JP6426532B2 - Virtual machine operation support system and virtual machine operation support method - Google Patents

Description

  The present invention relates to a virtual machine operation support system and a virtual machine operation support method, and more specifically to a technology for efficiently realizing performance stabilization in a virtual machine.

  In recent years, utilization of virtualization technology in information systems has become commonplace. Virtualization technology is a technology that abstracts computing resources such as servers. By using such virtualization technology, it is possible to operate a plurality of abstracted computers (hereinafter, virtual machines) on one physical server and to operate many virtual machines on a small number of physical servers. Become. By applying this virtualization, cost reduction due to the consolidation effect of virtual machines can be expected.

  However, if the consolidation rate of virtual machines is increased too much, many virtual machines will compete for computing resources on one physical server, which may cause performance degradation due to resource shortage. Therefore, in operation management of a virtualization environment, it becomes an important issue how to achieve both cost reduction by enhancement of consolidation rate and securing of performance stability.

  For example, in a system constituted by a plurality of physical servers, when a load bias occurs among the physical servers as a technique aiming to achieve both the stability of performance stability and the improvement of aggregation rate in such a virtualization environment. A technique (see Patent Document 1) or the like for moving a virtual server operating on a certain physical server to another physical server using a live migration technique has been proposed. This balances the load and reduces the risk of performance degradation.

US 8095929

  The above-described conventional technology is configured to determine the physical server to which the virtual server is moved based on the load information at the time of the movement examination. In such a configuration, although it is possible to attempt load leveling at the time of study of movement, it is not assumed that a load increase will occur in the physical server of which the virtual machine has already been moved. If such a situation occurs, it is necessary to re-execute live migration again despite the initial load leveling, ie, the execution of live migration.

  The execution of live migration consumes resources such as CPU and network bandwidth, and may adversely affect the performance of an application running on a virtual machine. Therefore, it is not preferable to perform live migration frequently.

  Therefore, an object of the present invention is to provide a technology for efficiently realizing performance stabilization in a virtual machine.

  Therefore, the present invention has an object of enabling selection of a destination host capable of stably providing resources over a long period of time in order to realize performance stabilization by implementing a small number of live migrations in optimizing placement of a virtual server. Do.

  The workload on the destination host selection in the administrator can be reduced. In addition, since the destination host can be determined based on the evaluation result in consideration of the continuity, it becomes possible to select a host that can afford the resources for a long period of time, suppressing frequent VM migration and performance. It becomes possible to stabilize.

A virtual machine operation support system according to the present invention, which solves the above-mentioned problems, comprises: a storage device holding at least reservation information of resources already allocated for each virtual machine as information on each physical server to which the virtual machine is allocated ; For each physical server, based on information on resources required for a predetermined virtual machine from a predetermined device regarding a predetermined virtual machine and information on each physical server held by the storage device, for each predetermined virtual machine a process of determining the continuity of the time can be secured to the appropriate resources from the predetermined timing at a predetermined algorithm, the predetermined virtual machine and information resources required for a predetermined period, each time for each physical server that the storage device is held On each physical server based on the reservation information of And determining process at a predetermined algorithm margin of resources can, an arithmetic unit for executing a process of displaying a predetermined device the margin and the information of the continuity obtained for each physical server by each judgment, the It is characterized by having.

Further, according to the virtual machine operation support method of the present invention, there is provided information including a storage device holding at least reservation information of resources already allocated for each virtual machine as information on each physical server which is an allocation destination of the virtual machine. For each physical server, the predetermined virtual server is obtained based on information on resources required by the virtual machine for a predetermined time acquired by the processing system from a predetermined device for a predetermined virtual machine and information on each physical server held by the storage device. a process of determining the continuity of the time can be secured to the appropriate resources from a predetermined time to machine types in a predetermined algorithm, and the predetermined virtual machine resources required for the predetermined time information, each physical server in which the storage device holds Directed to the predetermined virtual machine in each physical server based on the reservation information for each time related to A process of determining margin to be secured resource at a predetermined algorithm, the process of displaying the margin and the information of the continuity obtained for each physical server to a predetermined device by each determination, to run, that It features.

  According to the present invention, performance stabilization in a virtual machine can be efficiently realized.

It is a figure which shows the example of a network configuration containing the virtual machine operation assistance system in this embodiment. It is a figure showing the example of hardware constitutions of the virtual machine operation support system in this embodiment. It is a figure which shows the structural example of VM management information DB in this embodiment. It is a figure which shows the structural example of host management information DB in this embodiment. It is a figure which shows the example of a structure of resource amount DB scheduled for allocation in this embodiment. It is a figure which shows the structural example of the over-commit rate upper limit DB in this embodiment. It is a figure showing an example of composition of operation history DB in this embodiment. It is a figure which shows the structural example of resource usage prediction DB in this embodiment. It is a flowchart which shows process procedure example 1 of the virtual machine operation assistance method in this embodiment. It is a flowchart which shows process procedure example 2 of the virtual machine operation assistance method in this embodiment. It is a flowchart which shows process sequence example 3 of the virtual machine operation assistance method in this embodiment. It is a flowchart which shows process procedure example 4 of the virtual machine operation assistance method in this embodiment. It is a flowchart which shows process procedure example 5 of the virtual machine operation assistance method in this embodiment. It is a figure which shows the example 1 of a screen display in this embodiment. It is a figure which shows the screen display example 2 in this embodiment. It is a figure which shows the example 3 of a screen display in this embodiment. It is a figure which shows the example 4 of a screen display in this embodiment.

