CN103838634B - Method and system for dispatching number of virtual machines based on internal storage resource supplying - Google Patents

Abstract

The present invention provides a method and system for scheduling the number of virtual machines based on memory resource supply, including: determining the quantitative relationship between them according to the arrival rate of virtual machine creation requests and the service rate of virtual machine creation by the system, and determining the number of virtual machines that the system allows to create The optimal number of machines; modify the partner algorithm parameters of the host operating system memory resource scheduling subsystem, and divide the available memory of the system into k blocks; the host operating system memory resource scheduling subsystem schedules a block of k blocks of memory to the current virtual machine The guest operating system memory resource scheduling subsystem; the host operating system memory resource scheduling subsystem reclaims the memory of the virtual machine guest operating system memory resource scheduling subsystem. The present invention solves the method and system for scheduling memory resources between the memory resource scheduling subsystem of the host operating system and the memory resource scheduling subsystem of the guest operating system on the physical server of the data center, and effectively guides the number of scheduling virtual machines on the physical server .

Description

A method and system for scheduling the number of virtual machines based on memory resource supply

technical field

The invention relates to the field of operating systems, in particular to a method and system for scheduling the number of virtual machines based on memory resource supply.

Background technique

Virtualization technology is the cornerstone of cloud computing. By providing virtualized computing resources and storage resources, cloud service providers enable tenants to create thousands of virtual machines on numerous physical servers, and then create various types of virtual machines on these virtual machines. Operating system, so that different services of different operating systems meet the needs of various types of tenants. However, creating thousands of virtual machines on many physical servers requires solving many problems. From the perspective of memory resource scheduling, the current problems mainly include:

1) When the memory capacity of a single physical server is given, how to quantitatively estimate the number of virtual machines created on the server;

2) On the basis of 1), how to model the time from the time when the user requests to create a virtual machine to the time when the virtual machine is created after obtaining memory resources, that is, whether the user request rate and the service rate conform to a certain probability distribution;

3) On the basis of 2), how to allocate memory resources from the memory resource scheduling subsystem of the host operating system to the memory resource scheduling subsystem of the guest operating system under the host model virtual machine architecture, and how the memory resource scheduling subsystem of the guest operating system Release memory resources to the host operating system memory resource scheduling subsystem;

Of course, in a cloud data center, how to centrally solve the above three problems on many physical servers is also very tricky and complicated.

Contents of the invention

The purpose of the present invention is to provide a method and system for scheduling the number of virtual machines based on memory resource supply, so as to solve the problems existing in memory resource supply and scheduling when virtual machines are created on physical servers.

In order to solve the above technical problems, the present invention provides a system for scheduling the number of virtual machines based on memory resource supply, including: a host operating system memory resource scheduling subsystem and a guest operating system memory resource scheduling subsystem;

The host operating system memory resource scheduling subsystem includes a memory allocation module and a memory recovery module, wherein the memory allocation module is used to allocate memory blocks to the guest operating system memory resources by the host operating system memory resource scheduling subsystem The scheduling subsystem, the memory recovery module is used for the host operating system memory resource scheduling subsystem to recover the memory of the guest operating system memory resource scheduling subsystem after downtime;

The guest operating system memory resource scheduling subsystem includes a memory detection module and a memory release module, wherein the memory detection module is used for the guest operating system memory resource scheduling subsystem to detect the For the memory block assigned by the memory resource scheduling subsystem, the memory release module is used for the memory resource scheduling subsystem of the guest operating system to return the occupied memory to the host operating system memory resource scheduling subsystem when it is down.

At the same time, the present invention also provides a method for scheduling the number of virtual machines based on the supply of memory resources, using the system for scheduling the number of virtual machines based on the supply of memory resources, including:

Obtain the arrival rate of virtual machine creation requests; obtain the service rate of virtual machine creation by the system; determine the quantitative relationship between them according to the arrival rate of virtual machine creation requests and the service rate of virtual machine creation by the system, and determine the maximum number of virtual machines that the system allows to create Excellent quantity;

Modify the partner algorithm parameters of the memory resource scheduling subsystem of the host operating system, and divide the available memory of the system into k blocks;

The host operating system memory resource scheduling subsystem schedules a certain block of k blocks of memory to the guest operating system memory resource scheduling subsystem of the current virtual machine;

The host operating system memory resource scheduling subsystem reclaims the memory of the virtual machine guest operating system memory resource scheduling subsystem.

