JP4883492B2 - Virtual machine management system, computer, and program - Google Patents

Description

  The present invention relates to management of a virtual computer built on a computer.

  There is a technique for constructing a virtual computer (hereinafter referred to as “virtual machine”) by software on a computer using hardware resources of the computer. By constructing a plurality of virtual machines, a plurality of operating systems (OS) can be handled in parallel on one computer.

  In the above technique, an OS assigned to a virtual machine is generally called a guest OS. The software configuration connecting the virtual machine that executes the guest OS and the hardware of the computer differs depending on the standard of the virtual machine. For example, in a standard represented by “VMware GSX (registered trademark)” and “Virtual PC (registered trademark)”, an OS such as Windows (registered trademark) or Linux (registered trademark) operating on hardware is used as a host OS. Virtual software is interposed between the host OS and the virtual machine group.

  Further, there is a standard in which a control unit such as a kernel is interposed between a virtual machine group and hardware so that each guest OS of the virtual machine group accesses the hardware via a specific virtual machine. An example of this standard is “VMware ESX Server (registered trademark)”. In the same type of standard, such as “Xen (registered trademark)”, the OS executed by the specific virtual machine described above may be referred to as a host OS.

  When building a virtual machine on a server of a communication system or a computer system, an administrator manages the virtual machine of each server in a batch. As a management method, for example, there is a method described in Patent Document 1 described later. The method of this document is to handle commands executed by a plurality of guest OSs by a hypervisor in an integrated manner. Thereby, even when the command standard is different for each guest OS, the handling of the command becomes simple.

As another management method, for example, there is a method of periodically accessing a server from a management apparatus and acquiring information about a virtual machine. Another method is to install an agent program for monitoring each virtual machine on the server. In this method, for example, by setting an agent program in the guest OS of each virtual machine, the state of the virtual machine is reported to the management side.
JP 2005-135137 A

  In the above-described method of inquiring the state of the virtual machine to the server, frequent inquiries to the server are required to quickly detect the state change. Therefore, the load on the network or server side may increase.

  On the other hand, according to the method using the agent program, the state of the virtual machine is spontaneously reported from the server side. Therefore, an increase in the network load can be suppressed as compared with the method of periodically inquiring from the management side. However, an administrator or the like needs to install an agent program for each guest OS, and further requires setting work of parameters and the like for each installed agent. For this reason, particularly when managing a large-scale computer system or managing a large number of virtual machines, there arises a problem that it is difficult to reduce the management cost.

  SUMMARY An advantage of some aspects of the invention is that it provides a technique for preventing an increase in cost for managing a virtual machine.

A virtual machine management system according to the present invention includes a management device and a computer that are communicably connected to each other, and the computer includes a first virtual machine that executes a backbone OS and hardware of the computer via the backbone OS. A virtual machine group including a second virtual machine that executes a guest OS that accesses the hardware, an agent unit set in the core OS, and an interface for controlling the virtual machine group is provided to the agent unit. And a virtual machine control unit that relays between the hardware of the computer and the backbone OS, and the management device transmits a command to be executed by the agent unit with respect to the second virtual machine to the computer. And the agent unit provides an interface corresponding to a command from the management device from the virtual machine control unit. Possess means for executing the instruction by finely out and call interface, and means for reporting the state of the second virtual machine to the management device, the interface, the agent unit creates a copy of itself The agent unit further includes the replication unit to copy the agent unit to the guest OS when the command from the management apparatus represents setting the replication of the agent unit to the guest OS. It has the means to set, It is characterized by the above-mentioned .

A computer according to the present invention is a computer that is communicably connected to a management apparatus, and executes a first virtual machine that executes a backbone OS and a guest OS that accesses the hardware of the computer via the backbone OS A virtual machine group including the second virtual machine, an agent unit set in the core OS, an interface for controlling the virtual machine group to the agent unit, and hardware of the computer, A virtual machine control unit that relays between core OSs, and the agent unit has an interface corresponding to the command when the agent unit receives a command to be executed by the agent unit with respect to the second virtual machine from the management device. Means for executing the instruction by the interface called and called from the virtual machine control unit; The state of the second virtual machine to have a means for reporting to the management device, wherein the interface includes a replication interface for the agent unit creates a copy of itself, the agent unit is further the management device When the command from the above indicates that the copy of the agent unit is set in the guest OS, the copy interface has means for setting its copy in the guest OS .

