JP2018055160A - Virtual machine management device, method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the failure in a shift of a virtual machine.SOLUTION: A virtual machine management device comprises: a detection unit for detecting an access pattern indicating regularity of increase/decrease of access number, for each of a plurality of virtual machines operating in different physical machines; a specification unit for specifying combinations with low correlation in the access patterns between the plurality of virtual machines; and a determination unit for determining the virtual machine as a shift object and the physical machine as a shift destination, on the basis of the specified combination and free capacity of resource of each physical machine, and determining a time zone where the access number in the virtual machine as the shift object is reduced as a shift scheduled date, on the basis of the access pattern.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a virtual machine management apparatus, method, and program, and more particularly, to a virtual machine management apparatus, method, and program for migrating an operating virtual machine.

  Cloud providers provide servers via the Internet using server virtualization technology that operates multiple virtual machines on physical machines. In some cases, a cloud business provider performs live migration of a virtual machine in order to effectively use its own server resources, maintain a physical machine, and reduce power consumption. Here, live migration refers to a technology for migrating a virtual machine to another physical machine without stopping the operation of the virtual machine on a certain physical machine.

  Patent Document 1 discloses a technique in which a data relay device holds a data transfer addressed to a virtual machine for which a non-convergence of live migration is detected. Patent Document 2 discloses a technique for arranging virtual computers so that hardware resources of physical computers do not become excessive or insufficient as much as possible.

JP 2013-047920 A International Publication No. 2012/053393

  However, Patent Document 1 has a problem that if the access to the migration source virtual machine is concentrated during the live migration, the amount of data to be suspended increases and the migration may fail. Further, in Patent Document 2, since the resource state before the migration is taken into consideration, the above-mentioned problem due to the fluctuation during the implementation cannot be solved.

  The present invention has been made to solve such problems, and an object of the present invention is to provide a virtual machine management apparatus, method, and program for reducing failure of virtual machine migration.

A virtual machine management device according to the first aspect of the present invention includes:
For each of a plurality of virtual machines running in different physical machines, a detection unit that detects an access pattern indicating regularity of increase / decrease in the number of accesses,
A specifying unit that specifies a combination having a low correlation of the access patterns among the plurality of virtual machines;
The migration target virtual machine and the migration destination physical machine are determined based on the identified combination and the resource free capacity of each physical machine, and the migration target virtual machine in the migration target virtual machine is determined based on the access pattern. A determination unit that determines a time zone in which the number of accesses decreases as a scheduled transfer date and time,
Is provided.

The virtual machine management method according to the second aspect of the present invention includes:
For each of a plurality of virtual machines running in different physical machines, an access pattern indicating regularity of increase / decrease in the number of accesses is detected,
Identifying a combination with a low correlation of the access patterns among the plurality of virtual machines;
Based on the identified combination and the free capacity of the resources of each physical machine, determine a virtual machine to be migrated and a physical machine to be migrated,
Based on the access pattern, a time zone in which the number of accesses in the migration target virtual machine decreases is determined as a migration scheduled date and time.

A virtual machine management program according to the third aspect of the present invention includes:
For each of a plurality of virtual machines running in different physical machines, processing for detecting an access pattern indicating regularity of increase / decrease in the number of accesses,
A process of identifying a combination having a low correlation of the access patterns among the plurality of virtual machines;
A process of determining a migration target virtual machine and a migration destination physical machine based on the identified combination and the free capacity of the resource of each physical machine;
Based on the access pattern, a process for determining a time zone in which the number of accesses in the migration target virtual machine decreases as a scheduled migration date and time,
Is executed on the computer.

The virtual machine management device according to the fourth aspect of the present invention is:
A detection unit that detects a virtual machine whose number of accesses is increasing among a plurality of virtual machines operating in a physical machine;
Based on the number of accesses in the detected virtual machine, an estimated time when the resource is insufficient is calculated, and based on the data amount and network speed of the virtual machine, the migration time required for migration between physical machines of the virtual machine is calculated. Calculate, calculate the limit access number at the time of migration of the virtual machine based on the migration required time, calculate the migration required time for virtual machines other than the detected virtual machine,
Based on the estimated time, the required migration time, and the limit access number, a determination unit that determines a virtual machine to be migrated among the plurality of virtual machines and a scheduled migration date and time,
Is provided.

A virtual machine management method according to the fifth aspect of the present invention includes:
Detects virtual machines whose access count is increasing among multiple virtual machines running in physical machines.
Calculate an estimated time of resource shortage based on the number of accesses in the detected virtual machine,
Based on the amount of data of the virtual machine and the network speed, calculate the time required for migration of the virtual machine between physical machines,
Calculate the limit number of accesses at the time of migration of the virtual machine based on the migration required time,
Calculate the migration time for virtual machines other than the detected virtual machine,
Based on the estimated time, the required migration time, and the limit access number, a virtual machine to be migrated among the plurality of virtual machines and a scheduled migration date and time are determined.

  According to the present invention, it is possible to provide a virtual machine management apparatus, method, and program for reducing failure of virtual machine migration.

