CN101916207B - Energy saving method, device and system under desktop virtual environment - Google Patents

Abstract

The embodiment of the present invention discloses an energy-saving method, device, and system in a desktop virtualization environment, wherein the method includes: collecting specific event information corresponding to a virtual machine through a probe; virtual machine idle policy, determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine state information to the management server, wherein the virtual machine state information indicates that the virtual machine is in an idle state or a non-idle state ; receiving the standby command sent by the management server, and sending a corresponding standby command to the hardware layer, so as to control and make the physical host of the virtual machine host enter the standby state of suspending memory; thereby achieving the energy-saving effect while avoiding Network traffic congestion caused by live migration of a large number of virtual machines or loading of a large number of virtual machines.

Description

Energy saving method, device and system in desktop virtualization environment

technical field

The embodiments of the present invention relate to the technical field of virtualization, and in particular, to an energy-saving method, device and system in a desktop virtualization environment.

Background technique

Virtualization technology makes it possible to install multiple different operating systems on a physical server and realize the isolation of each operating system. Even if an operating system or an application exception on an operating system is abnormal, it will not affect other operating systems. Virtualization technology sets up a layer of "Virtual Machine Monitor (Virtual Machine Monitor, VMM)" between the traditional operating system and physical hardware resources to realize the monitoring of physical hardware resources, such as processors, memory, storage, networks and other peripherals. Control and management of resources; the traditional operating system runs on the VMM and loses control and access to physical resources, which is called a virtual machine (Virtual Machine, VM); VMM provides a virtual hardware environment for the virtual machine to correctly Run, and realize the scheduling, management and isolation of virtual machines.

Virtualization technology has brought great challenges to the traditional enterprise desktop office environment: personal office computers are replaced by virtual machines, and end users access the virtual machine desktop office through thin client (Thin Client, TC) or software terminal (Software Client, SC) Enterprises use virtualization management software to centrally manage and control physical servers, storage devices and network devices, and deploy virtual machines and management servers to form a desktop virtualization system.

Currently, in the desktop virtualization environment, there are two solutions for energy-saving management of virtual machines:

1. If a large number of virtual machines do not need to run any business at night, shut down all virtual machines and power off the corresponding physical hosts;

2. Centrally migrate low-load virtual machines (using "hot migration" technology) to a small number of physical machines through the dynamic load balancing mechanism, and power off the physical machines when there are no more virtual machines on the physical machines.

In the process of implementing the present invention, the inventor found that: in the existing solution 1, in the desktop virtualization environment, when a large number of users need to use virtual machines at the same time (such as during rush hour), the physical host needs to load a large number of virtual machines from the storage network. Virtual machines can easily lead to network traffic congestion, and virtual machines start slowly or fail to start; and in the existing solution 2, the resource usage of virtual machines is relatively stable, and the average resource load of physical hosts is relatively stable, so there are very few physical hosts that can be powered off. The energy-saving effect is not high; moreover, if a large number of virtual machine migrations will be triggered during the peak hours of work, a large number of "hot migration" will easily lead to network traffic congestion.

Contents of the invention

Embodiments of the present invention provide an energy-saving method, device, and desktop virtualization system in a desktop virtualization environment, so as to avoid network traffic congestion caused by hot migration or loading of a large number of virtual machines while achieving energy-saving effects.

On the one hand, an embodiment of the present invention provides a desktop virtualization system, including multiple physical hosts and management servers, wherein each physical host includes a hardware layer, and a virtual machine monitoring unit VMM running on the hardware layer runs on At least one virtual machine on the virtual machine monitoring unit VMM, and at least one probe, wherein:

The probe is used to: collect specific event information corresponding to the virtual machine; the virtual machine monitoring unit VMM is used to: determine the specific event information of the virtual machine collected by the probe and the obtained idle policy of the virtual machine. The current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine state information to the management server, wherein the virtual machine state information indicates that the virtual machine is in an idle state or a non-idle state;

The management server is configured to: send a standby command to the hosted physical host when the obtained virtual machine state information indicates that all virtual machines hosted on the hosted physical host are idle;

The virtual machine monitoring unit VMM is further configured to: receive a standby command issued by the management server, and send a corresponding standby command to the hardware layer, so as to control and make the physical host enter a standby state of suspending to memory.

On the other hand, an embodiment of the present invention provides a physical host, and the physical host includes: a hardware layer, a virtual machine monitoring unit VMM running on the hardware layer, and a virtual machine monitoring unit VMM running on the virtual machine monitoring unit VMM At least one virtual machine, and at least one probe where:

The probe is used to: collect specific event information corresponding to the virtual machine;

The virtual machine monitoring unit VMM is configured to: determine the current state of the virtual machine associated with the specific event information according to the specific event information of the virtual machine collected by the probe and the acquired idle policy of the virtual machine, and report to the management server Outputting corresponding virtual machine state information, receiving a standby command sent by the management server, and sending a corresponding standby instruction to the hardware layer, wherein the virtual machine state information indicates that the virtual machine is in an idle state or a non-idle state ;

The hardware layer is used for: responding to the standby instruction, controlling and causing the physical host to enter the standby state of suspending to memory.

On the other hand, an embodiment of the present invention provides a management server, including:

A policy management unit, configured to manage a virtual machine idle policy, and deliver the virtual machine idle policy to a physical host, where the virtual machine idle policy is used by the physical host to determine the status of a virtual machine hosted on the physical host;

The cloud management unit is configured to send a standby command to the hosted physical host when the obtained virtual machine status information indicates that all virtual machines hosted on the hosted physical host are idle, and the standby command Used to control and put the physical host into a suspend-to-ram standby state.

On the other hand, an embodiment of the present invention provides a method for energy saving in a desktop virtualization environment, the method comprising:

Collect specific event information corresponding to the virtual machine through the probe;

According to the collected specific event information of the virtual machine and the acquired virtual machine idle policy, determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine state information to the management server, wherein the virtual machine state The information indicates that the virtual machine is idle or not idle;

Receive the standby command sent by the management server, and send a corresponding standby command to the hardware layer, so as to control and make the physical host hosting the virtual machine enter the standby state of suspending memory.

It can be seen that the embodiment of the present invention provides an energy-saving solution in a desktop virtualization environment. A probe is deployed in each virtual machine, and the specific event information of the corresponding virtual machine is collected through the probe; the virtual machine monitoring unit VMM according to the probe The collected specific event information of the virtual machine and the idle policy of the virtual machine provided by the management server determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine state information to the management server; when the management server When sensing that all the virtual machines hosted on the hosted physical host are in an idle state, send a standby command to the hosted physical host; VMM receives the standby command issued by the management server, and sends the corresponding to the hardware layer Standby command, to control and make the physical host enter the standby state of suspending memory; it can be seen that through the solution of the embodiment of the present invention, the usage of the virtual machine can be identified, so that for example, after the enterprise is off work, it can make itself All the virtual machines hosted above are idle physical hosts to save energy. At the same time, because the physical hosts are in the standby state from suspend to memory, their memory supplies power and the memory status information is complete; Therefore, while achieving the energy-saving effect, it also avoids network traffic congestion caused by a large number of virtual machines "hot migration" or a large number of virtual machine loading when the enterprise is at work, thus ensuring the virtual machine Stable, reliable and high performance, which improves end-user experience satisfaction.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are For some embodiments of the present invention, those skilled in the art can also obtain other drawings based on these drawings without any creative work.