  According to the present invention, in order to manage a virtual machine group of computers, it is not necessary to install an agent program in each guest OS of the virtual machine group. Thereby, management cost can be held down.

  FIG. 1 shows a system configuration of an embodiment of the present invention. The system 101 includes a management apparatus 10 and a server 20 that are communicably connected via a network 80. The management device 10 is a computer terminal operated by an administrator or the like, and the server 20 is a computer on which a virtual machine to be managed is constructed. Although there is one server 20 in the illustrated configuration, a plurality of servers 20 may exist.

  The management apparatus 10 is installed with a management program 11 for communicating with an agent program 71 of the server 20 described later. The management program 11 corresponds to the management unit of the management device in the present invention. The management program 11 displays information reported from the agent program 71 using a GUI (Graphical User Interface). In addition, the management program 11 has a function of transmitting, to the server 20, policy information to be described later that defines a report to be performed by the agent program 71.

  The standard of the virtual machine applied to the server 20 is that a control unit is interposed between the virtual machine group and the hardware as in the above-mentioned “VMware ESX Server (registered trademark)” and “Xen (registered trademark)”. The standard is that the guest OS accesses the hardware through the OS of a specific machine. In this embodiment, an OS executed by the specific virtual machine is referred to as a host OS, and is handled as corresponding to the core OS in the present invention.

  The server 20 includes a virtual machine control unit 30 positioned in the same manner as the above-described control unit in terms of software. On the virtual machine control unit 30, a virtual machine A41, a virtual machine B42, and a virtual machine C43 are constructed as a virtual machine group. In the configuration of FIG. 1, the OS of the virtual machine A 41 is the host OS 51, and the OSs of the virtual machine B 42 and the virtual machine C 43 are the guest OS 61 and the guest OS 62. Therefore, the virtual machine A41 corresponds to the first virtual machine in the present invention, and the virtual machine B42 and the virtual machine C43 correspond to the second virtual machine.

  The virtual machine control unit 30 is interposed between the virtual machine group and the hardware 90, and relays data exchanged between them. When the virtual machine B42 or the virtual machine C43 of the guest OS accesses the hardware 90, they once access the virtual machine A41 of the host OS via the virtual machine control unit 30. Then, the virtual machine A41 accesses the hardware 90 via the virtual machine control unit 30.

  The server 20 includes an agent program 71 for communicating with the management program 11 of the management apparatus 10 described above. The agent program 71 corresponds to the agent unit in the present invention, and is installed in the host OS 51 by an administrator or the like in advance.

  The virtual machine control unit 30 includes an API (Application Program Interface) 300 that is an interface for the agent program 71 of the host OS 51 to control the virtual machines (42, 43) of the guest OSs. The API 300 provides the agent program 71 with functions for creating a new virtual machine, deleting it, starting it, stopping it, and monitoring the load state, and for communicating with the management apparatus 10. The agent program 71 calls a necessary one of these functions from the virtual machine control unit 30.

  FIG. 2 shows detailed configurations of the agent program 71 and the virtual machine control unit 30 in the server 20. The agent program 71 of the host OS 51 includes, as its functional configuration, a network input / output unit 711, a guest OS control unit 712, a policy control unit 713, a policy storage unit 714, a load monitoring control unit 715, and a failure monitoring control unit 716. . The virtual machine control unit 30 includes a network input / output unit 301, a guest OS management unit 302, and a guest OS monitoring unit 303 as the API 300 described above.

  When accessing the management program 11 on the management apparatus 10 side, the agent program 71 first calls the network input / output unit 301 from the virtual machine control unit 30 by the network input / output unit 711. The called network input / output means 301 connects to the management device 10 using a communication circuit (not shown) of the hardware 90 and communicates with the management program 11. In the case of access from the management program 11 to the agent program 71, the procedure is the reverse of the above.

  When the guest OS control unit 712 receives a command for creating / deleting / starting / stopping a guest OS from the management program 11, the guest OS control unit 712 calls the guest OS management unit 302 and executes the commanded process.

  The load monitoring control unit 715 and the failure monitoring control unit 716 use the guest OS monitoring unit 303 to monitor the state change of the virtual machine (42, 43) that executes the guest OS (61, 62). In the present embodiment, whether the load monitoring control unit 715 monitors a change in the load state based on the CPU usage rate or the memory usage amount of the virtual machine (42, 43) and reports whether the change is reported to the management apparatus 10 or not. The failure monitoring control unit 716 makes a determination. The failure monitoring control unit 716 determines whether or not it is necessary to report to the management apparatus 10 using the policy information that defines the criteria for this determination.