It is a block diagram which shows the whole structure containing the virtual machine management apparatus concerning Embodiment 1 of this invention. It is a flowchart which shows the flow of the virtual machine migration management method concerning Embodiment 1 of this invention. It is a block diagram which shows the whole structure containing the virtual machine management apparatus concerning Embodiment 2 of this invention. It is a flowchart which shows the flow of the virtual machine transfer management method concerning Embodiment 2 of this invention. It is a block diagram which shows the whole structure containing the virtual machine management apparatus concerning Embodiment 3 of this invention. It is a block diagram which shows the internal structure of the virtual machine management apparatus concerning Embodiment 3 of this invention. It is a figure which shows the example of the virtual machine transfer management information concerning Embodiment 3 of this invention. It is a figure which shows the example of the matching table concerning Embodiment 3 of this invention. It is a figure which shows the example of the network load condition concerning Embodiment 3 of this invention. It is a flowchart which shows the flow of the registration process concerning Embodiment 3 of this invention. It is a flowchart which shows the flow of the pattern analysis process concerning Embodiment 3 of this invention. It is a flowchart which shows the flow of the user requirement input process concerning Embodiment 3 of this invention. It is a flowchart which shows the flow of the migration plan process concerning Embodiment 3 of this invention. It is a flowchart which shows the flow of the migration plan process concerning Embodiment 3 of this invention. It is a flowchart which shows the flow of the migration plan process concerning Embodiment 3 of this invention.

  Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings. In the drawings, the same elements are denoted by the same reference numerals, and redundant description will be omitted as necessary for the sake of clarity.

<Embodiment 1>
FIG. 1 is a block diagram showing an overall configuration including a virtual machine management apparatus 1000 according to the first embodiment of the present invention. The virtual machine management apparatus 1000 is connected to the physical machines pmx, pmy, and pmz, and manages an arbitrary virtual machine that operates in these. Here, the virtual machine vma is running in the physical machine pmx and the virtual machine vmb is running in the physical machine pmy, but is not limited thereto. In the present embodiment, it is sufficient that a plurality of virtual machines operate in at least different physical machines.

  The virtual machine management apparatus 1000 includes a detection unit 1100, a specification unit 1200, and a determination unit 1300. The detection unit 1100 detects an access pattern indicating regularity of increase / decrease in the number of accesses and resource usage status for each of a plurality of virtual machines including the virtual machines vma and vmb. The specifying unit 1200 specifies a combination having a low access pattern correlation among a plurality of virtual machines. The determining unit 1300 determines the virtual machine to be migrated and the physical machine to be migrated based on the identified combination and the free capacity of the resources of each physical machine. Based on the access pattern, the determination unit 1300 determines a time zone in which the number of accesses in the migration target virtual machine decreases as the scheduled migration date and time.

  FIG. 2 is a flowchart showing a flow of the virtual machine migration management method according to the first exemplary embodiment of the present invention. First, the detection unit 1100 detects an access pattern indicating regularity of increase / decrease in the number of accesses and resource usage status for each of a plurality of virtual machines (S11). Next, the specifying unit 1200 specifies a combination having a low correlation of access patterns among a plurality of virtual machines (S12). For example, it is assumed that a combination of virtual machines vma and vmy is specified. Then, the determining unit 1300 determines the virtual machine to be migrated and the physical machine to be migrated based on the identified combination and the free capacity of the resources of each physical machine (S13). Based on the access pattern, the determination unit 1300 determines a time zone in which the number of accesses in the migration target virtual machine decreases as the scheduled migration date and time (S14).

  As described above, in this embodiment, in order to consider not only the current time but also the transition of the number of accesses during migration, the time zone in which the number of accesses in the migration target virtual machine decreases is migrated based on the access pattern. Determined as scheduled date and time. Therefore, it is possible to eliminate the influence of the load accompanying the migration of the virtual machine as much as possible, and to reduce the failure of the migration of the virtual machine.

<Embodiment 2>
FIG. 3 is a block diagram showing an overall configuration including the virtual machine management apparatus 2000 according to the second embodiment of the present invention. In FIG. 3, it is assumed that at least two virtual machines vma and vmb are operating in at least the physical machine pmx. The virtual machine management apparatus 2000 is connected to the physical machines pmx and pmy, and manages an arbitrary virtual machine operating on them.

  The virtual machine management apparatus 2000 includes a detection unit 2100, a calculation unit 2200, and a determination unit 2300. The detection unit 2100 detects a virtual machine whose number of accesses tends to increase among a plurality of virtual machines vma and vmb operating in the physical machine pmx. The calculation unit 2200 calculates an estimated time when the resource is insufficient based on the detected number of accesses in the virtual machine. In addition, the calculation unit 2200 calculates a migration required time required for migration of the virtual machine between physical machines based on the data amount of the virtual machine and the network speed. In addition, the calculation unit 2200 calculates the limit access number at the time of migration of the virtual machine based on the required migration time. In addition, the calculation unit 2200 calculates the migration required time for virtual machines other than the detected virtual machine. The determination unit 2300 determines a virtual machine to be migrated and a scheduled migration date and time among a plurality of virtual machines based on the estimated time, the migration required time, and the limit access number.

  FIG. 4 is a flowchart showing a flow of the virtual machine migration management method according to the second exemplary embodiment of the present invention. First, the detection unit 2100 detects a virtual machine whose number of accesses tends to increase among a plurality of virtual machines operating in the physical machine (S21).