FIG. 1a is a schematic diagram of a logical structure of a desktop virtualization system according to an embodiment of the present invention;

FIG. 1b is a schematic diagram of physical deployment of a desktop virtualization system according to an embodiment of the present invention;

FIG. 2 is a schematic structural diagram of a physical host according to an embodiment of the present invention;

3 is a schematic structural diagram of a probe embedded in a virtual machine according to an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of a management server according to an embodiment of the present invention;

FIG. 5 is a schematic flowchart of an energy saving method in a desktop virtualization environment according to an embodiment of the present invention;

FIG. 6 is a schematic flowchart of another energy saving method in a desktop virtualization environment according to an embodiment of the present invention;

FIG. 7 is a schematic flowchart of another energy saving method in a desktop virtualization environment according to an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Please refer to FIG. 1a, which is a schematic diagram of a logical structure of a desktop virtualization system according to an embodiment of the present invention. As shown in FIG. It can be referred to as a service host, shown as three physical hosts in the figure, but not limited thereto) and a management server 40, wherein each physical host includes a hardware layer (shown as 11, 21, 31 in the figure), running on the The virtual machine monitoring unit VMM on the hardware layer (indicated by 12, 22, 32 in the figure), at least one virtual machine running on the virtual machine monitoring unit VMM (indicated by VM1, VM2, VM3 in the figure, but is not limited to this), and at least one probe (shown as 13, 23, 33 in the figure, it should be noted that the deployment method of the probe is not limited to this), wherein:

The probe is used to: collect specific event information corresponding to the virtual machine;

The specific event information here may include hardware usage information and/or virtual machine user operation information;

The virtual machine monitoring unit VMM 12, 22, 32 is used for: according to the specific event information of the virtual machine collected by the probe, and the virtual machine idle policy provided by the management server 40, determine the current state of the virtual machine associated with the specific event information, And output corresponding virtual machine status information to the management server 40, wherein the virtual machine status information indicates that the virtual machine is in an idle state or a non-idle state;

And, the management server 40 is used for: when the obtained virtual machine status information indicates that all the virtual machines hosted on the hosted physical host (such as the physical host 10) are in an idle state, send a message to the hosted physical host (such as the physical host 10). Host 10) sends a standby command;

Correspondingly, the virtual machine monitoring unit VMM 12 is further used to: receive the standby command issued by the management server 40, and send a corresponding standby command to the hardware layer 11, so as to control and make the physical host 10 enter the standby state of suspending memory That is, the ACPI-S3 state.

The ACPI-S3 state here refers to the S3 state of ACPI, that is, the Suspend to RAM (suspend to memory) state, referred to as STR; the working state data before the system enters STR is stored in the memory. In the STR state, the power supply continues to supply power to the most necessary devices such as memory to ensure that data is not lost, while other devices are turned off, and the power consumption of the system is extremely low. Once the Power button (host power switch) is triggered, or through mouse and keyboard events, network card events, etc., the system will wake up, read data from the memory immediately and restore to the working state before STR. Among them, ACPI refers to Advanced Configuration and Power Interface, advanced configuration and power interface; this is a power management standard jointly developed by Intel, Microsoft and Toshiba; ACPI has a total of six states from S0 to S5 according to the degree of energy saving. ACPI-S3 is one of the standby states, the memory is powered, the CPU is not powered, and the energy consumption can be saved by more than 80%;

In one implementation, as shown in FIG. 1a, the probes 13, 23, and 33 are embedded in the corresponding virtual machine, and are specifically used to collect hardware usage information of the corresponding virtual machine and/or user operation information of the corresponding virtual machine;

Correspondingly, the virtual machine monitoring unit VMM 12, 22, 32 is specifically used to: determine the virtual machine idle policy provided by the management server 40 according to the hardware usage information and/or user operation information collected by the probes 13, 23, 33 The current state of the virtual machine associated with the hardware usage information and/or user operation information is output to the management server 40 corresponding virtual machine state information, wherein the virtual machine state information indicates that the virtual machine is in an idle state or a non-idle state.

In another implementation, the probes include front-end probes embedded in corresponding virtual machines (indicated by 13, 23, and 33 in FIG. 1a ) and back-end probes deployed in the virtual machine monitoring unit VMM. A needle (not shown in the figure), wherein, the front-end probe is used to collect user operation information of the corresponding virtual machine; the back-end probe is used to collect hardware usage information of the corresponding virtual machine;

Correspondingly, the virtual machine monitoring unit VMM 12, 22, 32 is specifically used for: according to the user operation information of the corresponding virtual machine collected by the front-end probe and the hardware usage information of the corresponding virtual machine collected by the back-end probe, and the management server 40 provides virtual machine idle policy, determine the current status of the virtual machine associated with the hardware usage information and user operation information, and output the corresponding virtual machine status information to the management server 40, wherein the virtual machine status information indicates that the virtual machine is idle state or non-idle state.

It should be noted that the hardware usage information of the virtual machine here represents hardware events that occur on the virtual machine, and the hardware events include but are not limited to: mouse events, keyboard events, network I/O events, and disk I/O events Etc., the user operation information of the virtual machine here represents the operation event that occurs on the virtual machine, and the operation event includes but not limited to: "idle" operation, "standby" operation, "sleep" operation performed by the user on the virtual machine operation and "disconnect" operation, etc.;

The virtual machine idle policy in the embodiment of the present invention includes: virtual machine idle time section, virtual machine idle category, virtual machine hardware usage information idle judgment condition, and virtual machine user operation information idle judgment condition. One or any of the four conditions Combination; the details can be flexibly determined according to the actual application.

For example: The virtual machine idle policy is: "The virtual machine idle category is idle" + "Idle time range 18:00-23:00" + "Mouse + keyboard without events for 30 minutes"||"User standby operation"; If virtual machine A belongs to the idle category, and the current time period is between 18:00-23:00, and virtual machine A has no mouse button event for 30 minutes, or virtual machine A is held by the user for standby operation, then virtual machine A Otherwise, virtual machine A is in a non-idle state; it should be noted that "+" here means "and" relationship, and "||" means "or" relationship.

Among them, the virtual machine idle time period refers to the time period during which the virtual machine can be idle, such as 20:00-06:00; the virtual machine idle category refers to whether the virtual machine can be idle; for example, virtual machine A is a non-idleable category , then no matter whether other conditions are satisfied, the virtual machine A will not be idle;

The virtual machine hardware usage information idle judgment condition refers to judging whether the virtual machine can be idle according to the occurrence frequency of hardware events represented by the reported virtual machine hardware usage information. Various virtual machine hardware events can be used in any combination. The hardware event Including but not limited to mouse events, keyboard events, network I/O events and disk I/O events, etc.;

The virtual machine user operation information idle judgment condition refers to determining whether the virtual machine that is executed with the operation represented by the virtual machine user operation information can be idle; specifically, if the operation represented by the reported virtual machine user operation information is an idle operation , then the virtual machine can be idle; if the operation indicated by the reported virtual machine user operation information is an operation within the predetermined operation range, then it is determined whether the virtual machine can be idle according to the set operation substitution rules; specifically, if the The "standby", "hibernation" or "disconnect" operation of the virtual machine can be replaced by the "idle" operation, and the virtual machine that has been executed with the "standby", "hibernation" or "disconnect" operation can be idle; if the virtual machine " "Standby", "Hibernation" or "Disconnect" operations cannot be replaced by "Idle" operations, and the virtual machines that are executed by "Standby", "Hibernation" or "Disconnect" operations cannot be idle; the scope of predetermined operations here includes " Standby operation, Hibernation operation or Disconnect operation.

It should be noted that, in order to enhance the flexibility of the virtual machine idle policy, for different virtual machine idle time periods and/or virtual machine idle categories, the corresponding virtual machine hardware usage information idle judgment conditions and/or virtual machine user operation information The idle determination conditions may be different.

