated in the mapping relation with the physical destination address of the data to be operated.

20. The physical machine according to claim 18 or 19, wherein in terms of converting a virtual source address in the offload command to a physical source address of the data to be operated on, the host is specifically configured to:

under the condition that the storage mode of the data to be operated is block storage, converting a virtual source address of the data to be operated into a physical source address of the data to be operated;

or,

and under the condition that the storage mode of the data to be operated is object storage, converting the virtual source address of the data to be operated into an object identifier, wherein the object identifier is an identifier of a data block corresponding to the data to be operated.

21. The physical machine according to claim 15, wherein when the offload command is the clear offload command, in an aspect that a specified operation is performed on the data to be operated according to a virtual destination address in the offload command and a physical source address of the data to be operated, the host is specifically configured to:

when the virtual disk is a virtual disk simulated by the data to be operated, clearing each bit in a bitmap of a physical source address of the data to be operated;

or,

when the virtual disk is a virtual disk simulated by a distributed shared storage system, releasing the object identifier of the data to be operated, and releasing the data block corresponding to the data to be operated;

or,

when the virtual disk is a virtual disk simulated by a cluster file system, releasing the data block of the data to be operated;

or,

and writing zero into a data block corresponding to the physical source address of the data to be operated when the mapping relation between the virtual source address of the data to be operated and the physical source address of the data to be operated does not exist.