  Next, the calculation unit 2200 calculates an estimated time of resource shortage based on the detected number of accesses in the virtual machine, and shifts between physical machines of the virtual machine based on the data amount and network speed of the virtual machine. The required migration time is calculated, the limit access number at the time of migration of the virtual machine is calculated based on the migration required time, and the migration required time is calculated for virtual machines other than the detected virtual machine (S22).

  Then, the determination unit 2300 determines a virtual machine to be migrated and a scheduled migration date and time among a plurality of virtual machines based on the estimated time, the migration required time, and the limit access number (S23).

  As described above, in this embodiment, when the number of accesses of a part of virtual machines in a certain physical machine is on an upward trend, the migration target virtual machine to be migrated from the limit access number of the virtual machine to another physical machine. The machine can be determined, and the estimated migration date and time can be determined using the estimated time when the resource is insufficient and the required migration time. Therefore, it is possible to eliminate the influence of the load accompanying the migration of the virtual machine as much as possible, and to reduce the failure of the migration of the virtual machine.

<Embodiment 3>
The third embodiment is a specific example of the first and second embodiments described above. Here, the problem to be solved will be described again. In addition, if the memory is frequently updated during live migration, the migration may fail because the differential copy of the memory cannot keep up with the update amount. Also, in an environment where the network performing the migration cannot be separated from the network for user data communication, the load caused by the migration affects the user data communication, resulting in a system failure due to a response speed delay or excessive request retry count. There is a possibility of inviting.

  Therefore, in this embodiment, the migration of virtual machines for the purpose of resource efficiency and the planned migration of virtual machines for the purpose of physical machine maintenance, etc. are targeted efficiently. In addition, the failure is reduced by satisfying the user system requirements. As a part of this, for example, the number of accesses is recorded by a virtual machine management apparatus for a certain period, and a virtual machine suitable for live migration is selected by detecting an access pattern.

  In particular, the determination unit determines, as the migration destination, a physical machine having a sufficient free capacity of the resource among physical machines to which the virtual machines related to the combination belong. Here, a physical machine having a sufficient free capacity of a resource means a physical machine having a capacity of a resource capable of operating both of the virtual machines related to the combination.

  Further, the determination unit determines, as a migration target, a virtual machine having a smaller number of accesses in a time zone in which the number of accesses decreases when all of the physical machines have sufficient free capacity of the resources. Is desirable. Further, when none of the physical machines has sufficient free capacity of the resource, the determination unit sets both virtual machines as migration targets, and selects a physical machine other than the physical machine to which the both virtual machines belong. It is desirable to determine the migration destination. In addition, the image processing apparatus further includes a reception unit that receives designation of a migration target virtual machine from the outside, and the determination unit detects the access pattern of the virtual machine when the detection unit does not detect the access pattern for the designated virtual machine. The timing when the decreasing tendency is detected in the number of accesses may be determined as the scheduled transfer date and time.

  FIG. 5 is a block diagram showing an overall configuration including the virtual machine management apparatus 10 according to the third embodiment of the present invention. The virtual machine management apparatus 10 is an information processing apparatus installed in a data center connected via a network N from a client C that uses a cloud environment. In the data center, physical machines 100, 200, and 300 are installed. The physical machine 100 includes, for example, virtual machines 101 to 103 and a virtual machine management unit 110. The physical machine 200 includes, for example, virtual machines 201 to 203 and a virtual machine management unit 210. The physical machine 300 includes a virtual machine management unit 310. The virtual machine management units 110, 210, and 310 control the operation of a virtual machine that operates in the physical machine to which the virtual machine management unit 110, 210, and 310 belongs.

  The virtual machine management apparatus 10 is connected to the physical machines 100, 200, and 300 via an internal network. Note that at least two or more physical machines may be installed in the data center. In addition, two or more virtual machine management devices 10 may be installed in the data center. In this case, the virtual machine management devices 10 can cooperate with each other.

  FIG. 6 is a block diagram showing an internal configuration of the virtual machine management apparatus 10 according to the third embodiment of the present invention. The virtual machine management device 10 includes an operation manager input unit 11, a migration command unit 12, a migration management unit 20, an auxiliary storage device 25, and a virtual machine management unit 30. The operation manager input unit 11 is connected to an external terminal and provides a management screen that can be operated. Based on the input from the migration management unit 20, the migration command unit 12 specifies a migration destination physical machine and sends a migration start command to the migration target virtual machine in the data center.

  The auxiliary storage device 25 is a storage device that can be accessed from the migration management unit 20 and the virtual machine management unit 30. The auxiliary storage device 25 stores virtual machine migration management information, network load status, and the like.

  FIG. 7 is a diagram showing an example of virtual machine migration management information according to the third exemplary embodiment of the present invention. The virtual machine migration management information is a table in which user requirements, priority, migration deadline date and time required for migration, and the like are associated with IP addresses. The IP address is an IP address of all virtual machines to be managed in the virtual machine management apparatus 10. The user requirement is an upper limit value of the number of retries specified by the user or an upper limit value of the response time. The user requirement is, for example, a threshold value indicating the upper limit of the number of retries and the response speed of the system that the cloud environment user presents when starting to use the cloud environment.