In one implementation, the virtual machine monitoring unit VMM 12, 22, 32 is specifically used for: when the virtual machine associated with the reported information belongs to the idle category, and the current time period is within the first idle time period, and the virtual machine The machine does not have a hardware event within the first threshold time, or the frequency of hardware events is less than the first threshold frequency, or the virtual machine is performing an idle operation, or the virtual machine is performing an operation within a predetermined operating range and the operation can Substitute with " idle " operation, then determine that this virtual machine is in idle state at present, and output the virtual machine status information that represents idle state to management server 40; Here hardware event comprises mouse event, keyboard event, network I/O event and One or more of disk I/O events;

In another implementation, the virtual machine monitoring unit VMM 12, 22, 32 is specifically used for: when the virtual machine associated with the reported information belongs to the idle category, and the current time period is within the second idle time period, and the The virtual machine does not have a hardware event within the second threshold time, or the frequency of hardware events is less than the second threshold frequency, or the virtual machine is idle, or the virtual machine is performing an operation within a predetermined operating range and the operation If it can be replaced by an "idle" operation, then it is determined that the virtual machine is currently in an idle state, and the virtual machine status information indicating the idle state is output to the management server 40;

In another implementation, the virtual machine monitoring unit VMM 12, 22, 32 is specifically used for: when the virtual machine associated with the reported information belongs to the idle category, and the virtual machine is executed within the predetermined operation range and the operation If the "idle" operation cannot be used instead, then it is determined that the virtual machine is currently in a non-idle state, and the virtual machine status information indicating the non-idle state is output to the management server 40;

In another implementation, the virtual machine monitoring unit VMM 12, 22, 32 is specifically used for: when the virtual machine associated with the reported information belongs to the non-idle category, then determine that the virtual machine is currently in the non-idle state, and report to the management server 40 outputting virtual machine status information indicating a non-idle state;

It should be understood that, the granularity of classifying virtual machines is not limited to: a class that can be idled and a class that cannot be idled. In the case of more fine-grained management, the idle category can be further divided into different subcategories, such as A, B, C.

Further, as shown in FIG. 1a, in the desktop virtualization system of the embodiment of the present invention, the management server 40 is further used to: determine the virtual machine that needs to be migrated and is in an unidleable state according to the virtual machine status information and the virtual machine migration policy. machine and the migration destination physical host, and send a migration command to the physical host hosting the virtual machine to be migrated; wherein, the migration command includes the ID of the virtual machine to be migrated and the information of the migration destination host. The virtual machine identification ID here is a unique string sequence assigned by VMM to each virtual machine, which is used to uniquely identify the virtual machine; it is called uuid under the Xen architecture; the destination host information of the migration can be the IP of the destination business host address.

Correspondingly, the virtual machine monitoring unit VMM 12, 22, 32 is further configured to: migrate the virtual machine corresponding to the virtual machine ID to the destination service host corresponding to the destination service host information according to the migration command sent by the management server 40 superior.

Let's introduce the virtual machine migration strategy:

The virtual machine migration strategy includes: a virtual machine migration time zone, a virtual machine migration judgment condition, and a migration target business host selection condition; The virtual machines in the "non-idle state" on the service hosts whose number is less than 2 and the total number of virtual machines is greater than the average number of virtual machines hosted by the physical host are migrated to the service host with the highest proportion of "non-idle state" virtual machines and the lowest resource utilization.

Wherein, the virtual machine migration judging condition refers to judging whether the current virtual machine needs to be migrated according to the state of the current virtual machine and the state information of the physical host hosting the current virtual machine. The state information of the physical host hosting the current virtual machine includes One or any combination of the total number of virtual machines hosted on the hosting physical host, the number of "idle" virtual machines, the average number of virtual machines;

The selection condition of the target business host refers to the selection of the target physical host for migration based on the status information of the physical host in the desktop virtualization environment. One or any combination of the number/proportion and resource utilization of business hosts; specifically, select a physical host that meets one or more of the following conditions as the target physical host for migration;

a. In the desktop virtualization environment, the total number of virtual machines hosted on the physical host is the largest, or the total number of virtual machines hosted on the physical host is greater than the first threshold number;

b. In the desktop virtualization environment, the proportion of "non-idle state" virtual machines hosted on the physical host is the largest, or the proportion of "non-idle state" virtual machines hosted on the physical host is greater than the first threshold percentage;

c. The resource utilization rate of the physical host is the lowest in the desktop virtualization environment, or the resource utilization rate of the physical host is lower than the second threshold percentage.

It should be understood that, in the embodiments of the present invention, in order to enhance the flexibility of migration, the corresponding virtual machine migration judgment conditions and/or migration target service host selection conditions may be different for different virtual machine migration time periods.

Further, please refer to FIG. 1b, which is a schematic diagram of physical deployment of a desktop virtualization system according to an embodiment of the present invention. As shown in FIG. 1b, in the desktop virtualization system according to this embodiment of the present invention, the management server 40 is further used to: respond The user logs in to the operation of the virtual machine, and sends a wake-up instruction to the physical host (such as the physical host 10) where the virtual machine hosts;

Correspondingly, the hardware layer 11 of the physical host 10 is used for: responding to the wake-up command sent by the management server 40, running the suspended memory, and activating the corresponding virtual machine, so that the virtual machine enters a manipulable state.

It should be noted that the wake-up instruction is an instruction sent directly by the management server 40 to the hardware layer of the physical host; it is received by the hardware layer (specifically, a network card), and triggers the CPU to be powered on, loaded and running.

In summary, the embodiment of the present invention provides a desktop virtualization system, deploying a probe in each virtual machine, and collecting specific event information of the corresponding virtual machine through the probe; the virtual machine monitoring unit VMM collects The specific event information of the virtual machine and the virtual machine idle policy provided by the management server, determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine status information to the management server; when the management server perceives When all the virtual machines hosted on the hosted physical host are idle, send a standby command to the hosted physical host; VMM receives the standby command issued by the management server, and sends the corresponding standby command to the hardware layer. Instructions to control and make the physical host enter the standby state of suspending memory; it can be seen that through the solution of the embodiment of the present invention, the usage of the virtual machine can be identified, so that, for example, after the enterprise is off work, it can make itself go online All the host virtual machines are idle physical hosts to save energy. At the same time, because the physical hosts are in the standby state of suspend to memory, their memory supplies power and the memory status information is complete; the business host is woken up, directly from the suspended The memory runs and quickly returns to the state before standby, so while achieving energy saving effects, it avoids network traffic congestion caused by "hot migration" of a large number of virtual machines or loading of a large number of virtual machines when the enterprise goes to work, thereby ensuring the virtual machine. Stable, reliable and high-performance, improving end-user experience satisfaction. Further, through the solution of the embodiment of the present invention, for example, for those virtual machines that still need to work at night, the virtual machines marked as "non-idle state" can be "hot migrated" to other business hosts; because the desktop virtualization system of the embodiment of the present invention The number of such virtual machines is small, and "hot migration" will not cause network traffic congestion.

Please refer to FIG. 2 , which is a schematic structural diagram of a physical host according to an embodiment of the present invention. The physical host can be applied to a desktop virtualization system. As shown in FIG. 2 , the physical host includes: a hardware layer 11 running on the hardware layer The virtual machine monitoring unit VMM 12 on the 11 and at least one virtual machine running on the virtual machine monitoring unit VMM 12 (shown as VM1 and VM2 in the figure, including but not limited to this), and at least one probe (shown as probe in the figure) Needle 13 is shown, but it should be noted that the deployment method of the probe is not limited to this), wherein:

The probe 13 is used to: collect specific event information corresponding to the virtual machine;

The virtual machine monitoring unit VMM 12 is used to: determine the current state of the virtual machine associated with the specific event information according to the specific event information of the virtual machine collected by the probe 13 and the acquired virtual machine idle policy, and output the information to the management server 40 The corresponding virtual machine status information is used to receive the standby command sent by the management server 40 and send the corresponding standby command to the hardware layer. Wherein, the virtual machine status information indicates that the virtual machine is in an idle state or a non-idle state; the acquisition mode of the virtual machine idle policy here may be the virtual machine idle policy issued by the management server 40 in response to the request of the physical host, or It may be actively pushed to the physical host by the management server 40, or may be a predetermined virtual machine idle policy, and so on.

The hardware layer 11 is configured to respond to the standby command, control and make the physical host enter into the standby state of suspending to memory, ie the ACPI-S3 state.