--- Network configuration ---
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a network configuration diagram including an operation management server 100 which is a virtual machine operation support system of the present embodiment. An operation management server 100 (hereinafter, operation management server), which is a virtual machine operation support system shown in FIG. 1, is a computer system that efficiently realizes performance stabilization in a virtual machine.

  In the network configuration illustrated in FIG. 1, the operation management server 100 is communicably connected to the operation management target system 1 and the operation terminal 200 via the networks 10 and 11. Among them, the operation management target system 1 is configured to include a plurality of hosts 21 to 23, 31 to 33, and the like. The operation terminal 200 is a terminal used by the operation manager of the operation management target system 1 described above.

  The hosts 21 to 23 in the operation management target system 1 described above are accommodated in the rack 20, and the hosts 31 to 33 are accommodated in the rack 30. In addition, virtual servers 25 to 27 which are virtual machines run on the hosts 21 to 23 in the rack 20, and virtual servers 35 to 37 run on the hosts 31 to 33 in the rack 30, respectively. Note that such a virtual server is referred to as a VM hereinafter.

  The operation management server 100 in such a network configuration is a server device that performs operation management of VMs in each host included in the operation management target system 1 described above, and moves a predetermined VM with occurrence of load imbalance among the hosts. That is, the information required to determine the necessity of so-called live migration is generated, and this is output to the operation terminal 200. The above-mentioned operation manager browses the above-mentioned information outputted from the operation management server 100 via the operation terminal 200, and can easily and accurately judge the virtual server to be moved, its movement necessity, and the movement destination Become. The operation terminal 200 may be provided, for example, for each operation manager.

--- Hardware configuration ---
The hardware configuration of the operation management server 100 of this embodiment is as follows. The operation management server 100 according to the present embodiment includes a storage device 101 configured of an appropriate non-volatile storage element such as a solid state drive (SSD) or a hard disk drive, a memory 103 configured of a volatile storage element such as a RAM, and the storage device 101. The program 102 stored in the memory is read out to the memory 103 and executed to perform overall control of the apparatus itself and to perform various determinations, calculations and control processing as well as the CPU 104 as an arithmetic unit and the networks 10 and 11 A communication device 105 responsible for communication processing of

  The above-described CPU 104 executes the program 102 to implement the functions of the resource reservation processing unit 110, the resource usage prediction unit 111, and the migration destination candidate evaluation unit 112. Details of each of these functions will be described later.

In the storage device 101, in addition to the program 102 for implementing the functions necessary as the virtual machine operation support system of the present embodiment, at least DBs 125 to 130 storing various information used for processing are stored. Specific examples of data configurations of these DBs 125 to 130 will be described later. When the operation management server 100 according to this embodiment is configured as a so-called stand-alone computer, in addition to the above-described configuration, an input device for receiving key input and voice input from the user, a display for displaying processing data, etc. An output device.

  The hosts 21 to 23, 31 to 33, and the like, and the operation terminal 200 also have a hardware configuration as a general computer device, similarly to the configuration of the operation management server 100 described above.

--- Data structure example ---
Next, an example of a data structure in a table used by the operation management server 100 of the present embodiment will be described. FIG. 3 is a diagram showing a configuration example of the VM management information DB 125 in the present embodiment. The VM management information DB 125 of the present embodiment illustrated in FIG. 3 is a database that stores management information regarding VMs operating on each host of the operation management target system 1 and VMs scheduled to be operated in the future.

  Each record in the VM management information DB 125 of FIG. 3 corresponds to one VM. In such a record, the value of the VM field 801 is an ID uniquely indicating the corresponding VM. Also, the value of the host field 802 is an ID uniquely indicating the host on which the VM is operating. The value of the operating time field 804 is time information indicating the operating time of the VM. Among them, the value of “start time” of the operation time field 804 is the operation start time of the corresponding VM. If this start time is a past time from the current time, it means that the corresponding VM is already in operation. On the other hand, when the start time is a future time from the current time, it indicates that the operation of the VM is reserved. In addition, the value of “end time” of the operating time field 804 is the scheduled end time of operation of the corresponding VM. If this end time is a future time from the current time, it indicates that the corresponding VM is scheduled to end operation at that time. On the other hand, when the end time is “−”, the corresponding VM indicates that the end schedule has not been determined. Also, the value of the allocation resource amount field 805 is the amount of resources to be allocated to the corresponding VM. Among them, “number of CPU cores” represents the number of CPU cores allocated to the corresponding VM, “CPU frequency” represents the frequency of each core allocated to the corresponding VM, and “memory” represents memory capacity allocated to the corresponding VM.