  The priority is for determining the transfer order. The priority can be specified by an integer value from 1 to 10, and the default value is 5. The priority is higher as the value is smaller. In addition to being designated by the user, the priority may be appropriately updated according to the access status or the like. The migration deadline date and time is the date and time of migration deadline for each virtual machine designated by the user. The required migration time is an estimated value of the required time required for migration between physical machines in a virtual machine.

  FIG. 9 is a diagram illustrating an example of a network load situation according to the third embodiment of the present invention. The network load status is an example of a resource usage status collected by the virtual machine management unit 30. For example, the path a is a network speed when the virtual machine 101 is accessed from the client C via the network N and the virtual machine management apparatus 10.

  The migration management unit 20 includes a user requirement management unit 22, a calculation unit 23, and a migration target management unit 24. The user requirement management unit 22 refers to the user requirements of the virtual machine migration management information stored in the auxiliary storage device 25 and monitors whether actual access satisfies the user requirements. The migration target management unit 24 selects a virtual machine to be migrated based on the priority of the virtual machine migration management information stored in the auxiliary storage device 25 and the calculated planned migration date and time, and issues a command to the migration command unit 12. put out.

  The calculation unit 23 receives information from the virtual machine management unit 30 and calculates the following values. For example, when the resource usage status of the virtual machine is increasing due to an increase in access or the like, the calculation unit 23 calculates an estimated time until resource depletion. That is, the calculation unit 23 calculates the estimated time when the resource is insufficient based on the number of accesses in the virtual machine whose number of accesses is increasing. In addition, the calculation unit 23 calculates the required migration time when performing migration of a virtual machine whose resource usage is increasing. In addition, when another virtual machine is operating on the same physical machine as the virtual machine whose resource usage is increasing, the calculation unit 23 also calculates the migration required time for the other virtual machine. Here, the calculation unit 23 may calculate the required migration time based on the transition of the data amount of the virtual machine and the network speed.

  Further, the calculation unit 23 calculates the limit number of accesses at the time of migration of the virtual machine based on the required migration time. In other words, the calculation unit 23 calculates the limit access number that is estimated to fail without the memory copy being caught up during live migration. Here, the calculation unit 23 may calculate the limit access number based on the transition of the network speed and the memory update amount. Further, when there is regularity of increase / decrease in the number of accesses of the virtual machine, the calculation unit 23 calculates a time zone in which the transition can be performed in the shortest time until the estimated time. In addition, when the number of accesses tends to decrease, the calculation unit 23 calculates a duration time in which the decreasing trend is predicted to continue.

  The virtual machine management unit 30 acquires information on all virtual machines in the data center. The virtual machine management unit 30 includes a resource management unit 31, an access management unit 32, a pattern analysis unit 34, and an auxiliary storage device 35. The auxiliary storage device 35 is a storage device that can be accessed from the migration management unit 20 and the virtual machine management unit 30. The auxiliary storage device 35 stores a number of accesses for each virtual machine, a resource holding amount, a resource usage status, an access pattern, a mapping table indicating a correlation of access patterns between virtual machines, and the like. FIG. 8 is a diagram illustrating an example of a matching table according to the third embodiment of the present invention. Here, the access pattern is information indicating the regularity of the increasing / decreasing tendency of the number of accesses. In particular, regularity means, for example, that when a specific time zone is compared on a daily basis, the increase or decrease in the number of accesses is close, or the interval or number of increases or decreases is repeated a certain number of times. It means that The auxiliary storage device 35 may store information indicating the presence / absence of an access pattern for each virtual machine.

  The resource management unit 31 measures the resource holding amount and usage status of each virtual machine, and saves and manages them in the auxiliary storage device 35. Here, the resource usage status includes, for example, a CPU (Central Processing Unit) usage rate, a memory usage rate, an HDD (Hard Disk Drive) usage rate, a network bandwidth, and the like. Moreover, the resource management part 31 detects it, when a resource usage condition exists in the upward tendency or the decreasing tendency.

  The access management unit 32 measures the number of accesses, the response time, and the number of retries of each virtual machine, and stores them in the auxiliary storage device 35 for management. Further, the access management unit 32 detects when the number of accesses tends to increase or decrease.

  The pattern analysis unit 34 refers to the auxiliary storage device 35, analyzes changes in the number of accesses and resource usage of each virtual machine for a certain period, and detects the presence or absence of regularity (access pattern) along the time axis. . Here, the pattern analysis unit 34 stores the analysis result and the presence / absence of regularity in the auxiliary storage device 35. The pattern analysis unit 34 may calculate a correlation between the number of accesses of each virtual machine and the resource usage rate.

  Then, the pattern analysis unit 34 searches for a virtual machine in which the regularity is detected and searches for a partner whose total resource usage rate is almost the same as the other regular virtual machines, and deals with them. Addition (mapping) is performed and the auxiliary storage device 35 stores the mapping table. Specifically, the pattern analysis unit 34 calculates the correlation of access patterns between virtual machines. Then, a combination in which the correlation between the access patterns is equal to or less than the threshold is specified, and the correlation between the combination and the access pattern is registered in the auxiliary storage device 35 as a mapping table.

  In addition, the pattern analysis unit 34 periodically checks both access patterns of the virtual machines related to the combination and the correlation thereof, and when the access pattern is no longer detected or when the correlation of the access pattern exceeds a threshold value. Deletes the mapping table registration.