It should be noted that, in one implementation mode, the probe 13 is embedded in the corresponding virtual machine, specifically embedded in the guest operating system Guest OS of the corresponding virtual machine, and has two levels: a, the same level as the driver Drive; b. It is at the same level as Application; the probe 13 is specifically used to collect hardware usage information of the corresponding virtual machine and/or user operation information of the corresponding virtual machine;

Correspondingly, the virtual machine monitoring unit VMM 12 is specifically configured to: determine the hardware usage information and/or user operation information of the corresponding virtual machine collected by the probe 13, and the idle policy of the virtual machine provided by the management server 40. /or the current state of the virtual machine associated with the user operation information, and output corresponding virtual machine state information to the management server 40, wherein the virtual machine state information indicates that the virtual machine is in an idle state or a non-idle state. Specifically, output the status information of each virtual machine to the management server 40 through message communication or shared file access;

It should be noted that the hardware usage information of the virtual machine here represents hardware events that occur on the virtual machine, and the hardware events include but are not limited to: mouse events, keyboard events, network I/O events, and disk I/O events Etc., the user operation information of the virtual machine here represents the operation event that occurs on the virtual machine, and the operation event includes but not limited to: "idle" operation, "standby" operation, "sleep" operation performed by the user on the virtual machine operation and "disconnect" operation, etc.; as a supplement to this device, the virtual machine can provide an "idle" operation option or menu for the user to actively perform the "idle" operation of the virtual machine;

In yet another implementation, as shown in FIG. 3, front-end probes 13aa are respectively deployed in virtual machines VM1 and VM2, and back-end probes 13b, front-end probes 13aa and The back-end probe 13b can communicate through an event channel. In other words, the probe includes a front-end probe 13aa embedded in the corresponding virtual machine and a back-end probe 13b deployed in the virtual machine monitoring unit VMM, wherein the front-end probe 13aa It is specifically used to collect user operation information of the corresponding virtual machine; the back-end probe 13b is specifically used to collect the hardware usage information of the corresponding virtual machine; it should be understood that the probes are not necessarily divided into "front-end probe" and "back-end probe" probe".

Correspondingly, the virtual machine monitoring unit VMM12 is specifically used for: according to the user operation information of the corresponding virtual machine collected by the front-end probe 13a and the hardware usage information of the corresponding virtual machine collected by the back-end probe 13b, and the virtual machine provided by the management server 40 The idle policy is to determine the current state of the virtual machine associated with the hardware usage information and user operation information, and output the corresponding virtual machine state information to the management server 40, wherein the virtual machine state information indicates that the virtual machine is in an idle state or not idle state.

Preferably, the virtual machine monitoring unit 12 is further configured to receive the migration command sent by the management server 40, and migrate the virtual machine corresponding to the virtual machine ID contained in the migration command to the destination service host information contained in the migration command on the corresponding destination service host. And, the hardware layer 11 is further configured to: receive a wake-up command sent by the management server 40, respond to the wake-up command, run the suspended memory, activate the corresponding virtual machine, and make the virtual machine enter a manipulable state. Specifically, the wake-up instruction is an instruction sent directly by the management server to the hardware layer of the physical host; it is received by the hardware layer (specifically, a network card), and triggers the CPU to be powered on, loaded and running.

The following describes how the probe collects virtual machine hardware usage information and/or virtual machine user operation information through the following methods:

1. User operation information corresponding to the virtual machine: the probe intercepts the corresponding user operation event by registering the corresponding operation event processing function with the Guest OS;

2. Corresponding to the mouse and keyboard information in the hardware usage information of the virtual machine: the probe intercepts the corresponding events by registering the event processing functions of the mouse and keyboard with the Guest OS; network I/O information: the probe registers the corresponding events with the Guest OS The driver interface intercepts network I/O events; disk I/O information: the probe registers the corresponding driver interface with the Guest OS to intercept disk I/O events;

After the virtual machine hardware usage information and/or virtual machine user operation information is collected, it can be transmitted to the virtual machine monitoring unit VMM 12, 22, 32 through an event channel or message communication;

The virtual machine idle policy is described below:

The virtual machine idle policy here includes: one or any combination of four conditions: virtual machine idle time period, virtual machine idle category, virtual machine hardware usage information idle judgment condition and virtual machine user operation information idle judgment condition; It can be flexibly determined according to actual applications.

For example: The virtual machine idle policy is: "The virtual machine idle category is idle" + "Idle time range 18:00-23:00" + "Mouse + keyboard without events for 30 minutes"||"User standby operation"; If virtual machine A belongs to the idle category, and the current time period is between 18:00-23:00, and virtual machine A has no mouse button event for 30 minutes, or virtual machine A is held by the user for standby operation, then virtual machine A Otherwise, virtual machine A is in a non-idle state; it should be noted that "+" here means "and" relationship, and "||" means "or" relationship.

Among them, the virtual machine idle time period refers to the time period during which the virtual machine can be idle, such as 20:00-06:00; the virtual machine idle category refers to whether the virtual machine can be idle; for example, virtual machine A is a non-idleable category , then no matter whether other conditions are satisfied, the virtual machine A will not be idle;

The virtual machine hardware usage information idle judgment condition refers to judging whether the virtual machine can be idle according to the occurrence frequency of hardware events represented by the reported virtual machine hardware usage information. Various virtual machine hardware events can be used in any combination. The hardware event Including but not limited to mouse events, keyboard events, network I/O events and disk I/O events, etc.;

The virtual machine user operation information idle judgment condition refers to determining whether the virtual machine that is executed with the operation represented by the virtual machine user operation information can be idle; specifically, if the operation represented by the reported virtual machine user operation information is an idle operation , then the virtual machine can be idle; if the operation indicated by the reported virtual machine user operation information is an operation within the predetermined operation range, then it is determined whether the virtual machine can be idle according to the set operation substitution rules; specifically, if the The "standby", "hibernation" or "disconnect" operation of the virtual machine can be replaced by the "idle" operation, and the virtual machine that has been executed with the "standby", "hibernation" or "disconnect" operation can be idle; if the virtual machine " "Standby", "Hibernation" or "Disconnect" operations cannot be replaced by "Idle" operations, and the virtual machines that are executed by "Standby", "Hibernation" or "Disconnect" operations cannot be idle; the scope of predetermined operations here includes " Standby operation, Hibernation operation or Disconnect operation.

It should be noted that, in order to enhance the flexibility of the virtual machine idle policy, for different virtual machine idle time periods and/or virtual machine idle categories, the corresponding virtual machine hardware usage information idle judgment conditions and/or virtual machine user operation information The idle determination conditions may be different.

In one implementation, the virtual machine monitoring unit VMM 12 is specifically configured to: when the virtual machine associated with the reported information belongs to the idle category, and the current time period is within the first idle time period, and the virtual machine is in the first idle time period, No hardware event occurs within the threshold time, or, the frequency of occurrence of hardware events is less than the first threshold frequency, or, the virtual machine is performed an idle operation, or the virtual machine is performed an operation within a predetermined operation range and the operation can be "idle" If the operation is replaced, it is determined that the virtual machine is currently in an idle state, and the virtual machine status information indicating the idle state is output to the management server 40, and is used to receive the standby command issued by the management server 40, and send a corresponding standby command to the hardware layer. Instructions to control and make the physical host enter the ACPI-S3 state; the hardware events here include one or more of mouse events, keyboard events, network I/O events and disk I/O events;

In another implementation, the virtual machine monitoring unit VMM 12 is specifically used for: when the virtual machine associated with the reported information belongs to the idle category, and the current time period is within the second idle time period, and the virtual machine is in the second idle time period. No hardware event occurs within the second threshold time, or, the frequency of occurrence of hardware events is less than the second threshold frequency, or, the virtual machine is executed with an idle operation, or the virtual machine is executed with an operation within a predetermined operation range and the operation can be performed as "idle " operation instead, then determine that the virtual machine is currently in an idle state, and output virtual machine state information indicating the idle state to the management server 40, and be used to receive the standby command issued by the management server 40, and send the corresponding to the hardware layer Standby command, to control and make the physical host enter the ACPI-S3 state, and to receive the standby command issued by the management server 40, and send a corresponding standby command to the hardware layer, to control and make the physical host enter ACPI-S3 state;

In another implementation, the virtual machine monitoring unit VMM 12 is specifically used for: when the virtual machine associated with the reported information belongs to the category that can be idled, and the virtual machine is executed within the predetermined operation range and the operation cannot be "idle " operation instead, determine that the virtual machine is currently in a non-idle state, and output virtual machine status information indicating the non-idle state to the management server 40, and be used to receive the standby command issued by the management server 40, and send a corresponding message to the hardware layer The standby command to control and make the physical host enter the ACPI-S3 state;

In another implementation, the virtual machine monitoring unit VMM 12 is specifically used for: when the virtual machine associated with the reported information belongs to the non-idleable category, then determine that the virtual machine is currently in a non-idleable state, and output to the management server 40 indicating that it cannot be idle. The status information of the virtual machine in the idle state, and the standby command for receiving the management server 40, and sending the corresponding standby command to the hardware layer, so as to control and make the physical host enter the ACPI-S3 state;

It should be understood that, the granularity of classifying virtual machines is not limited to: a class that can be idled and a class that cannot be idled. In the case of more fine-grained management, the idle category can be further divided into different subcategories, such as A, B, C.