  In the present embodiment, as described above, the policy information is provided from the management apparatus 10 side. When the server 20 receives the policy information from the management apparatus 10, the policy control unit 713 stores it in the policy storage unit 714.

  The operation of the system 101 having the above configuration will be described. First, operations of the agent program 71 and the management program 11 when a new virtual machine is created in the server 20 will be described with reference to the flowchart shown in FIG.

  When an instruction to construct a new virtual machine in the server 20 is input to the management apparatus 10, the management program 11 transmits information indicating the instruction to the server 20 (step S1). This information describes the type of guest OS assigned to the virtual machine to be created this time.

  When the server 20 receives the command information (step S2), the agent program 71 calls the guest OS management unit 302 from the API 300 by the guest OS control unit 712. Then, using the called guest OS management unit 302, a new virtual machine and a guest OS to be executed by the virtual machine are created on the virtual machine control unit 30 (step S3).

  Further, the agent program 71 calls the guest OS monitoring unit 303 from the API 300 by the load monitoring control unit 715, and issues an event when the load state of the new virtual machine changes, so that the guest OS monitoring unit 303 is issued. Is set (step S4). As a result, the newly created virtual machine is added to the monitoring target of the agent program 71.

  Policy information registration will be described with reference to the flowchart shown in FIG. As described above, the policy information defines the standard for reporting from the agent program 71 to the management program 11.

  The management program 11 of the management device 10 transmits policy information to be registered in the agent program 71 and a registration command for the policy information to the server 20 (step S11). The policy information describes what load state the virtual machine (61, 62) of the guest OS should report to the management apparatus 10 side. Specifically, for example, the upper limit value of the load such as the CPU usage rate and the memory usage rate described above, and the fact that the load to be monitored should be reported when the load exceeds the upper limit value are described.

  When the server 20 receives the above policy information (step S12), the agent program 71 stores it in the policy storage unit 714 by the policy control unit 713 (step S13). At this time, an identifier or the like is added and stored so that the current policy information is related to the guest OS monitoring unit 303 or related to monitoring of the load state.

  The operation for monitoring the load state of the virtual machines (61, 62) in the server 20 will be described with reference to the flowchart shown in FIG. The monitoring target is a virtual machine (FIG. 3: S4) for which the load monitoring control unit 715 has set the issuance of an event to the guest OS monitoring unit 303.

  The guest OS monitoring unit 303 called by the load monitoring control unit 715 monitors the load state of each virtual machine (61, 62). When an increase in load is detected for a certain virtual machine (step S21: Yes), an event is issued to the agent program 71 (step S22). The issued event is detected by the failure monitoring control unit 716.

  When the failure monitoring control unit 716 recognizes that the load of a certain virtual machine has increased due to the above event, the failure monitoring control unit 716 acquires policy information regarding the load state from the policy control unit 713 (step S23). Then, the upper limit value indicated by the acquired policy information is compared with the load value of the virtual machine corresponding to the current event. As a result of the comparison, if the load value does not exceed the upper limit (step S24: No), it is determined not to report to the management apparatus 10 (step S25), and the load state of each virtual machine is continuously monitored.

  On the other hand, when the load value exceeds the upper limit (step S24: Yes), it is determined that a report to the management apparatus 10 should be performed (step S26). In this case, the agent program 71 calls the network input / output unit 301 of the API 300, and transmits information indicating that the virtual machine is in an overload state to the management apparatus 10 (step S27). As a result, it is reported to the management apparatus 10 that a certain virtual machine is in an overload state, and that fact is displayed on the screen of the management apparatus 10.

  According to the present embodiment, in managing the virtual machine group of the server 20, the work of installing the agent program 71 in each of the guest OSs (61, 62) is unnecessary. Therefore, the management cost can be suppressed. Further, since the agent program 71 of the host OS 51 actively acquires the information of each guest OS (61, 62) and reports it to the management apparatus 10 as necessary, the burden on the management apparatus 10 side is reduced.