  Further, the pattern analysis unit 34 uses the moving average method to determine whether the number of accesses and the resource consumption are correlated, and determines whether the trend is increasing or decreasing, and the subsequent duration of the trend Predict.

  Here, an operation at the initial stage of installation when the virtual machine management apparatus 10 is installed in a data center that has started operation will be described. FIG. 10 is a flowchart showing the flow of registration processing according to the third embodiment of the present invention. First, immediately after the virtual machine management device 10 is activated, the virtual machine management unit 30 acquires information on IP addresses and network load conditions of all virtual machines to be managed in the data center (S41). For example, the resource management unit 31 of the virtual machine management unit 30 measures the network load status based on the packet throughput and the like, and stores it in the auxiliary storage device 25. After that, the resource management unit 31 periodically updates the network load status.

  Then, the virtual machine management unit 30 creates a table with the same number of rows as the acquired IP address of the virtual machine in the auxiliary storage device 25 as virtual machine migration management information. Then, the virtual machine management unit 30 adds the acquired IP address information to the table (S42).

  FIG. 11 is a flowchart showing a flow of pattern analysis processing according to the third embodiment of the present invention. The resource management unit 31 and the access management unit 32 start monitoring the resource usage rate and the number of accesses of each virtual machine (S51). These pieces of information are collected in association with the virtual machine information obtained by the virtual machine management unit 30.

  The resource management unit 31 and the access management unit 32 transmit the acquired monitoring target information to the pattern analysis unit 34 as needed. The pattern analysis unit 34 performs pattern analysis based on the received monitoring target information. Then, the pattern analysis unit 34 stores the information to be monitored and the analysis result in the auxiliary storage device 35 (S52). The information to be monitored and the analysis result may be deleted after a certain period of time.

  FIG. 12 is a flowchart showing the flow of the user requirement input process according to the third embodiment of the present invention. First, the operation manager inputs user requirements to the corresponding system (virtual machine) from the operation manager input unit 11 (S61). The user requirement management unit 22 updates the auxiliary storage device 25 by adding the input user requirement information to the virtual machine migration management information (S62).

  Next, an operation when a migration of a virtual machine is planned by the operation manager will be described. In this case, since the maintenance of the physical machine is assumed, it is assumed that the migration target virtual machine is determined. FIG. 13 is a flowchart showing the flow of the migration plan process according to the third embodiment of the present invention.

  First, the operation manager designates the priority and the migration deadline date and time for the migration target virtual machine A from the operation manager input unit 11 (S71). For example, the operation manager inputs a higher value for the priority of the virtual machine A, that is, a smaller value. In response to this, the migration management unit 20 updates the auxiliary storage device 25 by adding the specified priority and migration deadline date to the virtual machine migration management information.

  Next, the pattern analysis unit 34 checks whether or not there is an access pattern for the corresponding virtual machine A (S72). Specifically, the pattern analysis unit 34 refers to the auxiliary storage device 35 to determine whether there is an access pattern related to the virtual machine.

  If it is confirmed in step S72 that there is no access pattern, the process proceeds to step S76. When the access management unit 32 detects that the number of accesses of the virtual machine A is decreasing, it notifies the migration management unit 20. When notified that the number of accesses of the virtual machine A is in a decreasing trend, the calculation unit 23 of the migration management unit 20 calculates a prediction of a decreasing trend duration. In addition, the calculation unit 23 calculates the migration required time of the virtual machine A. Then, the migration target management unit 24 issues a command to start migration of the virtual machine A when the calculated duration time is longer than the migration required time. That is, the timing at which the decreasing tendency of the number of accesses is detected is determined as the scheduled transfer date (S76). Note that the migration destination is an arbitrary physical machine other than the physical machine on which the virtual machine A operates.

  When it is confirmed in step S72 that there is an access pattern, the migration target management unit 24 sets a time zone with a small number of accesses as a recommended migration date and time based on the access pattern (S73). The recommended transition date and time is separately before the estimated time until resource depletion calculated by the calculation unit 23. Then, the migration target management unit 24 determines whether the recommended migration date / time is before the migration deadline date / time (S74). If the recommended migration date is before the migration deadline date, the recommended migration date is set as the scheduled migration date (S75). On the other hand, if it is determined in step S74 that the recommended migration date / time is not earlier than the migration deadline date / time, the process proceeds to step S76 and the above-described processing is performed.

  If it is confirmed in step S72 that there is an access pattern, the process proceeds to step S77 in parallel with step S73. The pattern analysis unit 34 refers to the auxiliary storage device 35 and checks whether there is a virtual machine B having a pattern corresponding to the access pattern of the virtual machine A in the mapping table (S77). When there is a virtual machine B corresponding to the virtual machine A, the resource management unit 31 determines whether the physical machine to which the virtual machine B belongs has sufficient resource free space (S78). When there is sufficient resource free capacity, the migration target management unit 24 sets the migration destination to the physical machine to which the virtual machine B belongs (S79).

  If there is no virtual machine B corresponding to the virtual machine A in step S77, or if there is not enough resource free space in the physical machine to which the virtual machine B belongs in step S78, the migration target management unit 24 sets the migration destination. An arbitrary physical machine with sufficient resource free space is set (S80).