  FIG. 4 is a view showing a configuration example of the host management information DB 126 in the present embodiment. The host management information DB 126 of the present embodiment illustrated in FIG. 4 is a database storing information on each host included in the operation management target system 1.

  Each record of the host management information DB 126 illustrated in FIG. 4 is a record corresponding to one host. In this record, the value of the host field 811 is an ID uniquely indicating the corresponding host. Further, the value of the rack field 812 is an ID uniquely indicating a rack that accommodates the corresponding host. Also, the value of the capacity field 813 represents the capacity of physical resources possessed by the corresponding host.

  Of the capacity field 813, "CPU core number" indicates the number of CPU cores that the host has, "CPU frequency" indicates the frequency of each CPU core that the host has, and "memory" indicates the host. Represents the memory capacity that it has.

  For example, if the number of CPU cores is "12" and the frequency of one CPU is "2.0 GHz" as the capacity of resources related to a host, this host can provide VMs as resources based on CPU. Is 12 × 2 = 24 GHz.

FIG. 5 is a diagram showing a configuration example of the scheduled allocation resource amount DB 127 in the present embodiment. The scheduled allocation resource amount DB 127 of the present embodiment illustrated in FIG. 5 is a database storing information indicating the total amount of resources to be allocated to each VM operating on the host in each host.

  Each record of the scheduled allocation resource amount DB 127 illustrated in FIG. 5 indicates the scheduled allocation resource amount at a certain time for the corresponding host. Among these values, the value of the time field 821 represents a target time. Further, the value of the allocated resource amount field 822 is a total value of each resource amount of CPU and memory allocated to each VM that is scheduled to be operated at the corresponding time. The value of the overcommit rate field 823 is a ratio of the allocated resource amount 822 to the capacity of the resource possessed by the host.

  For example, if the frequency of the CPU scheduled to be allocated to each VM for a certain date and time is “6 GHz”, “4 GHz”, “2 GHz” (extract a record for the relevant date and time from the VM management information DB 125 With regard to the CPU among the allocated resource amounts of the corresponding date and time, 6 + 4 + 2 = 12 GHz. Also, in this case, if the CPU capacity among the resources related to the host is “number of CPU cores = 12” and “CPU frequency = 2.0 GHz” (extract the record for the host from the host management information DB 126 and acquire The corresponding host can provide up to 12 × 2.0 = 24 GHz CPU resources. In this case, the over-commit rate 823 has a value of (12 GHz / 24 GHz) × 100 = 50%.

  Thus, for example, when the over-commit rate is "50%", it indicates that 50% of resources of the capacity of resources possessed by the corresponding host are to be allocated. If you allocate more resources than the capacity of the resources of the host (that is, if you are over-committing), the over-commit rate exceeds 100%.

  FIG. 6 is a diagram showing a configuration example of the over-commit rate upper limit DB 128 in the present embodiment. The over-commit rate upper limit DB 128 of the present embodiment illustrated in FIG. 6 is a database storing the upper limit value of the over-commit rate permitted in the operation management target system 1. Of the overcommit rate upper limit DB 128, the value of the CPU field 831 represents the upper limit value of the overcommit rate of the CPU, and the value of the memory field 823 represents the upper limit value of the overcommit rate of the memory. Such an over-commit rate upper limit DB 128 is determined in advance in accordance with requirements such as cost and performance stability that can be tolerated by the operation manager, and is stored in the storage device 101.

  FIG. 7 is a view showing a configuration example of the operation history DB 129 in the present embodiment. The operation history DB 129 of the present embodiment illustrated in FIG. 7 is a database that stores, for each host, information on the amount of resources used by each VM on the corresponding host. That is, each record of the operation history DB 129 illustrated in FIG. 7 represents the usage record of each resource such as CPU and memory at a certain time in the past for the corresponding host.

  Among these values, the value of the time field 901 represents the target time. Further, the value of the CPU usage amount field 902 is a total value of the resource amount of CPU used by each VM on the corresponding host at the corresponding time. Further, the value of the memory field 903 is a total value of the resource amount of memory used by each VM on the corresponding host at the corresponding time.

FIG. 8 is a diagram showing a configuration example of the resource usage prediction DB 130 in the present embodiment. The resource usage prediction DB 130 of this embodiment illustrated in FIG. 7 stores the predicted value of the resource usage of each host generated by the operation management server 100 according to a predetermined algorithm based on the data of the operation history DB 129 described above. Database. That is, each record of the resource usage prediction DB 130 illustrated in FIG. 8 represents a usage prediction value of each resource such as CPU and memory at a certain time in the future for the corresponding host.

  Among these values, the value of the time field 911 represents the target time. Further, the value of the CPU usage amount field 912 is a total value of the CPU resource amount that will be used by each VM on the corresponding host at the corresponding time. Also, the value of the memory field 913 is a total value of the resource amount of memory that will be used by each VM on the corresponding host at the corresponding time. The generation of the resource usage prediction DB 130 will be described later.

---Example of resource reservation process---
Hereinafter, the actual procedure of the virtual machine operation support method in the present embodiment will be described based on the drawings. Various operations corresponding to the virtual machine operation support method described below are realized by a program that the above-described operation management server 100 as a virtual machine operation support system reads out to a memory or the like and executes. The program is composed of code for performing various operations described below.