  FIG. 6 shows a system configuration of another embodiment of the present invention. The difference between the system 102 of the present embodiment and the above-described system 101 (FIG. 1) is that a duplication unit 717 is added to the agent program 71 of the system 102, and an agent duplication unit 304 is added to the API (300) of the virtual machine control unit 30. Is included. The duplication unit 717 calls the agent duplication unit 304 to set duplication of the agent program 71 in the guest OS 61 and the guest OS 62.

  The operation of the present embodiment will be described along the flowchart shown in FIG. When a command for copying the agent program 71 of the server 20 to the guest OS is input to the management device 10, the management program 11 transmits the command to the agent program 71 of the server 20 (step S31).

  When the server 20 receives the instruction information (step S32), the agent program 71 calls the agent duplication unit 304 from the API 300 by the duplication unit 717. Then, it creates a copy (72, 73) of itself (step S33), and sets it in the guest OS 61 and guest OS 62 (step S34). Thereby, as shown in FIG. 6, an agent program 72 is set in the guest OS 61, and an agent program 73 is set in the guest OS 62.

  Each guest OS (61, 62) recognizes and starts the set agent program (72, 73) (step S35). The activated agent program (72, 73) calls the guest OS monitoring means 303 of the API 300, and changes the setting of the event issuance destination regarding the change in the load state from the agent program 71 of the host OS 51 to its own address (72, 73). (Step S36). Thereby, an event related to a change in the load state of each virtual machine (42, 43) is notified to the agent program (72, 73) of the virtual machine.

  Thereafter, the agent programs (72, 73) copied to the guest OSs (61, 62) make their own judgments regarding load monitoring of their machines and reporting to the management apparatus 10, but with the management apparatus 10 This communication is performed via the virtual machine A41 of the host OS 51.

  According to the present embodiment, the load management of the guest OS (61, 62) is performed by the agent program (72, 73) copied to each, so that the load can be distributed accordingly. In addition, since the agent program 71 of the host OS 51 automatically installs the agent program for each guest OS, installation work for each guest OS by an administrator or the like is not necessary.

  The implementation of the present invention is not limited to the above-described embodiment, and various modifications are possible. For example, the monitoring contents of the virtual machine are not limited to changes in the load state, and can be appropriately set, for example, whether or not a restart or stall has occurred, or whether or not the creation of a new virtual machine has been completed without any problem .

  Further, although the policy information is provided from the management apparatus 10 in the above embodiment, it may be set when the agent program 71 is installed in the host OS 51. Then, when the setting value needs to be updated, new policy information is transmitted from the management apparatus 10 to the server 20.

  Furthermore, in the above embodiment, the failure monitoring control unit 716 determines whether to report to the management apparatus 10 that the status of the virtual machine has changed, but this determination is omitted, that is, every time a status change occurs. , You may make it report.

  The present invention is not limited to being implemented as a system or apparatus, and may be implemented as a computer program based on the software configuration of the server 20 and a recording medium storing the program.

1 is a configuration diagram of a system in an embodiment of the present invention. It is a block diagram regarding the structure of the server in embodiment of this invention. It is a flowchart regarding the new creation of the virtual machine in the embodiment of the present invention. It is a flowchart regarding registration of policy information in the embodiment of the present invention. It is a flowchart regarding the monitoring of the virtual machine in the embodiment of the present invention. It is a block diagram of the system in other embodiment of this invention. It is a flowchart regarding the duplication process of the agent program in other embodiment of this invention.

Explanation of symbols

101,102 system
80 network
10: Management device, 11: Management program
20: Server, 30: Virtual machine control unit, 300: API, 41, 42, 43: Virtual machines A, B, C, 51: Host OS, 61, 62: Guest OS, 71: Agent program, 90: Hardware

Claims (11)