  Next, an operation when migration of a virtual machine is planned by pattern detection will be described. FIG. 14 is a flowchart showing the flow of the migration plan process according to the third embodiment of the present invention. As a premise, it is assumed that the priority of the virtual machines A and B to be processed is high.

  First, it is assumed that the pattern analysis unit 34 analyzes the change in the number of accesses of the virtual machine A along the time axis and determines that an access pattern exists (S81). Next, the pattern analysis unit 34 refers to the auxiliary storage device 35 and checks whether there is a virtual machine B having a pattern corresponding to the access pattern of the virtual machine A in the mapping table (S82). If there is no virtual machine B corresponding to the virtual machine A, the process is terminated without migrating (S83).

  On the other hand, when there is a virtual machine B corresponding to the virtual machine A in step S82, the resource management unit 31 determines whether there is sufficient resource free capacity in both of the physical machines to which the virtual machines A and B belong. Determine (S84).

  When both of the physical machines have sufficient resource free capacity, the migration target management unit 24 sets the physical machine to which the virtual machine with the smaller number of accesses in the time zone in which the number of accesses decreases belongs to the migration target. Furthermore, the migration target management unit 24 sets a time zone with a low number of accesses as the scheduled migration date (S85).

  On the other hand, in step S84, when both of the physical machines do not have sufficient resource free space, the resource management unit 31 has sufficient resources for one of the physical machines to which each of the virtual machines A and B (for example, the virtual machine B) belongs. It is determined whether there is free space (S86).

  When there is sufficient resource free space in one of the physical machines, the migration target management unit 24 sets the virtual machine A as the migration target. Further, the migration target management unit 24 sets a time zone during which the number of accesses of the virtual machine A is reduced to the scheduled migration date (S87).

  On the other hand, in step S86, when there is no sufficient resource free space in any of the physical machines, the migration target management unit 24 sets both (virtual machines A and B) as migration targets. Then, the migration target management unit 24 sets the time zone in which the access numbers of the virtual machines A and B are reduced to the respective scheduled migration dates and times (S88).

  Next, a description will be given of a case where migration is planned when an increase in the number of accesses is detected. Here, it is assumed that a plurality of virtual machines are operating in one physical machine. Also, as a premise, the priority of the virtual machines A, B, and C to be processed is high. FIG. 15 is a flowchart showing the flow of the migration planning process according to the third embodiment of the present invention.

First, it is assumed that the access management unit 32 detects a rising trend in the number of accesses of the virtual machine A (S91). Then, when the increase in the number of accesses continues, the calculation unit 23 calculates an estimated time d at which there is a risk of resource exhaustion. Furthermore, the calculation unit 23 calculates a required transition time Ta required for live migration from the upward trend and the load state of the network. Further, the calculation unit 23 calculates the limit access number Nm at which the memory copy cannot catch up during the live migration from the value of the required transition time Ta. Further, the calculation unit 23 calculates a migration deadline date du that can realize live migration in a state where the user requirements are satisfied (S92). The transition deadline date du is calculated so as to be before the estimated time d.

Thereafter, the virtual machine management unit 30 determines whether other virtual machines (referred to as virtual machines B and C) exist on the same physical machine as the virtual machine A (S93). If it does not exist, the process is terminated without shifting (S94).

  On the other hand, if it exists in step S93, the calculation unit 23 calculates the required migration times Tb and Tc required for live migration of the virtual machines B and C (S95). Then, the migration target management unit 24 compares the current access count Nr of the virtual machine A with the limit access count Nm, and determines whether Nr is greater than Nm (S96).

  When Nr is larger than Nm, the migration target management unit 24 sets the virtual machines B and C as migration targets. Then, the migration target management unit 24 sets the scheduled migration date and time before the migration deadline date and time du (S97).

  On the other hand, when Nr is Nm or less, the migration target management unit 24 sets the virtual machine A as the migration target. Then, the migration target management unit 24 sets the scheduled migration date and time before the migration deadline date and time du (S98).

  Thus, according to the present embodiment, the live migration failure probability can be reduced. In addition, cloud business providers can reduce virtual machine resource usage. Furthermore, it is possible to solve the resource shortage without taking the risk of failure of live migration even for a system whose load is increasing. Further, by considering the response time and the upper limit of the number of retries as user requirements, the reliability of the system can be maintained even under the influence of live migration.

<Other embodiments>
The priority may be updated. For example, when an increase in the number of accesses occurs in an operating system, first, a correlation between the number of accesses and the resource usage status in the access management unit 32 and the resource management unit 31 is detected. When the calculation unit 23 determines that there is a possibility of resource exhaustion, the migration target management unit 24 determines the migration expiration date and time from the estimated time until resource exhaustion. The transfer target management unit 24 further determines the priority based on the time limit.

  In addition, when the above-described virtual machine management apparatus 1000 or the like is placed under the management of a plurality of data centers connected via a network, the virtual machine management apparatus 1000 or the like arranged for each data center is linked. Applicable.

  Note that the present invention is not limited to the above-described embodiment, and can be changed as appropriate without departing from the spirit of the present invention.

  In the above-described embodiments, the present invention has been described as a hardware configuration, but the present invention is not limited to this. The present invention can also realize arbitrary processing by causing a CPU (Central Processing Unit) to execute a computer program.