Further, in the method for scheduling the number of virtual machines based on memory resource supply, the arrival rate of obtaining virtual machine creation requests is obtained by the following formula:

When n->∞,

in

P _m (t) represents the probability that there are m virtual machine requests to be created at time interval t, λ represents the arrival rate of new virtual machine creation requests when virtual machine requests are created, and n represents the time cut time within time interval t Divide into n equal parts for infinitesimal reduction.

Further, in the method for scheduling the number of virtual machines based on the supply of memory resources, the service rate of virtual machines created by the acquisition system is obtained by the following formula:

P{X>t}=e ^-μt wherein, μ represents the virtual machine service rate created by the system when the virtual machine requests to be created.

Further, in the method for scheduling the number of virtual machines based on memory resource supply, the quantitative relationship between them is determined according to the arrival rate of virtual machine creation requests and the service rate of virtual machine creation by the system through the following formula:

k: The memory block provided by the memory resource scheduling subsystem of the host operating system

m: the number of virtual machines requested to be created

p _m : the probability of requesting to create m virtual machines

Further, in the method for scheduling the number of virtual machines based on memory resource supply, the k blocks of memory refer to continuous or discontinuous memory blocks, and continuous within a block.

Further, in the method for scheduling the number of virtual machines based on memory resource supply, the step of modifying the partner algorithm parameters of the memory resource scheduling subsystem of the host operating system and dividing the available memory of the system into k blocks includes:

Allocating k blocks of memory in the partner system of the current host operating system memory resource scheduling subsystem for the guest operating system memory resource scheduling subsystem of the virtual machine, and locking the k blocks of memory by using a lock mechanism;

An item is added in the physical memory description area of the guest operating system memory resource scheduling subsystem of each virtual machine to describe the memory block information currently used for creating the virtual machine.

Further, in the method for scheduling the number of virtual machines based on memory resource supply, the host operating system memory resource scheduling subsystem schedules a certain block of k blocks of memory to the guest operating system memory resource scheduling subsystem of the current virtual machine Steps include:

The memory resource scheduling subsystem of the host operating system sequentially selects a block of k blocks of memory, and identifies the starting page frame number of the physical page table of this block of memory;

The guest operating system memory resource scheduling subsystem of the virtual machine loads a block of memory, reads the information of the memory block, and simultaneously updates the number of physical page frames and the starting point of the physical page table of the memory management module of the current guest operating system memory resource scheduling subsystem. Data structures such as page frame number, page table mapping relationship, memory area, memory node, etc., after the update is completed, send a completion message to the memory resource scheduling subsystem of the host operating system;

After receiving the completion message, the memory resource scheduling subsystem of the host operating system clears the block of memory from the memory management pool of the memory resource scheduling subsystem of the host operating system.

Further, in the method for scheduling the number of virtual machines based on memory resource supply, the step of the host operating system memory resource scheduling subsystem reclaiming the memory of the virtual machine guest operating system memory resource scheduling subsystem includes:

After the memory resource scheduling subsystem of the guest operating system of the virtual machine receives the downtime scheduling, it packs the number of physical page frames it has obtained and the starting page frame number of the physical page table into a message;

The guest operating system memory resource scheduling subsystem of the virtual machine sends an interprocessor interrupt to the host operating system memory resource scheduling subsystem, and delivers the message to the host operating system memory resource scheduling subsystem;

After the host operating system memory resource scheduling subsystem receives the interrupt between the processors, it executes a predefined interrupt processing function between the processors. System message, and according to this message, the memory of the virtual machine guest operating system memory resource scheduling subsystem that needs to be shut down is reintroduced into the memory management pool of the host operating system memory resource scheduling subsystem.

A method and system for scheduling the number of virtual machines based on memory resource supply provided by the present invention has the following beneficial effects: it solves the problem between the memory resource scheduling subsystem of the host operating system and the memory resource scheduling subsystem of the guest operating system on the physical server of the data center. The method and system for scheduling memory resources in between, and at the same time established the relationship between the request rate of virtual machine creation and the service rate of the system, effectively guiding the number of virtual machines to be scheduled on the physical server, and reasonably reducing the performance of the physical server consumption.