Regarding the specific implementation of the above functional units, reference may be made to the description of the method embodiments.

Each unit of the device in the embodiment of the present invention can be integrated into one body, or can be deployed separately. The above units can be combined into one unit, and can also be further divided into multiple sub-units.

In summary, the embodiment of the present invention provides a physical host, deploying probes in each virtual machine, and collecting specific event information of the corresponding virtual machine through the probes; the virtual machine monitoring unit VMM collects virtual machine specific event information and the acquired virtual machine idle policy, determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine state information to the management server; when the management server perceives that the hosted When all virtual machines hosted on the physical host are in an idle state, a standby command is sent to the hosted physical host; VMM receives the standby command issued by the management server, and sends a corresponding standby command to the hardware layer to control And make the physical host enter the standby state of suspending to the memory; it can be seen that through the scheme of the embodiment of the present invention, the usage of the virtual machine can be identified, so that for example, after the enterprise is off work, all the hosts on itself can be The virtual machines are all idle physical hosts to save energy. At the same time, because the physical hosts are in the standby state of suspend to memory, their memory is powered and the memory status information is complete; the business host is awakened and runs directly from the suspended memory, quickly Restore to the state before standby, so while achieving the energy-saving effect, avoid the network traffic congestion caused by the "hot migration" of a large number of virtual machines or the loading of a large number of virtual machines when the enterprise goes to work, thereby ensuring the stability and reliability of the virtual machine. performance, improving end-user experience satisfaction. Further, through the solution of the embodiment of the present invention, for example, for those virtual machines that still need to work at night, the virtual machines marked as "non-idle state" can be "hot migrated" to other business hosts; because the desktop virtualization system of the embodiment of the present invention The number of such virtual machines is small, and "hot migration" will not cause network traffic congestion.

Please refer to FIG. 4, which is a schematic structural diagram of a management server according to an embodiment of the present invention. As shown in FIG. 4, the management server 40 includes a policy management unit 41 and a cloud management unit 42, wherein:

The policy management unit 41 is used to manage the idle policy of the virtual machine, and deliver the idle policy of the virtual machine to the physical host, and the idle policy of the virtual machine is used by the physical host to determine the status of the virtual machine hosted on the physical host; It should be noted that, generally, when the system is initialized, the policy management unit 41 issues the virtual machine idle policy. For the specific content of the virtual machine idle policy here, please refer to the foregoing embodiments, and details are not repeated here.

The cloud management unit 42 is configured to send a standby command to the hosted physical host when the obtained virtual machine state information indicates that all the virtual machines hosted on the hosted physical host are idle, and the standby command Used to control and put the physical host into a suspend-to-ram standby state. Wherein, the virtual machine state information indicates that the virtual machine is in an "idle state" or a "non-idle state";

In one implementation manner, the policy management unit 41 is further configured to manage the virtual machine migration policy; for the specific content of the virtual machine migration policy here, please refer to the foregoing embodiments, which will not be repeated here.

The cloud management unit 42 is further used for: according to the acquired virtual machine state information and virtual machine migration strategy, determine the virtual machine that needs to be migrated and is in the non-idle state and the migration destination service host, and send the host to the physical host that hosts the virtual machine that needs to be migrated. A migration command; wherein, the migration command includes the ID of the virtual machine to be migrated and the information of the destination host for migration. The virtual machine identification ID here is a unique string sequence assigned by VMM to each virtual machine, which is used to uniquely identify the virtual machine; for example, it is called uuid under the Xen architecture; the destination host information of the migration here can specifically be the destination IP address of the service host.

In one implementation, the cloud management unit 42 is further configured to: respond to the operation command of the user to log in to the virtual machine, and issue a wake-up instruction to the physical host hosting the virtual machine, and the wake-up instruction is used to make the virtual machine enter a controllable state; it should be noted that, as shown in FIG. 1b, the user logs in to the corresponding virtual machine through SC or TC, and the corresponding operation command is delivered to the management server 40 (specifically, the cloud management unit 42), and the management server 40 (specifically, the cloud management unit 42) 42) In response to the operation command, send a wake-up instruction to the physical host 10, 20, 30 of the virtual machine host, specifically send a wake-up instruction to the hardware layer of the physical host of the virtual machine host through a network message, and the wake-up instruction A wake-up command for the network card.

In practical applications, the management server in this embodiment of the present invention may be a physical server or a virtual machine.

Regarding the specific implementation of the above functional units, reference may be made to the description of the method embodiments.

Each unit of the device in the embodiment of the present invention can be integrated into one body, or can be deployed separately. The above units can be combined into one unit, and can also be further divided into multiple sub-units.

To sum up, the embodiment of the present invention provides a management server that, when the obtained virtual machine state information indicates that all the virtual machines hosted on the hosted physical host are in the "idle state" state, sends The physical host sends a standby command, so that the physical host responds to the standby command and enters the standby state that is suspended to the memory, that is, the ACPI-S3 state; it can be seen that through the solution of the embodiment of the present invention, it is possible to make some business hosts standby and save energy. The ACPI-S3 state of the business host, its memory power supply, the memory status information is complete; the business host is woken up, directly runs from the suspended memory, and quickly restores to the state before standby, so while achieving the energy saving effect, it is avoided. When a large number of virtual machines are "hot migrated" or the network traffic congestion caused by a large number of virtual machine loading, the stability, reliability and high performance of the virtual machines are ensured, and the end user experience satisfaction is improved.

Further, through the solution of the embodiment of the present invention, for example, for those virtual machines that still need to work at night, the virtual machines marked as "non-idle state" can be "hot migrated" to other business hosts; because the desktop virtualization system of the embodiment of the present invention The number of such virtual machines is small, and "hot migration" will not cause network traffic congestion.

Please refer to FIG. 5 , which provides a schematic flowchart of an energy-saving method in a desktop virtualization environment according to an embodiment of the present invention. The method can be applied to a physical host in a desktop virtualization system as shown in FIGS. 1a and 1b. The physical host includes : a hardware layer, a virtual machine monitoring unit VMM running on the hardware layer and at least one virtual machine running on the virtual machine monitoring unit VMM, and at least one probe, as shown in Figure 5, the method may include the following steps:

S501. Collect specific event information corresponding to the virtual machine through a probe;

S502. According to the collected specific event information of the virtual machine and the obtained idle policy of the virtual machine, determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine state information to the management server, wherein the The virtual machine state information indicates that the virtual machine is in an idle state or a non-idle state;

Here, the way to obtain the virtual machine idle policy can be the virtual machine idle policy issued actively by the management server, the virtual machine idle policy issued by the management server in response to the request of the physical host, or the predetermined virtual machine idle policy. strategy, etc.

The virtual machine idle policy here includes: one or any combination of four conditions: virtual machine idle time period, virtual machine idle category, virtual machine hardware usage information idle judgment condition and virtual machine user operation information idle judgment condition; It can be flexibly determined according to actual applications. For a detailed description of the virtual machine idle policy, please refer to the foregoing embodiments, and details are not repeated here.

S503. Receive a standby command sent by the management server, and send a corresponding standby command to the hardware layer, so as to control and make the physical host hosting the virtual machine enter the standby state of suspending memory, that is, the ACPI-S3 state.

In one implementation manner, S501 specifically includes: collecting hardware usage information and/or user operation information of the virtual machine through a probe embedded in the corresponding virtual machine; correspondingly, S502 specifically includes: according to the collected corresponding virtual machine The hardware usage information of the machine and/or the user operation information of the corresponding virtual machine and the idle policy of the virtual machine provided by the management server determine the current status of the virtual machine associated with the hardware usage information and/or user operation information, and report to the management The server outputs corresponding virtual machine state information.