  In the above example, the program can be stored and supplied to a computer using various types of non-transitory computer readable media. Non-transitory computer readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic recording media (for example, flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (for example, magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / W, DVD (Digital Versatile Disc), BD (Blu-ray (registered trademark) Disc), semiconductor memory (for example, mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, RAM ( Random Access Memory)). The program may also be supplied to the computer by various types of transitory computer readable media. Examples of transitory computer readable media include electrical signals, optical signals, and electromagnetic waves. The temporary computer-readable medium can supply the program to the computer via a wired communication path such as an electric wire and an optical fiber, or a wireless communication path.

A part or all of the above embodiments can be described as in the following supplementary notes, but is not limited thereto.
(Appendix 1)
For each of a plurality of virtual machines running in different physical machines, a detection unit that detects an access pattern indicating regularity of increase / decrease in the number of accesses,
A specifying unit that specifies a combination having a low correlation of the access patterns among the plurality of virtual machines;
The migration target virtual machine and the migration destination physical machine are determined based on the identified combination and the resource free capacity of each physical machine, and the migration target virtual machine in the migration target virtual machine is determined based on the access pattern. A determination unit that determines a time zone in which the number of accesses decreases as a scheduled transfer date and time,
A virtual machine management device.
(Appendix 2)
The determination unit
The virtual machine management device according to claim 1, wherein, among the physical machines to which the virtual machines related to the combination belong, a physical machine having a sufficient free capacity of the resource is determined as the migration destination.
(Appendix 3)
The determination unit
The virtual machine management device according to attachment 2, wherein a physical machine having a resource capacity capable of operating both of the virtual machines according to the combination is a physical machine having a sufficient free capacity of the resource.
(Appendix 4)
The determination unit
The virtual machine having a smaller number of accesses in the time zone in which the number of accesses decreases is determined as the migration target when any of the physical machines has sufficient free capacity of the resource. Virtual machine management device.
(Appendix 5)
The determination unit
If none of the physical machines has sufficient free space in the resource, both virtual machines are set as the migration target, and a physical machine other than the physical machine to which the both virtual machines belong is determined as the migration destination. The virtual machine management device according to any one of appendix 2 or 3.
(Appendix 6)
It further includes a reception unit that accepts designation of the virtual machine to be migrated from outside,
The determination unit
If the access pattern is not detected for the designated virtual machine by the detection unit, the timing at which a decreasing tendency is detected in the number of accesses of the virtual machine is determined as the scheduled migration date and time. The virtual machine management device according to claim 1.
(Appendix 7)
For each of a plurality of virtual machines running in different physical machines, an access pattern indicating regularity of increase / decrease in the number of accesses is detected,
Identifying a combination with a low correlation of the access patterns among the plurality of virtual machines;
Based on the identified combination and the free capacity of the resources of each physical machine, determine a virtual machine to be migrated and a physical machine to be migrated,
Based on the access pattern, a time zone in which the number of accesses in the migration target virtual machine decreases is determined as a scheduled migration date and time,
Virtual machine management method.
(Appendix 8)
For each of a plurality of virtual machines running in different physical machines, processing for detecting an access pattern indicating regularity of increase / decrease in the number of accesses,
A process of identifying a combination having a low correlation of the access patterns among the plurality of virtual machines;
A process of determining a migration target virtual machine and a migration destination physical machine based on the identified combination and the free capacity of the resource of each physical machine;
Based on the access pattern, a process for determining a time zone in which the number of accesses in the migration target virtual machine decreases as a scheduled migration date and time,
Virtual machine management program that causes a computer to execute.
(Appendix 9)
A detection unit that detects a virtual machine whose number of accesses is increasing among a plurality of virtual machines operating in a physical machine;
Based on the number of accesses in the detected virtual machine, an estimated time when the resource is insufficient is calculated, and based on the data amount and network speed of the virtual machine, the migration time required for migration between physical machines of the virtual machine is calculated. Calculate, calculate the limit access number at the time of migration of the virtual machine based on the migration required time, calculate the migration required time for virtual machines other than the detected virtual machine,
Based on the estimated time, the required migration time, and the limit access number, a determination unit that determines a virtual machine to be migrated among the plurality of virtual machines and a scheduled migration date and time,
A virtual machine management device.
(Appendix 10)
Detects virtual machines whose access count is increasing among multiple virtual machines running in physical machines.
Calculate an estimated time of resource shortage based on the number of accesses in the detected virtual machine,
Based on the amount of data of the virtual machine and the network speed, calculate the time required for migration of the virtual machine between physical machines,
Calculate the limit number of accesses at the time of migration of the virtual machine based on the migration required time,
Calculate the migration time for virtual machines other than the detected virtual machine,
Based on the estimated time, the required migration time, and the limit access number, determine a virtual machine to be migrated among the plurality of virtual machines, and a scheduled migration date and time,
Virtual machine management method.
(Appendix 11)
A process for detecting a virtual machine whose number of accesses is increasing among a plurality of virtual machines running in a physical machine;
A process of calculating an estimated time of resource shortage based on the number of accesses in the detected virtual machine;
A process of calculating a migration required time required for migration of the virtual machine between physical machines based on the data amount and network speed of the virtual machine;
A process of calculating a limit access number at the time of migration of the virtual machine based on the migration required time;
Processing for calculating the migration required time for virtual machines other than the detected virtual machine;
Based on the estimated time, the required migration time, and the limit access number, a process for determining a migration target virtual machine and a scheduled migration date and time among the plurality of virtual machines;
Virtual machine management program that causes a computer to execute.