Description of drawings

FIG. 1 is a schematic diagram of a system for scheduling the number of virtual machines based on memory resource supply in a preferred embodiment of the present invention;

2 is a schematic diagram of a method for scheduling the number of virtual machines based on memory resource supply in a preferred embodiment of the present invention;

Fig. 3 is a schematic diagram of an arrival rate distribution curve of a virtual machine creation request according to a preferred embodiment of the present invention.

detailed description

The method and system for scheduling the number of virtual machines based on the supply of memory resources proposed by the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Advantages and features of the present invention will be apparent from the following description and claims. It should be noted that all the drawings are in a very simplified form and use imprecise scales, and are only used to facilitate and clearly assist the purpose of illustrating the embodiments of the present invention.

Please refer to FIG. 1 , which is a schematic diagram of a system for scheduling the number of virtual machines based on memory resource supply in a preferred embodiment of the present invention. This embodiment uses ubuntu-12.04.3LTS as the host operating system memory resource scheduling subsystem, and ubuntu-12.04.3LTS implements an application system as shown in Figure 1 for the guest operating system memory resource scheduling subsystem. The kernel version of this application system The number is Linux-3.8.0, and the CPU model is Intel(R) Xeon(R) E56202.4GHz. The system mainly adds a memory allocation module and a memory recovery module in the memory management of the host operating system memory resource scheduling subsystem, wherein the memory allocation module is used for the host operating system memory resource scheduling subsystem to allocate memory blocks to client operations The system memory resource scheduling subsystem, the memory recovery module is used for the host operating system memory resource scheduling subsystem to reclaim the memory of the guest operating system memory resource scheduling subsystem after downtime; the system is installed in the guest operating system memory resource scheduling subsystem In the memory management, a memory detection module and a memory release module are mainly added, wherein the memory detection module is used for the memory resource dispatching subsystem of the guest operating system to detect the memory allocated from the memory resource dispatching subsystem of the host operating system during the startup process. Block, the memory release module is used for the memory resource scheduling subsystem of the guest operating system to return the occupied memory to the memory resource scheduling subsystem of the host operating system at the moment of downtime.

Please refer to FIG. 2 , which is a schematic diagram of a method for scheduling the number of virtual machines based on memory resource supply according to a preferred embodiment of the present invention. As shown in Figure 2, the present invention provides a method for scheduling the number of virtual machines based on memory resource supply, including the following steps:

A) Obtain the arrival rate of the virtual machine creation request, obtain the service rate of the virtual machine created by the system, determine the quantitative relationship between them according to the arrival rate of the virtual machine creation request and the service rate of the virtual machine created by the system, and determine the allowed creation of the system The optimal number k of virtual machines;

A1) Obtain the arrival rate of virtual machine creation requests;

S: A system consisting of a single physical server and multiple virtual machines created on it;

N(t): At a certain time t, the number of virtual machines requested by the user to be created, and its value is also the queue length of the virtual machine requested by the user at this moment;

P _m (t): the probability that there are m virtual machine requests to create at time interval t;

λ _m : When there are m virtual machine requests in S, the arrival rate of new virtual machine creation requests;

μ _m : When there are m virtual machine requests in S, the service rate of creating virtual machines in S.

In the time interval t, divide the time into n equal parts for infinitesimal minimization, set Δt=t/n, if the arrival rate is λ within Δt time, then a virtual machine creation request can be identified as λ* when it arrives Δt, the probability that no virtual machine request is created is 1-λ*Δt. Since the creation time of the user request is independent of each other, that is, the user's request to create a virtual machine arrives independently and randomly, so within the time interval t of n Δt, there is a request to create a virtual machine or no request to create a virtual machine can be considered as n independent experiments, so the probability of m virtual machine creation requests in n Δt is:

When n->∞,

Formula Ⅰ

Among them, P _m (t) represents the probability that there are m virtual machine requests to be created at time interval t, λ represents the arrival rate of new virtual machine creation requests when virtual machine requests are created, and n represents the number of virtual machine creation requests within time interval t. Time is divided into n equal parts for infinitesimal minimization.

Formula I shows that within the time interval t, the process of user request to create a virtual machine obeys the Poisson process with parameter λt.

Table 1 is a statistical sample table of the number of requests created by virtual machines and the frequency of occurrence in a unit time (with 30s as a unit time) in the system shown in FIG. 1 .