  FIG. 9 is a flow chart showing a processing procedure example 1 of the virtual machine operation support method in the present embodiment. Here, two types of administrators, an application administrator and an infrastructure administrator, exist as users who operate the operation terminal 200 of the present embodiment. Among them, the application administrator is a user who performs operation management operation on an application operating on a VM. On the other hand, an infrastructure administrator is a user who manages infrastructures such as hosts and networks. In the operation terminal 200 that exchanges data with the operation management server 100 in the flow illustrated in FIG. 9, it is assumed that an application administrator who desires resource reservation for a VM operates among the users described above. That is, the processing procedure involved in resource reservation by the application manager is the flow of FIG.

  In this case, the operation terminal 200 receives a resource reservation request for a predetermined VM from the application manager, and transmits the request to the operation management server 100 via the network 11 (S411).

  On the other hand, the operation management server 100 receives the above-mentioned resource reservation request via the network 11, executes resource reservation processing described later (S412), and sends the result of the reservation processing back to the operation terminal 200. Then (S413), the process ends. In the operation management server 100, it is the resource reservation processing unit 110 that executes this resource reservation processing.

  Here, the details of the above-described step S412, that is, the resource reservation process will be described. FIG. 10 is a flowchart showing a processing procedure example 2 of the virtual machine operation support method in the present embodiment, and more specifically, it is a flowchart showing the details of the resource reservation process in the operation management server 100.

  In this case, first, the resource reservation processor 110 of the operation management server 100 receives a resource reservation request from the operation terminal 200 that has received an instruction from the application manager (S501). In the resource reservation request received here, the reservation start time and reservation end time corresponding to the reservation period for which resources are to be reserved for the VM, and the allocated resource amount indicating the resource amount to be secured (for example, the number of CPU cores, CPU frequency, memory Each information of) is included at least.

  Next, the resource reservation processing unit 110 of the operation management server 100 reads out information on all hosts included in the operation management target system 1 from the host management information DB 126, and for each host, the following steps S503 to S506 are performed. A series of each process is executed (S502).

  Among these, the resource reservation processing unit 110 executes, for example, in ascending order of host IDs (eg, the value of the host field 811 in the host management information DB 126 of FIG. 4) among the hosts from which the information is read out in step S502. For each selected processing target host, each record is referred to in the scheduled allocation resource amount DB 127, and each value of the allocation resource amount field 822 is extracted from each record in which the time of the time field 821 is included in the above-mentioned reservation period. (S503).

  Also, in step S504, the resource reservation processing unit 110 (currently resource allocation) is allocated to the allocation scheduled resource amount (for each other VM) (for example, CPU: 12 GHz, memory: 10 GB) extracted in step S503 described above. Resource amount (for example, CPU: 20 GHz, memory: 20 GB) indicated by the above resource reservation request (for example, CPU: 32 GHz, memory: 20 GB) If the processing target host plays all VMs, the overcommit rate in the reservation period is the upper limit (the value indicated by the overcommit rate upper limit DB 128 in FIG. 6. For example, CPU: 400%, memory: 100%) Is determined (S504).

  As a result of the determination, if the total resource amount described above exceeds the overcommit rate upper limit (S504: No), the resource reservation processing unit 110 ends the processing for the processing target host, returns the processing to step S502, and is not yet performed. Execute processing for the next host of processing.

  On the other hand, as a result of the above determination, if the total resource amount does not exceed the over-commit rate upper limit value (S504: Yes), the resource reservation processing unit 110 advances the process to step S505.

  In step S505, the resource reservation processing unit 110 calculates an average scheduled resource allocation amount in the above-described reservation period. Here, the average scheduled resource allocation amount is a time average of the total resource amount obtained in step S504 for the reservation period. For example, if the total resource amount is CPU: 32 GHz, memory: 20 GB and the reservation period is 2 hours, the average planned resource allocation amount is CPU: 16 GHz / hour, memory: 10 GB / hour.

  Subsequently, in step S506, the resource reservation processing unit 110 uses the ID of the processing target host targeted for processing up to the above-mentioned step S505 and the respective information of the average planned resource allocation amount obtained in step S505 as a memory of allocation destination candidate host. Save on 103.

  If each of the processes in steps S503 to S506 is completed for all hosts (S507), the resource reservation processing unit 110 advances the process to S508.

  In step S <b> 508, the resource reservation processing unit 110 determines the presence / absence of information of the allocation destination candidate host in the memory 103. As a result of the determination, when the information on the allocation destination candidate host does not exist in the memory 103 (S508: No), the resource reservation processing unit 110 proceeds with the process to step S512 to display a message indicating the reservation failure. Send to to finish the process. On the other hand, as a result of the above determination, if the information on the allocation destination candidate host exists in the memory 103 (S508: Yes), the resource reservation processing unit 110 advances the process to S509.

  In step S509, the resource reservation processing unit 110 arranges the host with the smallest average resource allocation planned amount obtained in step S505 among the allocation destination candidate hosts holding the information in the memory 103 with respect to the next VM as the allocation destination host. Decide on.