A management device and a computer connected to each other so as to communicate with each other;
The computer includes a virtual machine group including a first virtual machine that executes a backbone OS and a second virtual machine that executes a guest OS that accesses the hardware of the computer via the backbone OS; A virtual machine control unit that provides an interface for controlling the virtual machine group to the agent unit and relays between the hardware of the computer and the backbone OS;
The management device includes a management unit that transmits an instruction to be executed by the agent unit to the computer with respect to the second virtual machine,
The agent unit calls an interface corresponding to a command from the management device from the virtual machine control unit, and executes the command through the called interface, and reports the status of the second virtual machine to the management device. have a means for,
The interface includes a replication interface for the agent unit to create a copy of itself, and the agent unit further indicates that a command from the management apparatus sets the replication of the agent unit to the guest OS. In this case, the virtual machine management system further comprises means for setting its own copy in the guest OS by the copy interface .   The interface includes a monitoring interface for the agent unit to recognize a state change of the second virtual machine, and the agent unit further receives a command from the management device to change the state of the second virtual machine. Means for recognizing a change in the state of the second virtual machine by the monitoring interface, and a means for determining whether or not to report the recognized change in state to the management device based on policy information. The virtual machine management system according to claim 1.   The monitoring interface is an interface for the agent unit to recognize a change in the load of the second virtual machine due to execution of the guest OS, and the agent unit is configured to detect the second virtual machine by the monitoring interface. When recognizing that the load has changed, the load value of the second virtual machine is compared with the set value indicated by the policy information, and if the load value exceeds the set value, the load of the second virtual machine The virtual machine management system according to claim 2, wherein the virtual machine management system determines to report to the management apparatus that the increase has occurred.   The interface includes a management interface for the agent unit to newly create a second virtual machine in the virtual machine group, and the agent unit further includes a command from the management device for a new second virtual machine. 4. The method according to claim 1, further comprising means for adding a second virtual machine and a guest OS to be executed by the machine to the virtual machine group by the management interface when representing creation. 5. The described virtual machine management system.   The management unit of the management device has means for transmitting the policy information to the computer, and the agent unit has means for recording policy information from the management device in association with the monitoring interface. The virtual machine management system according to any one of claims 2 to 4. A computer that is communicably connected to a management device,
A virtual machine group including a first virtual machine that executes a basic OS and a second virtual machine that executes a guest OS that accesses the computer hardware via the basic OS, and is set in the basic OS An agent unit, and a virtual machine control unit that provides an interface for controlling the virtual machine group to the agent unit and relays between the hardware of the computer and the backbone OS,
The agent unit calls an interface corresponding to the command from the virtual machine control unit when the command to be executed by the agent unit with respect to the second virtual machine is received from the management apparatus, and the command is called by the interface that is called. We possess means for executing, and means for reporting the state of the second virtual machine to the management device,
The interface includes a replication interface for the agent unit to create a copy of itself, and the agent unit further indicates that a command from the management apparatus sets the replication of the agent unit to the guest OS. In this case, the computer has means for setting its own copy in the guest OS by the copy interface . The interface includes a monitoring interface for the agent unit to recognize a state change of the second virtual machine, and the agent unit further receives a command from the management device to change the state of the second virtual machine. Means for recognizing a change in the state of the second virtual machine by the monitoring interface, and a means for determining whether or not to report the recognized change in state to the management device based on policy information. The computer according to claim 6 . The monitoring interface is an interface for the agent unit to recognize a change in the load of the second virtual machine due to execution of the guest OS, and the agent unit is configured to detect the second virtual machine by the monitoring interface. When recognizing that the load has changed, the load value of the second virtual machine is compared with the set value indicated by the policy information, and if the load value exceeds the set value, the load of the second virtual machine The computer according to claim 7 , wherein it is determined to report to the management apparatus that the increase has occurred. The interface includes a management interface for the agent unit to newly create a second virtual machine in the virtual machine group, and the agent unit further includes a command from the management device for a new second virtual machine. to any one of claims 6 to 8, characterized in that it comprises means for adding a guest OS to execute the second virtual machine and the machine by the management interface to the virtual machine group to represent the creation Listed calculator. The computer according to any one of claims 7 to 9 , wherein the agent unit includes means for recording the policy information in association with the monitoring interface when the policy information is received from the management device. . The computer,
A virtual machine group including a first virtual machine that executes a basic OS and a second virtual machine that executes a guest OS that accesses the computer hardware via the basic OS, and is set in the basic OS Providing an agent unit and an interface for controlling the virtual machine group to the agent unit and functioning as a virtual machine control unit that relays between the hardware of the computer and the backbone OS;
When the command to be executed by the agent unit with respect to the second virtual machine is received from the management device, the interface corresponding to the command is called from the virtual machine control unit and the command And a means for reporting the status of the second virtual machine to the management device,
The interface includes a replication interface for the agent unit to create a copy of itself, and the agent unit further indicates that a command from the management apparatus sets the replication of the agent unit to the guest OS. In this case, the copy interface causes the copy to function as means for setting the copy to the guest OS .

Images