In another implementation, S501 specifically includes: collecting the user operation information of the corresponding virtual machine by the front-end probe deployed in the corresponding virtual machine, and collecting the corresponding virtual machine by the back-end probe deployed in the virtual machine monitoring unit VMM Correspondingly, S502 specifically includes: according to the user operation information of the corresponding virtual machine collected by the front-end probe, the hardware usage information of the corresponding virtual machine collected by the back-end probe, and the virtual machine provided by the management server The idle policy determines the current state of the virtual machine associated with the hardware usage information and the user operation information, and outputs the corresponding virtual machine state information to the management server.

Preferably, the method in the embodiment of the present invention further includes: receiving a migration command sent by the management server, and migrating the virtual machine corresponding to the virtual machine ID contained in the migration command to the target service contained in the migration command On the destination service host corresponding to the host information. Wherein, the migration command includes the ID of the virtual machine to be migrated and the information of the destination host of the migration. The virtual machine identification ID here is a unique string sequence assigned by VMM to each virtual machine, which is used to uniquely identify the virtual machine; for example, it is called uuid under the Xen architecture; the destination host information for migration can be specifically the destination business host IP address.

Further, the method in the embodiment of the present invention further includes: receiving a wake-up command sent by the management server, responding to the wake-up command, running the suspended memory, activating the corresponding virtual machine, and making the virtual machine enter a manipulable state, where the wake-up command Issued by the management server in response to the operation of the user logging in to the virtual machine.

The hardware usage information of the virtual machine represents a hardware event occurring on the virtual machine, and the hardware event includes: one or more of a mouse event, a keyboard event, a network I/O event, and a disk I/O event; The user operation information of the virtual machine indicates an operation event that occurs on the virtual machine, and the operation event includes: one or more of an idle operation, a standby operation, a hibernation operation, and a disconnection operation performed by the user on the virtual machine ;

The virtual machine idle policy includes: one or any combination of virtual machine idle time period, virtual machine idle category, virtual machine hardware use information idle judgment condition and virtual machine user operation information idle judgment condition;

Wherein, the virtual machine idle time section refers to the time section during which the virtual machine can be idle; the virtual machine idle category refers to whether the virtual machine can be idle; the virtual machine hardware usage information idle judgment condition refers to Determine whether the virtual machine can be idle according to the frequency of occurrence of hardware events represented by the virtual machine hardware usage information; Whether the operating virtual machine can be idled. For specific content and implementation, reference may be made to the description of the foregoing embodiments, and details are not repeated here.

In one implementation, according to the collected hardware usage information and/or user operation information and the acquired virtual machine idle policy, determine the current status of the virtual machine associated with the hardware usage information and/or user operation information, and send the information to the management The server outputs the corresponding virtual machine status information, including:

When the virtual machine associated with the collected hardware usage information and/or user operation information belongs to the idle category, and the current time period is within the first idle time period, and no hardware event occurs on the virtual machine within the first threshold time , or, the frequency of occurrence of hardware events is less than the first threshold frequency, or, the virtual machine is executed an idle operation, or the virtual machine is executed an operation within a predetermined operating range and the operation can be replaced by an "idle" operation, then it is determined that the virtual machine The machine is currently in an idle state, and outputs virtual machine status information indicating the idle state to the management server; the hardware events here include one or more of mouse events, keyboard events, network I/O events, and disk I/O events ;

Or, when the virtual machine associated with the collected hardware usage information and/or user operation information belongs to the idle category, and the current time period is within the second idle time period, and the virtual machine does not occur within the second threshold time A hardware event, or, the hardware event occurrence frequency is less than the second threshold frequency, or, the virtual machine is performed an idle operation, or the virtual machine is performed an operation within a predetermined operation range and the operation can be replaced by an "idle" operation, then it is determined The virtual machine is currently in an idle state, and outputs virtual machine state information indicating the idle state to the management server; wherein, the predetermined operation range includes standby operation, hibernation operation or disconnection operation.

Or, when the virtual machine associated with the collected hardware usage information and/or user operation information belongs to the idle category, and the virtual machine is executed within the predetermined operation range and the operation cannot be replaced by the "idle" operation, then determine The virtual machine is currently in a non-idle state, and outputs virtual machine status information indicating the non-idle state to the management server;

Or, when the virtual machine associated with the collected hardware usage information and/or user operation information belongs to the non-idleable category, it is determined that the virtual machine is currently in a non-idleable state, and the virtual machine status information indicating the non-idleable state is output to the management server .

To sum up, the embodiment of the present invention provides an energy-saving method in a desktop virtualization environment, in which a probe is deployed in each virtual machine, and the specific event information of the corresponding virtual machine is collected through the probe; The specific event information of the virtual machine and the virtual machine idle policy provided by the management server determine the current state of the virtual machine associated with the specific event information, and output the corresponding virtual machine status information to the management server; The standby command sends the corresponding standby command to the hardware layer to control and make the physical host enter the standby state of suspend-to-memory; it can be seen that through the solution of the embodiment of the present invention, the usage of the virtual machine can be identified, thereby For example, after the enterprise is off work, all the virtual machines hosted by itself can be idled to save energy. At the same time, because the physical host is in the standby state of suspending to memory, its memory is powered and its memory status information is complete. ;The business host is woken up, runs directly from the suspended memory, and quickly restores to the state before standby, so while achieving energy saving effects, it avoids the "hot migration" of a large number of virtual machines or the loading of a large number of virtual machines when the enterprise goes to work, for example. Network traffic congestion, thereby ensuring the stability, reliability and high performance of virtual machines, and improving end-user experience satisfaction.

Further, through the solution of the embodiment of the present invention, for example, for those virtual machines that still need to work at night, the virtual machines marked as "non-idle state" can be "hot migrated" to other business hosts; because the desktop virtualization system of the embodiment of the present invention The number of such virtual machines is small, and "hot migration" will not cause network traffic congestion.

The method of the embodiment of the present invention will be introduced in detail below in combination with practical applications:

Please refer to FIG. 6 , which is a schematic flow diagram of another energy-saving method in a desktop virtualization environment according to an embodiment of the present invention, which can be applied in an office environment of a desktop virtualization system to trigger a virtual machine standby process during the user's off-duty time period scenario, as shown in Figure 6, the method includes:

S601, the process starts;

In the initial state, the virtual machine on the service host runs normally.

S602, the user accesses the desktop of the virtual machine through TC (Thin Client, thin client) or SC (Software Client, software terminal), and manually executes the idle operation of the virtual machine;

It should be noted that the virtual machine may provide an "idle" operation option or menu for the user to actively perform the "idle" operation of the virtual machine;

S603. The probe embedded in the virtual machine of the service host collects the hardware usage information of the virtual machine and the user operation information of the virtual machine, and transmits the hardware usage information and user operation information of the virtual machine to the VMM of the service host;

In this embodiment, the hardware usage information of the virtual machine represents hardware events occurring on the virtual machine, and the hardware events include mouse events, keyboard events, network I/O events and disk I/O events, etc., the virtual machine The user operation information of the machine represents an operation event occurring on the virtual machine, and the operation event is an "idle" operation event performed by the user on the virtual machine.

It should be noted that there is no clear sequence relationship between S602 and S603 here. It should be understood that after the virtual machine is started, the collection action continues;

S604. The VMM of the service host determines the status of the virtual machine on the VMM layer according to the virtual machine hardware usage information and virtual machine user operation information combined with the virtual machine idle policy delivered by the management server, and notifies the management server of the status information of the virtual machine; , the virtual machine status information indicates that the corresponding virtual machine is in the "idle state" or "non-idle state";

S605. The management server monitors whether all virtual machines hosted on the service host are in an idle state, and when the obtained virtual machine status information indicates that a few virtual machines hosted on the service host are in a "non-idle state", go to step S606; otherwise , when the obtained virtual machine status information indicates that all virtual machines hosted on the service host are in the "idle state", go to step S609;

It should be noted that the monitoring here may be regular monitoring, periodic monitoring, or real-time monitoring.