Subsequently, in step S510, the resource reservation processing unit 110 generates a record in the VM management information DB 125 and adds the reservation information indicated by the current resource reservation request received in step S501. Further, the resource reservation processing unit 110, in the resource assignment scheduled resource amount DB 127, shows the allocated resource amount indicated by the above resource reservation request for the record of the time 821 corresponding to the above-mentioned reservation period among The value obtained by multiplying the number of cores by the CPU frequency and the value of the memory) is added to the value of the allocated resource amount 822 and corrected.

  In step S511, the resource reservation processing unit 110 transmits a message indicating that the reservation is successful as the processing result up to this point, that is, the reservation result, to the operation terminal 200, and ends the processing.

--- An example of processing involved in VM migration ---
Next, the flow involved in the process of VM placement optimization will be described based on the drawings. FIG. 11 is a flowchart showing the processing procedure example 3 of the virtual machine operation support method in the present embodiment. In this case, the operation terminal 200 receives a resource usage prediction request from the infrastructure administrator, and transmits this to the operation management server via the network 11 (S421). For example, the operation terminal 200 may automatically execute this process every predetermined time, or it is recognized that the operation uneven distribution alert between the hosts in the operation management target system 1 is notified from the operation management server 100. It may be started by an instruction from the infrastructure administrator, and the process trigger is not limited.

  On the other hand, in response to the reception of the above-mentioned prediction request, the operation management server 100 executes a process of predicting the resource usage (S422), and returns the prediction result to the operation terminal 200 (S423). Details of the resource usage prediction process will be described later.

  On the other hand, the infrastructure administrator browses the above-mentioned prediction result of the resource usage from the operation management server 100 with the operation terminal 200, and among the hosts of the operation management target system 1, the presence or absence of a host expected to cause a resource shortage is Check. As a result, when recognizing the existence of a host expected to cause a resource shortage, this infrastructure administrator moves a predetermined VM of the corresponding host to another host using live migration technology before the resource shortage occurs. And load balancing to prevent resource shortages from occurring.

  The infrastructure administrator refers to the prediction result received by the operation terminal 200 in step S 423 described above, and determines the time to move with the VM to be moved. The operation terminal 200 which has received such a determination result by the input device or the like transmits this to the operation management server 100 via the network 11 (S424).

  On the other hand, the operation management server 100 evaluates a host as a migration destination candidate of the corresponding VM based on the migration target VM and migration time indicated by the judgment result received from the operation terminal 200 (S 425). It transmits to the operating terminal 200 (S426).

  On the other hand, the operation terminal 200 receives the above evaluation result from the operation management server 100, outputs it to a display device such as a display, and provides it to the infrastructure administrator for viewing. In this case, the infrastructure administrator browses the evaluation result received by the operation terminal 200, determines the host to which the VM is to move, and inputs a request for moving reservation of the VM to the operation terminal 200. The operation terminal 200 having received this transmits a request for moving reservation of the corresponding VM to the operation management server 100 (S427).

  In this case, based on the reservation request for VM migration received from the operation terminal 200, the operation management server 100 performs VM migration reservation processing appropriately using the existing technology such as live migration described above (S428).

  Here, details of the above-described step S 422, that is, the process of resource usage prediction will be described. FIG. 12 is a flowchart showing the processing procedure example 4 of the virtual machine operation support method in the present embodiment, and more specifically, is a flowchart showing processing by the resource usage prediction unit 111 in the operation management server 100.

  In this case, first, the resource usage prediction unit 111 of the operation management server 100 receives a prediction request by the infrastructure administrator from the operation terminal 200 (S601).

  Next, the resource usage prediction unit 111 executes a series of processes from the following step S603 to S604, that is, for all hosts included in the operation management target system 1 (S602). The method of selecting a host in this case is the same as that illustrated in FIG.

  Among these, in step S603, the resource usage prediction unit 111 extracts a record related to the target host from the stored information of the operation history DB 129, and applies the operation history information indicated by the corresponding record to a predetermined prediction algorithm. In the host, the resource usage of the prediction period from the present time to the future of the predetermined future is predicted, and the prediction result is stored in the resource usage prediction DB 130. As the above-mentioned algorithm adopted for such resource usage prediction, for example, it corresponds to a method generally used in time-series data analysis such as SARIMA (Seasonal Auto Regressive Integrated Moving Average) model, Holt-Winters method, etc. An algorithm may be used.

  Subsequently, in step S604, the resource usage prediction unit 111 extracts, from the VM management information DB 125, a record in which the start time and the end time of the operation time field 804 are included in the prediction period to be predicted in step S603 described above. Then, based on the value of the allocated resource amount 805 indicated by the corresponding record, the correction processing regarding the predicted value of the resource usage is performed. In this correction processing, when the presence of a VM scheduled to stop resource allocation can be specified from the record in the VM management information DB 125 in the above-described prediction period, the resource usage amount for the allocation suspension is subtracted from the predicted value. , And update the corresponding record of the resource usage forecast DB 130. Also, when resource allocation is started in the above-mentioned forecast period, that is, when the existence of a VM scheduled to start operation can be specified from the record of the VM management information DB 125, the resource usage amount for the assignment start is used as a forecast value. It adds and updates the applicable record of resource usage forecast DB130.