The number of requests created by the virtual machine per unit time frequency of occurrence 0 52 1 43

2 twenty three 3 and above 13

Table 1

The arrival rate distribution curve of the virtual machine creation request shown in Figure 3 is obtained from Table 1. This curve reflects that the arrival rate of the virtual machine creation request conforms to the Poisson process. For the Poisson process, the sample mean is the maximum of the parameter λ Likelihood estimates, so,

A2) Obtain the service rate of the virtual machine created by the system;

Because on the physical server, the service time of the virtual machine created by the system has no memory, that is, within s and t at any point, the service time of the virtual machine created by the system does not affect each other. If the service rate of the virtual machine created by the system is μ, according to the relevant knowledge of probability theory, it can be concluded that the service time of the virtual machine created by the system is an exponential distribution with parameter μ, that is, the following formula II is established:

for

Formula II

Table 2 is a statistical sample table of the service time and occurrence frequency of the virtual machine created by the system in a certain test as shown in FIG. 1 .

service hours frequency of occurrence 0-30 35 30-60 38

60-90 44 90-120 4

Table 2

The service rate of the system from Table 2 is:

A3) Determine the quantitative relationship between them according to the arrival rate of the virtual machine creation request and the service rate of the virtual machine created by the system.

Assuming that there are k service lines on the physical server that can accept user requests to create virtual machines, the k service lines mean that the system S can provide k blocks of available memory to satisfy the user's request to create a virtual machine.

If λ _m > μ _m , it indicates that the arrival rate of user requests to create virtual machines exceeds the service rate of virtual machines created by the system. If this situation accumulates, it will inevitably lead to a large number of user requests queuing in system S, that is, system S The number of available memory blocks provided is the bottleneck; if λ _m < μ _m , it indicates that the service rate of creating virtual machines by the system exceeds the arrival rate of user requests to create virtual machines. The memory block is idle, that is, the system S is idling, wasting electric energy. Therefore, in order to avoid the above two situations, the following formula IV should be established:

For p _m to exist, there must be

When system S provides k blocks of available memory: ,λ _m =λ _m ≥0

so:

then:

Formula IV

A4) Determine the optimal number k of virtual machines that the system allows to create.

It is easy to draw from Formula IV that when the number m of virtual machines requested by users is less than the available memory block k provided by system S, the service rate of system S depends on m, and when the number m of virtual machines requested by users is greater than that of system S When the available memory block k provided by S, the service rate of system S depends on k, when the number m of virtual machines requested by users is equal to the available memory block k provided by system S, the number of virtual machines running on a single physical server reached the optimum.

From the sum of A1) and A2), it is obtained that λ>μ, that is, the arrival rate of system virtual machine creation requests is greater than the service rate of the system, and the optimal number of virtual machines that the system allows to create depends on the available memory block of the physical server currently being tested ( The memory of the physical server for the test is 64GB, and the available memory block is 30, and the guest operating system memory resource scheduling subsystem of each virtual machine uses a 2GB memory block). Reduce the arrival rate of virtual machine creation requests so that λ is close to μ, and it can be determined that the optimal number of virtual machines that the system allows to create is 30.

B) Modify the partner algorithm parameters of the host operating system memory resource scheduling subsystem, and divide the available memory of the system into k blocks. The k blocks of memory refer to that the blocks of memory can be continuous or discontinuous, but the block is continuous of. Its realization method is:

B1) Allocate k blocks of memory in the partner system of the current host operating system memory resource scheduling subsystem for the guest operating system memory resource scheduling subsystem of the virtual machine, and use a lock mechanism to lock the k blocks of memory;

When this step is implemented, adjust parameters such as MAX_ORDER in the kernel file of the host operating system memory resource scheduling subsystem, such as "include/linux/mmzone.h", so that the system can allocate large blocks of continuous memory, and use memory locks to prevent these memory Swap out to external storage.

B2) An item is added in the physical memory descriptor area of the guest operating system memory resource scheduling subsystem of each virtual machine to describe the memory block information currently used for creating the virtual machine.

In order to enable the memory resource scheduling subsystem of the guest operating system of the virtual machine to detect the distribution of memory blocks that can be provided at present, when this step is implemented, the kernel file of the memory resource scheduling subsystem of the guest operating system such as "arch/x86 /kernel/E820.C" and other files to reconstruct data structures such as e820map, explain the starting address of the physical memory available to the guest operating system memory resource scheduling subsystem of the virtual machine, the maximum page frame number and other information, and then construct the page table, initialize the page directory and a series of operations.

C) The memory resource scheduling subsystem of the host operating system schedules a certain block of k blocks of memory to the guest operating system memory resource scheduling subsystem of the current virtual machine. Its realization method is:

C1) The host operating system memory resource scheduling subsystem sequentially selects a certain block of k blocks of memory, and identifies the starting page frame number of the physical page table of this memory;

During the implementation of this step, modify the kernel file such as "mm/page_alloc.c" of the memory resource scheduling subsystem of the host operating system, and allocate a piece of available memory therefrom.