S607. Determine whether the virtual machine hosted on the physical host in the "non-idle state" satisfies the virtual machine migration policy, and if not, go to step S605; otherwise, go to step S607;

The virtual machine migration strategy includes: virtual machine migration time zone, virtual machine migration judgment conditions, and migration target business host selection conditions; for the introduction of the specific content of the virtual machine migration strategy, please refer to the foregoing embodiments, and will not be repeated here :

For example: if the current time period is within the virtual machine migration time period, the number of virtual machines in the "non-idle state" hosted on the physical host is less than 2 and the total number of virtual machines hosted on the physical host is greater than the preset average virtual machine number, and there is another physical host in the desktop virtualization system that has the highest proportion of virtual machines in the non-idle state and the lowest resource utilization, then the virtual machine in the "non-idle state" hosted on the physical host meets the virtual machine migration strategy; otherwise , the virtual machine in the "non-idle state" hosted on the physical host does not meet the virtual machine migration policy.

S607. The management server sends a migration command to the VMM of the physical host;

Wherein, the migration command includes the ID of the virtual machine to be migrated and the information of the destination host of the migration. The virtual machine identification ID here is a unique string sequence assigned by VMM to each virtual machine, which is used to uniquely identify the virtual machine; for example, it is called uuid under the Xen architecture; the destination host information for migration can be specifically the destination business host IP address.

S608, the VMM of the physical host responds to the migration command, and migrates the virtual machine in the "non-idle state" hosted on the physical host (that is, the virtual machine corresponding to the virtual machine ID) to the proportion of the virtual machine in the "non-idle state" on the destination service host with the highest and lowest resource utilization (that is, the destination service host corresponding to the destination service host information).

S609. The management server issues a standby command to the VMM of the service host;

S610. After receiving the standby command, the VMM of the service host sends a corresponding standby command to the processor of the hardware layer of the service host, so as to control and make the physical host enter the standby state of suspending memory;

S611. The physical host enters a standby state of suspend-to-memory;

The standby state here means that the processor of the business host is running in ACPI-S3 mode. Except for the memory that remains active, other devices such as the processor, hard disk, and network card are powered off or run on weak power;

S612, the process ends.

To sum up, the embodiment of the present invention provides an energy-saving method in a desktop virtualization environment. A "probe" is set in the virtual machine, and the virtual machine hardware usage information and/or virtual machine user operations are collected through the "probe". Information, according to the hardware usage information and/or user operation information and virtual machine idle policy, determine the current state of the virtual machine, and output virtual machine state information, wherein the state includes "idle state" or "non-idle state" and after receiving the standby command issued, send the standby command to the hardware layer to control and make the physical host enter the ACPI-S3 state; it can be seen that by the scheme of the embodiment of the present invention, the usage of the virtual machine can be identified , so that, for example, after the enterprise is off work, some business hosts can be kept on standby to save energy. At the same time, because the business hosts in the ACPI-S3 state are powered by memory, the memory status information is complete; the business hosts are woken up and run directly from the suspended memory. , and quickly return to the state before standby, so while achieving energy saving effects, avoid network traffic congestion caused by "hot migration" of a large number of virtual machines or loading of a large number of virtual machines when the enterprise is at work, thereby ensuring the stability and reliability of virtual machines and high performance, improving end-user experience satisfaction.

Further, through the solution of the embodiment of the present invention, the migration operation is performed on only a small number of virtual machines in the "not idle state" on the service hosts, so that these "not idle state" virtual machines are gathered on a small number of service hosts, so that more A large number of service hosts are on standby to save energy; since the number of such virtual machines in the desktop virtualization system of the embodiment of the present invention is small, "migration" will not cause network traffic congestion.

Please refer to FIG. 7 , which is a schematic flow chart of another energy-saving method in a desktop virtualization environment according to an embodiment of the present invention, which can be applied in an office environment of a desktop virtualization system to trigger a virtual machine wake-up process during the user's working hours. scenario, as shown in Figure 7, the method includes:

S701, the process starts;

S702. During working hours, when the user needs to use the virtual machine, he first logs in to the terminal, and the terminal is connected to the management server.

The terminal form can be a web page or a client program.

S703. The management server checks the status of the virtual machine; when the virtual machine is in an idle state, go to step S704; otherwise, go to step S707;

S704. The management server checks the status of the service host hosting the virtual machine. When the service host hosting the virtual machine is on standby, go to step S705; otherwise, go to step S706;

S705. The management server sends a wake-up command to the hardware layer of the service host to wake up the service host; the wake-up command is a network card wake-up command;

The initiator sends a data packet of a specific format to the host to be awakened, and the network card of the host to be awakened has special circuit processing logic to process the data packet, and then sends a power-on signal through a dedicated connection to notify the motherboard to start up. This technology requires the initiator and the awakened to be in the same subnet, and the initiator needs to know the MAC address of the awakened.

S706. After waking up, the service host directly runs the suspended memory, and activates a corresponding virtual machine of the host on itself, so that the virtual machine enters a manipulable state.

S707. Log in to the virtual machine.

S708, the process ends.

In summary, the embodiment of the present invention provides an energy-saving method in a desktop virtualization environment. Since the service host in the ACPI-S3 state is powered by its memory, the memory status information is complete; The memory runs and quickly returns to the state before standby, so while achieving energy saving effects, it avoids network traffic congestion caused by "hot migration" of a large number of virtual machines or loading of a large number of virtual machines when the enterprise goes to work, thereby ensuring the virtual machine. Stable, reliable and high-performance, improving end-user experience satisfaction.

It should be noted that the terms first and second used in the description of the foregoing embodiments do not mean to limit the sequence, but are only for convenience of distinction.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be realized by instructing related hardware (such as a processor) through a computer program, and the program can be stored in a computer-readable memory In the medium, when the program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, the storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM) or a random access memory (Random Access Memory, RAM), etc.

The foregoing is only a specific embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, some improvements and modifications can also be made without departing from the principle of the present invention. It should be regarded as the protection scope of the present invention.

Claims (9)

1. desktop virtual system, comprise many physical hosts and management server, wherein every physical host comprises hardware layer, operate in the virtual machine monitoring unit VMM on the described hardware layer, operate in a plurality of virtual machines on the described virtual machine monitoring unit VMM, and a plurality of probes, described a plurality of probes are corresponding with described a plurality of virtual machines, wherein:

Described probe specifically is embedded in the corresponding virtual machine, is used for collecting the hardware use information of corresponding virtual machine and/or the user's operation information of corresponding virtual machine; Perhaps, described probe comprises the front end probe that is embedded in the corresponding virtual machine and the rear end probe that is deployed among the described virtual machine monitoring unit VMM, and described front end probe is used for collecting the user's operation information of corresponding virtual machine; Described rear end probe is used for collecting the hardware use information of corresponding virtual machine; Wherein, the hardware use information of described virtual machine is represented the hardware event that takes place on the described virtual machine, and described hardware event comprises: one or more in mouse event, KeyEvent, network I/O I/O event and the disk I/O I/O event; The user's operation information of described virtual machine is represented the Action Events that takes place on the described virtual machine, and described Action Events comprises: one or more in idle operation, standby operation, sleep operation and the opening operation that the user carries out virtual machine;

Described virtual machine monitoring unit VMM is used for: belong to the classification of can leaving unused when described hardware uses the virtual machine of information and/or user's operation information association, and current slot is in described virtual machine section standby time, and hardware event does not take place in this virtual machine in threshold time, perhaps, the hardware event occurrence frequency is less than the first threshold frequency, perhaps, this virtual machine is performed idle operation, perhaps the operation and this operation that are performed in the scheduled operation scope of this virtual machine can substitute with idle operation, then determine the current idle state that is in of this virtual machine, and represent the virtual machine state information of idle state to described management server output; Described scheduled operation scope comprises standby operation, sleep operation or opening operation; Perhaps, be used for: belong to the classification of to leave unused when described hardware uses the virtual machine of information and/or user's operation information association, and operation and this operation that this virtual machine is performed in the scheduled operation scope can not substitute with idle operation, then determine the current non-idle state that is in of this virtual machine, and represent the virtual machine state information of non-idle state to described management server output; Perhaps, be used for: belong to the classification of to leave unused when described hardware uses the virtual machine of information and/or user's operation information association, then determine the current non-idle state that is in of this virtual machine, to the virtual machine state information of described management server output expression non-idle state; Wherein, described virtual machine section standby time refers to the time section that virtual machine can leave unused; Whether the idle classification of described virtual machine refers to virtual machine and can leave unused;

Described management server is used for: when all virtual machines of the host of institute are idle state on the virtual machine state information that obtains is represented by host's physical host, sent standby command to described by host's physical host;

Described virtual machine monitoring unit VMM is further used for: receive the standby command that described management server issues, and send corresponding standby command to described hardware layer, to control and to make this physical host enter the holding state that suspends to internal memory.