  When the processes of steps S603 to S604 described above have been completed for all hosts (S605), the resource usage prediction unit 111 advances the process to step S606.

In the loop started from step S606 , the resource usage prediction unit 111 is based on the information of the rack 812 of each host indicated by the host management information DB 126 (not shown in FIG. 4, but also includes the arrangement position in the rack). The ID of each host is arranged in the corresponding section of the matrix (drawing data held in advance in the storage device 101) according to the physical shelf configuration of each rack, and a heat map canvas is generated (S607) . Figure 14 shows a specific example of a heat map canvas 1010 that is generated by the resource use amount prediction unit 111 at step S60 7.

In the heat map canvas 1010 illustrated in FIG. 14, one mass in the matrix represents one host, and a character string described in the mass represents the ID of the host. Also, the bold line represents the range of the rack. In the example of the heat map canvas 1010 illustrated in FIG. 14, for example, hosts having IDs “H01” to “H08” are stored in the same rack.

Subsequently resource usage prediction unit 111, only the number of days of the forecast period, repeats the processing in steps S60 7 described above, to produce a heat map canvas 1010 number of days. Here, the case where the prediction period is 7 days will be described as an example. Therefore, it is assumed that seven heat map canvases 1010 are generated by repeating the process of the above-mentioned step S607 seven times from the first day to the seventh day.

Next, in step S608, the resource usage amount prediction unit 111, for each mass of the heat map canvas 1010 generated in step S60 7 described above, the size of the resource use amount prediction value for each host that obtained through step S604 The heat map 1020 is generated by selectively applying a predetermined color according to the height. FIG. 15 shows a specific example of the heat map 1020 generated in step S608.

  In this case, the resource usage prediction unit 111 calculates, for each host included in the operation management target system 1, the average value of the resource usage prediction values at the dates targeted by the heat map 1020, and the size of the average value. Select a predetermined color according to. For example, "red" when the resource usage predicted value is larger than the first reference value, "yellow" when it falls between the first reference value and the second reference value, and "blue" when it is smaller than the second reference value. Choose.

Subsequently, in step S610, the resource usage amount prediction unit 111, generated in step S609 from Step S60 6 above, generates display data for screen display heatmaps 1020 number of days of the forecast period, the prediction results which And transmit to the operation terminal 200. FIG. 16 shows an example of the screen data generated in step S610. The screen data example of FIG. 16 shows a form in which a heat map 1031 of a predetermined day designated by the time slider 1032 operated by the infrastructure administrator via the interface of the operation terminal 200 is displayed at the center of the screen. In this way, by displaying the heat map covering multiple days continuously and in an identifiable manner in one screen, the infrastructure administrator who browses this can easily identify the date and host at which resource usage increases. Becomes possible.

  Next, the process by the migration destination candidate evaluation unit 112 of the operation management server 100, that is, the process of step S425 described above will be described. FIG. 13 is a flowchart showing the processing procedure example 5 of the virtual machine operation support method in the present embodiment.

  In this case, it is assumed that the infrastructure administrator refers to the prediction result presented in the above-described step S 423 with the operation terminal 200 and determines the VM to be moved and the movement time. On the other hand, the operation terminal 200 that has received the determination items by the input device transmits the information of the VM to be moved and the movement time to the operation management server 100 in step S424 described above.

  On the other hand, the migration destination candidate evaluation unit 112 of the operation management server 100 receives the information on the migration target VM and migration start time specified by the infrastructure administrator via the operation terminal 200 (S701), and accordingly, the VM The management information DB 125 extracts a record related to the migration target VM, and acquires information included in the corresponding record (S702). Specifically, each value of the allocation resource amount 805 and the end time of the operation time 804 regarding the migration target VM is acquired and stored in the memory 103.

Next, the migration destination candidate evaluation unit 112 executes a series of processes from step S704 to S708 below for all the hosts included in the operation management target system 1 to migrate the migration target VM described above. The aptitude evaluation as the host above is performed (S703). The selection method of the processing target, ie, the evaluation target host is the same as that described in each of the above-mentioned flows.

  Among them, in step S704, the migration destination candidate evaluation unit 112 uses the operation start time received in step S701 as the time series data to be used in the subsequent processing from the resource usage prediction DB 130 in the operation time acquired in step S702. A series of resource usage forecast values (CPU usage 912 related to the relevant time, values in the memory 913) are extracted during the time to the end time.

  Next, in step S 705, the migration destination candidate evaluation unit 112 refers to the host management information DB 126 to acquire each value of the capacity 813 of the evaluation target host.

In step s 706, the destination candidate evaluating unit 112 calculates the resource amount of margin in the evaluation target host. The margin indicates how large the capacity value of the corresponding host is with respect to the resource usage predicted value indicated by the time-series data from the above-mentioned transfer start time to the operation end time.