C2) The guest operating system memory resource scheduling subsystem of the virtual machine starts loading after obtaining a block of memory allocated by C1, reads the description information of the memory block in step B2 during the loading detection memory stage, and updates the current guest operating system memory resource scheduling Subsystem memory management module physical page frame number, physical page table start page frame number, page table mapping relationship, memory area, memory node and other data structures, after the update is completed, send a completion message to the host operating system memory resource scheduling subsystem;

C3) After receiving the completion message, the memory resource scheduling subsystem of the host operating system clears a piece of memory assigned by C1 from the memory management pool of the memory resource scheduling subsystem of the host operating system.

In the above C2) and C3), the message sent by the memory resource scheduling subsystem of the guest operating system to the memory resource scheduling subsystem of the host operating system is realized by an interrupt (IPI) interrupt between processors, which needs to be implemented in the memory resource scheduling subsystem of the host operating system Preset an interrupt handler to handle this message.

D) The memory resource scheduling subsystem of the host operating system reclaims the memory of the guest operating system memory resource scheduling subsystem of the virtual machine. Its realization method is:

D1) After the guest operating system memory resource scheduling subsystem of the virtual machine receives the downtime scheduling, it packs the number of physical page frames it has obtained and the initial page frame number of the physical page table into a message;

D2) The guest operating system memory resource scheduling subsystem of the virtual machine sends an interprocessor interrupt (IPI) to the host operating system memory resource scheduling subsystem, and delivers the message in D1) to the host operating system memory resource scheduling subsystem system;

D3) After the host operating system memory resource scheduling subsystem receives the interrupt between the processors, it executes a predefined interrupt processing function between the processors. The function of this function is to analyze D2) and deliver it to the host operating system memory resource According to the message of the dispatching subsystem, the memory of the guest operating system memory resource dispatching subsystem of the virtual machine that needs to be shut down is brought back into the memory management pool of the memory resource dispatching subsystem of the host operating system according to the message.

In summary, the present invention solves the method and system for scheduling memory resources between the memory resource scheduling subsystem of the host operating system and the memory resource scheduling subsystem of the guest operating system on the physical server of the data center, and at the same time establishes a request for creating a virtual machine The relationship between the rate and the service rate of the system effectively guides the scheduling of the number of virtual machines on the physical server to reasonably reduce the energy consumption of the physical server.

The above description is only a description of the preferred embodiments of the present invention, and does not limit the scope of the present invention. Any changes and modifications made by those of ordinary skill in the field of the present invention based on the above disclosures shall fall within the protection scope of the claims.

Claims (9)

1. a kind of system based on memory source supply scheduling virtual machine quantity is it is characterised in that include：In host operating system Deposit scheduling of resource subsystem and client operating system memory source scheduler subsystem；

Described host operating system memory source scheduler subsystem includes memory allocating module and Memory recycle module, wherein, institute State memory allocating module for being grasped to described client by described host operating system memory source scheduler subsystem storage allocation block Make system memory resource scheduler subsystem, described Memory recycle module is used for described host operating system memory source dispatches subsystem The internal memory of the client operating system memory source scheduler subsystem united after reclaiming machine of delaying；

Described client operating system memory source scheduler subsystem includes probe the memory module and internal memory release module, wherein, institute State probe the memory module to detect from described in start-up course for described client operating system memory source scheduler subsystem The memory block that the assignment of host operating system memory source scheduler subsystem comes, described internal memory release module is used for described client behaviour Make the internal memory that system memory resource scheduler subsystem taken in the machine moment of delaying and return to host operating system memory source to adjust Degree subsystem.

2. a kind of method supplying scheduling virtual machine quantity based on memory source, using as claimed in claim 1 based on internal memory The system of resource provision scheduling virtual machine quantity is it is characterised in that include：

Obtain the arrival rate of virtual machine creating request；Obtain system creation virtual machine service speed；Asked according to virtual machine creating Arrival rate and system creation virtual machine service speed determine quantitative relation between them, and determine that system allows the void creating The optimal number of plan machine；

The buddy algorithm parameter of modification host operating system memory source scheduler subsystem, the free memory of system is divided into k Block；

Host operating system memory source scheduler subsystem dispatches a certain piece of guest operation system to current virtual machine of k block internal memory System memory source scheduler subsystem；

Host operating system memory source scheduler subsystem reclaims virtual machine client operation system memory source scheduler subsystem Internal memory.