2. system according to claim 1, it is characterized in that, described management server is further used for: according to described virtual machine state information and the virtual machine (vm) migration strategy that obtains, need to determine the virtual machine that is in non-idle state and the migration target physical main frame of migration, and sending the migration order to the virtual machine host's of needs migrations physical host, described migration order comprises that the virtual machine of needs migration identifies the destination host information of ID and migration;

The virtual machine monitoring unit VMM of described physical host is further used for: according to described migration order, described virtual machine sign ID corresponding virtual machine is moved on the target physical main frame of described destination host information correspondence.

3. system according to claim 1 is characterized in that, described management server is further used for: the response user logins the operational order of virtual machine, sends to described virtual machine host's physical host and wakes instruction up;

The hardware layer of described physical host is used for: respond the described instruction that wakes up, the internal memory that operation is hung up activates the corresponding virtual machine, but makes this virtual machine enter operation state.

4. physical host, it is characterized in that, described physical host comprises: hardware layer, operate in the virtual machine monitoring unit VMM on the described hardware layer, operate in a plurality of virtual machines on the described virtual machine monitoring unit VMM, and a plurality of probes, described a plurality of probes are corresponding with described a plurality of virtual machines, wherein:

Described probe specifically is embedded in the corresponding virtual machine, is used for collecting the hardware use information of corresponding virtual machine and/or the user's operation information of corresponding virtual machine; Perhaps, described probe comprises the front end probe that is embedded in the corresponding virtual machine and the rear end probe that is deployed among the described virtual machine monitoring unit VMM, and described front end probe is used for collecting the user's operation information of corresponding virtual machine; Described rear end probe is used for collecting the hardware use information of corresponding virtual machine; Wherein, the hardware use information of described virtual machine is represented the hardware event that takes place on the described virtual machine, and described hardware event comprises: one or more in mouse event, KeyEvent, network I/O I/O event and the disk I/O I/O event; The user's operation information of described virtual machine is represented the Action Events that takes place on the described virtual machine, and described Action Events comprises: one or more in idle operation, standby operation, sleep operation and the opening operation that the user carries out virtual machine;

Described virtual machine monitoring unit VMM is used for: belong to the classification of can leaving unused when described hardware uses the virtual machine of information and/or user's operation information association, and current slot is in described virtual machine section standby time, and hardware event does not take place in this virtual machine in threshold time, perhaps, the hardware event occurrence frequency is less than the first threshold frequency, perhaps, this virtual machine is performed idle operation, perhaps the operation and this operation that are performed in the scheduled operation scope of this virtual machine can substitute with idle operation, then determine the current idle state that is in of this virtual machine, and represent the virtual machine state information of idle state to management server output; And, receive standby command, send corresponding standby command to described hardware layer, described scheduled operation scope comprises standby operation, sleep operation or opening operation; Perhaps, be used for: belong to the classification of to leave unused when described hardware uses the virtual machine of information and/or user's operation information association, and operation and this operation that this virtual machine is performed in the scheduled operation scope can not substitute with idle operation, then determine the current non-idle state that is in of this virtual machine, and to the virtual machine state information of management server output expression non-idle state, and, receive standby command, send corresponding standby command to described hardware layer; Perhaps, be used for: belong to the classification of to leave unused when described hardware uses the virtual machine of information and/or user's operation information association, then determine the current non-idle state that is in of this virtual machine, virtual machine state information to management server output expression non-idle state, and, receive standby command, send corresponding standby command to described hardware layer; Wherein, described virtual machine section standby time refers to the time section that virtual machine can leave unused; Whether the idle classification of described virtual machine refers to virtual machine and can leave unused;

Described hardware layer is used for: respond described standby command, control and make this physical host enter the holding state that suspends to internal memory.

5. physical host according to claim 4, it is characterized in that, described virtual machine monitoring unit VMM is further used for: receive described management server and send the migration order, the virtual machine sign ID corresponding virtual machine that comprises in the described migration order is moved on the target physical main frame of the destination host information correspondence that comprises in the described migration order.

6. physical host according to claim 4 is characterized in that, described hardware layer is further used for: receive the instruction that wakes up of described management server transmission, wake instruction up in response to this, the internal memory that operation is hung up activates the corresponding virtual machine, but makes this virtual machine enter operation state.

7. the power-economizing method under the desktop virtual environment, it is characterized in that, be applied to comprise hardware layer, operate in the virtual machine monitoring unit VMM on the described hardware layer, operate in a plurality of virtual machines on the described virtual machine monitoring unit VMM, and the physical host of a plurality of probes, described a plurality of probes are corresponding with described a plurality of virtual machines, and this method comprises:

Collect the hardware use information of corresponding virtual machine and/or the user's operation information of corresponding virtual machine by the probe that is embedded in the corresponding virtual machine; Perhaps, described probe comprises by being embedded in front end probe in the corresponding virtual machine and collects the user's operation information of corresponding virtual machine and collect the hardware of corresponding virtual machine and use information by being deployed in rear end probe among the described virtual machine monitoring unit VMM; Wherein, the hardware use information of described virtual machine is represented the hardware event that takes place on the described virtual machine, and described hardware event comprises: one or more in mouse event, KeyEvent, network I/O I/O event and the disk I/O I/O event; The user's operation information of described virtual machine is represented the Action Events that takes place on the described virtual machine, and described Action Events comprises: one or more in idle operation, standby operation, sleep operation and the opening operation that the user carries out virtual machine;

When using the virtual machine of information and/or user's operation information association, described hardware belongs to the classification of to leave unused, and current slot is in described virtual machine section standby time, and hardware event does not take place in this virtual machine in threshold time, perhaps, the hardware event occurrence frequency is less than the first threshold frequency, perhaps, this virtual machine is performed idle operation, perhaps the operation and this operation that are performed in the scheduled operation scope of this virtual machine can substitute with idle operation, then determine the current idle state that is in of this virtual machine, and represent the virtual machine state information of idle state to management server output; Described scheduled operation scope comprises standby operation, sleep operation or opening operation;

Perhaps, when using the virtual machine of information and/or user's operation information association, described hardware belongs to the classification of to leave unused, and operation and this operation that this virtual machine is performed in the scheduled operation scope can not substitute with idle operation, then determine the current non-idle state that is in of this virtual machine, and represent the virtual machine state information of non-idle state to management server output;

Perhaps, belong to the classification of to leave unused when described hardware uses the virtual machine of information and/or user's operation information association, then determine the current non-idle state that is in of this virtual machine, to the virtual machine state information of management server output expression non-idle state; Wherein, described virtual machine section standby time refers to the time section that virtual machine can leave unused; Whether the idle classification of described virtual machine refers to virtual machine and can leave unused;

Receive the standby command that described management server sends, and send corresponding standby command to hardware layer, to control and to make this virtual machine host's physical host enter the holding state that suspends to internal memory.

8. method according to claim 7 is characterized in that, described method further comprises:

Receive the migration order that described management server sends, the virtual machine sign ID corresponding virtual machine that comprises in the described migration order is moved on the target physical main frame of the destination host information correspondence that comprises in the described migration order.

9. method according to claim 7 is characterized in that, described method further comprises: receive the instruction that wakes up of described management server transmission, wake instruction up in response to this, the internal memory that operation is hung up activates the corresponding virtual machine, but makes this virtual machine enter operation state.

Images