Here, due to the capacity C of the evaluation target host acquired in step S705, the allocated resource amount R acquired in step S702, and the resource usage predicted data P _{i at} time i extracted in step S704, a margin at time i is obtained. The degree S _i can be defined by the following equation 1.
S _i = C−R−P _i ... Number 1

The time zone in which the margin S _i is a positive value represents that it is expected that sufficient resources can be allocated to the moving VM, and the larger the value, the higher the certainty. Conversely, a time zone in which the margin S _i is a negative value represents that it is likely that sufficient resources can not be allocated to the moving VM. The migration destination candidate evaluation unit 112 calculates the integrated value _{i i} S _i of the margin in the evaluation target period, and sets this as the evaluation value of the margin of the target host (S 707).

  Next, in step S 708, the migration destination candidate evaluation unit 112 evaluates the continuity D of the time in which resources can be secured. The continuity D is determined as follows.

If S _O 00, let D be the time at which S _O <0 first.

In the case of · _S O <0, the first in the _{S O} ≧ 0 time and D.

  When the value of the continuity D described above is large, it indicates that the time when resources can be secured (or can not be secured) is expected to last for a long time. Conversely, when the value of continuity D is small, it means that it is likely that the time for which resources can be secured (or can not be secured) will not last long.

After performing evaluation on all hosts by the series of processes from step S703 to step S709 (S709), the movement destination candidate evaluation unit 112 advances the process to step S710, and creates screen data displaying the evaluation result. , And transmits this to the operation terminal 200.
FIG. 17 shows an example of screen data generated in step S710. Here, the heat map 1040 is created by coloring the heat map canvas 1010 created in step S 606 based on the margin obtained in step S 707 and the continuity determined in step S 708, and this is used as screen data. Shows the form included in.

The heat map 1040 is obtained by selecting and setting a color for each block of the heat map canvas 1010 according to the size of the margin obtained for the corresponding host. For example, the moving destination candidate evaluation unit 112 “blue” when the margin is larger than the first standard, “yellow” when smaller than the first standard and larger than the second standard, and “red” when smaller than the second standard. Perform color selection operation etc. Furthermore, the migration destination candidate evaluation unit 112 determines the density of each color set in the corresponding cell according to the continuity obtained for each host. In this case, for example, the moving destination candidate evaluation unit 112 darkens the corresponding color as the continuity is higher than the predetermined reference, and lightens the corresponding color as the continuity is lower than the predetermined reference.

  As described above, by generating the heat map 1040 indicating the degree of margin and the continuity and transmitting it to the operation terminal 200 and displaying it, the infrastructure administrator who browsed this can perform the above-described operation at the time of selecting the destination host. The destination host may be selected according to the degree of continuity and the degree of margin, and the load of such work can be greatly reduced.

  In the example of the heat map 1040 illustrated in FIG. 17, the migration destination candidate evaluation unit 112 displays the movement target VM designated by the infrastructure administrator from the operation terminal 200 on the screen as a circle 1041 icon. A circle 1101 indicating a migration target VM is displayed in a matrix representing a host on which the VM is operating. As described above, the thick lines in the heat map 1040 represent racks. As a result, the infrastructure administrator can easily identify the migration target VM, the host including the migration target VM and the rack, and the hosts included in the rack. Therefore, when it is desired to move within the same rack considering network performance etc., it is possible to easily narrow down the destination hosts. In the example of FIG. 17, it is easily understood that the migration target VM is operating on the host “H32” and the hosts “H31” to “H38” exist in the rack.

  Furthermore, as described above, the color on the heat map 1040 represents the evaluation result of the destination host. Since the resource margin is expressed in color and continuity is expressed in color, the infrastructure administrator is displayed in dark blue when it is desired to select a host with sufficient resources for a long time as a destination. Find a host that is

  In this manner, the infrastructure administrator refers to the evaluation result transmitted from the operation management server 100 with the operation terminal 200 to select the destination host, and the selection result is received via the operation terminal 200. Send to

  The operation management server reserves the migration processing of the migration target VM to the migration destination host instructed from the operation terminal 200, and executes the VM migration at the time designated by the corresponding reservation. This VM migration may be performed by adopting the live migration already described.

  By the processing described above, the migration destination host in VM migration for load leveling can be evaluated from the viewpoint of resource margin and continuity, and the result can be presented to the user. A user who recognizes such an evaluation result can reduce the workload involved in selecting the migration destination host of the migration target VM. In addition, since the destination host can be determined based on the evaluation result in consideration of the continuity and the degree of margin, it becomes possible to efficiently and accurately select a host with sufficient resources for a long time, and VM migration is frequently performed. It is possible to suppress the occurrence and stabilize the performance.

  As mentioned above, although the best mode etc. for carrying out the present invention were concretely explained, the present invention is not limited to this, and can be variously changed in the range which does not deviate from the gist.

  According to the present embodiment, performance stabilization in a virtual machine can be efficiently realized.

At least the following matters will be made clear by the description of the present specification. That is, in the virtual machine operation support system of the present embodiment, the storage device holds at least reservation information of resources already allocated for each virtual machine as information on the physical server, and the arithmetic device In the determination of the continuity, the predetermined in each physical server based on information of the resource required for the predetermined virtual machine at a predetermined time and the reservation information for each physical server held by the storage device for each time A process of further determining the margin of resources to be secured for virtual machines with a predetermined algorithm, and displaying each information of the margin and the continuity obtained for each physical server by each determination on a predetermined device It may be executed.