1000 virtual machine management apparatus 1100 detection unit 1200 identification unit 1300 determination unit pmx physical machine pmy physical machine pms physical machine vma virtual machine vmb virtual machine 2000 virtual machine management apparatus 2100 detection unit 2200 calculation unit 2300 determination unit 10 virtual machine management apparatus 100 physical Machine 101 virtual machine 102 virtual machine 103 virtual machine 110 virtual machine management unit 200 physical machine 201 virtual machine 202 virtual machine 203 virtual machine 210 virtual machine management unit 300 physical machine 310 virtual machine management unit 11 operation administrator input unit 12 migration command unit DESCRIPTION OF SYMBOLS 20 Migration management part 22 User requirement management part 23 Calculation part 24 Migration object management part 25 Auxiliary storage device 30 Virtual machine management part 31 Resource management part 32 Access management part 3 Pattern analysis unit 35 the auxiliary storage device C client N Network

Claims (10)

For each of a plurality of virtual machines running in different physical machines, a detection unit that detects an access pattern indicating regularity of increase / decrease in the number of accesses,
A specifying unit that specifies a combination having a low correlation of the access patterns among the plurality of virtual machines;
The migration target virtual machine and the migration destination physical machine are determined based on the identified combination and the resource free capacity of each physical machine, and the migration target virtual machine in the migration target virtual machine is determined based on the access pattern. A determination unit that determines a time zone in which the number of accesses decreases as a scheduled transfer date and time,
A virtual machine management device. The determination unit
The virtual machine management apparatus according to claim 1, wherein a physical machine having a sufficient free capacity of the resource among physical machines to which the virtual machines related to the combination belong is determined as the migration destination. The determination unit
The virtual machine management apparatus according to claim 2, wherein a physical machine having a capacity of a resource capable of operating both of the virtual machines related to the combination is a physical machine having a sufficient free capacity of the resource. The determination unit
The virtual machine having a smaller number of accesses in a time zone in which the number of accesses decreases is determined as the migration target when any of the physical machines has sufficient free space for the resource. Virtual machine management device. The determination unit
If none of the physical machines has sufficient free space in the resource, both virtual machines are set as the migration target, and a physical machine other than the physical machine to which the both virtual machines belong is determined as the migration destination. The virtual machine management apparatus according to claim 2. It further includes a reception unit that accepts designation of the virtual machine to be migrated from outside,
The determination unit
The timing at which a decrease tendency is detected in the number of accesses of the virtual machine is determined as the scheduled migration date and time when the access unit does not detect the specified virtual machine by the detection unit. The virtual machine management device according to any one of the above. For each of a plurality of virtual machines running in different physical machines, an access pattern indicating regularity of increase / decrease in the number of accesses is detected,
Identifying a combination with a low correlation of the access patterns among the plurality of virtual machines;
Based on the identified combination and the free capacity of the resources of each physical machine, determine a virtual machine to be migrated and a physical machine to be migrated,
Based on the access pattern, a time zone in which the number of accesses in the migration target virtual machine decreases is determined as a scheduled migration date and time,
Virtual machine management method. For each of a plurality of virtual machines running in different physical machines, processing for detecting an access pattern indicating regularity of increase / decrease in the number of accesses,
A process of identifying a combination having a low correlation of the access patterns among the plurality of virtual machines;
A process of determining a migration target virtual machine and a migration destination physical machine based on the identified combination and the free capacity of the resource of each physical machine;
Based on the access pattern, a process for determining a time zone in which the number of accesses in the migration target virtual machine decreases as a scheduled migration date and time,
Virtual machine management program that causes a computer to execute. A detection unit that detects a virtual machine whose number of accesses is increasing among a plurality of virtual machines operating in a physical machine;
Based on the number of accesses in the detected virtual machine, an estimated time when the resource is insufficient is calculated, and based on the data amount and network speed of the virtual machine, the migration time required for migration between physical machines of the virtual machine is calculated. Calculate, calculate the limit access number at the time of migration of the virtual machine based on the migration required time, calculate the migration required time for virtual machines other than the detected virtual machine,
Based on the estimated time, the required migration time, and the limit access number, a determination unit that determines a virtual machine to be migrated among the plurality of virtual machines and a scheduled migration date and time,
A virtual machine management device. Detects virtual machines whose access count is increasing among multiple virtual machines running in physical machines.
Calculate an estimated time of resource shortage based on the number of accesses in the detected virtual machine,
Based on the amount of data of the virtual machine and the network speed, calculate the time required for migration of the virtual machine between physical machines,
Calculate the limit number of accesses at the time of migration of the virtual machine based on the migration required time,
Calculate the migration time for virtual machines other than the detected virtual machine,
Based on the estimated time, the required migration time, and the limit access number, determine a virtual machine to be migrated among the plurality of virtual machines, and a scheduled migration date and time,
Virtual machine management method.

Images