3. the method based on memory source supply scheduling virtual machine quantity as claimed in claim 2 is it is characterised in that described obtain The arrival rate taking virtual machine creating request is obtained by below equation：

$p_m\left(t\right)=c_n^m{(\mathrm{\&lambda;}t/n)}^m{(1-\mathrm{\&lambda;}t/n)}^{n-m}$

Work as n->During ∞,

P_mT () represents the probability having m virtual machine request to create in time interval t, when λ represents that virtual machine request creates, new The arrival rate of virtual machine creating request, n represents in time interval t, will be cut into n decile the time and carry out infinitesimalization.

4. the method based on memory source supply scheduling virtual machine quantity as claimed in claim 3 is it is characterised in that described obtain System creation virtual machine service speed is taken to obtain by below equation：

P{X>T }=e^-μt；

Wherein, P { X>T } represent virtual machine request create when, the service speed of system creation virtual machine, and P { X>T } represent random Event X is in condition X>The probability occurring under t, t is certain constant, and P { X>T } obey e^-μtExponential, μ be exponential e^-μtParameter and for system creation virtual machine arrival rate.

5. as claimed in claim 4 the method for scheduling virtual machine quantity is supplied it is characterised in that described based on memory source Arrival rate and system creation virtual machine service speed according to virtual machine creating request determine that the quantitative relation between them is excessively following Formula obtains：

k:The memory block that host operating system memory source scheduler subsystem provides

m:Request creates the number of virtual machine

p_m:Request creates the probability of m virtual machine

P₀For virtual machine creating in the unit time request number of times be 0 probability.

6. as claimed in claim 2 the method for scheduling virtual machine quantity is supplied based on memory source it is characterised in that described K block internal memory refers between the block of internal memory continuously or discontinuously, is continuous in block.

7. the method based on memory source supply scheduling virtual machine quantity as claimed in claim 2 is it is characterised in that described repair Change the buddy algorithm parameter of host operating system memory source scheduler subsystem, the free memory of system is divided into the step of k block Rapid inclusion：

In the buddy system of current host operating system memory source scheduler subsystem, distribution k block internal memory is used for the visitor of virtual machine Family operating system memory scheduling of resource subsystem, pins described k block internal memory using lock mechanism；

Increase a use in the physical memory descriptor area of the client operating system memory source scheduler subsystem of each virtual machine It is currently used in the internal memory block message of this virtual machine creating in description.

8. as claimed in claim 7 the method for scheduling virtual machine quantity is supplied it is characterised in that described place based on memory source Master operating system memory source scheduler subsystem dispatches a certain piece of client operating system internal memory to current virtual machine of k block internal memory The step of scheduling of resource subsystem includes：

Host operating system memory source scheduler subsystem order selects a certain piece of k block internal memory, and identifies the thing of this block internal memory The initial page frame number of reason page table；

The client operating system memory source scheduler subsystem of virtual machine loads a certain piece of internal memory, and reads this internal memory block message, Update physical page frame quantity, the physical page table of current client operating system memory source scheduler subsystem memory management module simultaneously Initial page frame number, page table mapping relations, region of memory, the data structure such as memory node, after the completion of renewal, send one and complete Message is to host operating system memory source scheduler subsystem；

After host operating system memory source scheduler subsystem receives the message completing, by this block internal memory from host operating system Clear out in the memory management pond of memory source scheduler subsystem.

9. as claimed in claim 8 the method for scheduling virtual machine quantity is supplied it is characterised in that described place based on memory source Master operating system memory source scheduler subsystem reclaims the internal memory of virtual machine client operation system memory source scheduler subsystem Step includes：

After the client operating system memory source scheduler subsystem of virtual machine receives machine scheduling of delaying, by its acquired physical page frame Quantity, the initial page frame number of physical page table are packaged into a message；

The interruption that the client operating system memory source scheduler subsystem of virtual machine sends between a processor operates to host System memory resource scheduler subsystem, and by described message dilivery to host operating system memory source scheduler subsystem；

Host operating system memory source scheduler subsystem is had no progeny in receiving between this processor, and execution one is fixed in advance Interrupt processing function between the processor of justice, the function of this function is to be delivered to host operating system internal memory money described in parsing The message of source scheduler subsystem, and according to this message, the virtual machine client operation system memory source being currently needed for the machine of delaying is adjusted The internal memory of degree subsystem reintegrates in the memory management pond of host operating system memory source scheduler subsystem.