  According to this, with regard to the physical server to which the virtual machine is to be moved, the information in charge of operation of the virtual machine is further taken into consideration, along with the continuity of the time required to secure the resources necessary for the virtual machine. Can be presented to The operation personnel who viewed this can easily and surely recognize whether or not the necessary resources can be secured continuously for the future in a predetermined period after movement, as well as at the time of movement study of the virtual machine. Thus, it is possible to accurately determine the destination of the virtual machine and its availability. As a result, performance stabilization in a virtual machine can be realized efficiently.

  Further, in the virtual machine operation support system of the present embodiment, the storage device holds at least history information of past resource usage as the information on the physical server, and the computing device determines the continuity. When performing this, apply history information of past resource usage held in the storage device to a predetermined algorithm, and execute prediction of future resource usage after the predetermined time for each physical server, Also, based on the predicted value of the future resource usage amount in each physical server and the information of the resource required for the predetermined virtual machine at the predetermined time, for each physical server, from the predetermined time to the future for the predetermined virtual machine The continuity of the time during which the corresponding resource can be secured may be determined by a predetermined algorithm.

  According to this, it is possible to present, to the person in charge of operation of the virtual machine or the like, the above-mentioned continuity based on the resource usage predicted accurately based on the past history. The person in charge of operation who has read this can more accurately determine the virtual machine destination and its availability. As a result, performance stabilization in a virtual machine can be realized efficiently.

  Further, in the virtual machine operation support system according to the present embodiment, the storage device includes, as information on the physical server, history information of resource usage in the past, and reservation information of resources already allocated at each time for each virtual machine. In the determination of the continuity, the arithmetic unit applies history information of the past resource usage held in the storage unit to a predetermined algorithm, and determines the future of the future after the predetermined time. Each virtual machine at the corresponding time of the corresponding physical server indicated by the reservation information with respect to the predicted value of the future resource usage of each physical server obtained by performing prediction of resource usage for each physical server and obtaining the prediction The correction is performed by applying the information on the use / non-use of the resource according to the predetermined virtual machine and the predetermined virtual machine after the correction. On the basis of the resource information required for the period, the continuity of the time in which the resource can be secured for the predetermined virtual machine from the predetermined time to the future is determined by a predetermined algorithm for each physical server It is also good.

  According to this, it is possible to present, to the person in charge of operation of the virtual machine or the like, the above-mentioned continuity based on the resource usage predicted more accurately based on the past history. The person in charge of operation who has read this can more accurately determine the virtual machine destination and its availability. As a result, performance stabilization in a virtual machine can be realized efficiently.

Further, in the virtual machine operation support system of the present embodiment, the storage device holds at least a display object indicating an arrangement position of each physical server in a predetermined space as the information on the physical server, and the arithmetic device During the process of displaying the continuity information on a predetermined device, the display data is read from the storage device, and the continuity indicated by the continuity information on the physical server for the area for each physical server in the display object is indicated The display mode may be set in accordance with the height and height, and the display object after the setting may be output to the predetermined device.

  According to this, the above-mentioned continuity in each physical server can be presented to a person in charge of operation of a virtual machine or the like in a form that can be visually recognized easily. The person in charge of operation who has browsed this can judge the virtual machine destination and its availability more efficiently and accurately. As a result, performance stabilization in a virtual machine can be realized efficiently.

  Further, in the virtual machine operation support system according to the present embodiment, the storage device includes, as information related to the physical server, reservation information of resources already allocated for each virtual machine at each time, and an arrangement position of each physical server in a predetermined space And at least the display object indicating the information, the computing device, when determining the continuity, information of the resource required by the predetermined virtual machine at a predetermined time, and each physical server held by the storage device The physical server further executes a process of determining the margin of the resource to be secured for the predetermined virtual machine by a predetermined algorithm based on the reservation information for each time related to In the process of displaying, the display object is read from the storage device, and the display object is For each physical server area to be displayed, the display form is set according to each of the continuity level indicated by the information on the continuity with respect to the physical server and the margin level indicated by the information on the margin, and display after the setting An object for output may be output to the predetermined device.

  According to this, it is possible to present, to a person in charge of operation of a virtual machine or the like, the above-mentioned continuity and allowance in each physical server in a form that can be visually recognized easily. The person in charge of operation who has browsed this can judge the virtual machine destination and its availability more efficiently and accurately. As a result, performance stabilization in a virtual machine can be realized efficiently.

  Further, in the virtual machine operation support method of the present embodiment, the information processing system holds, in the storage device, at least reservation information of resources already allocated for each virtual machine as information on the physical server. In each physical server, based on the information of the resource required by the predetermined virtual machine at a predetermined time and the reservation information for each physical server held by the storage device at the time of determining the continuity, A process of further determining the margin of resources to be secured for a predetermined virtual machine by a predetermined algorithm, and displaying each information of the margin and the continuity obtained for each physical server by each determination on a predetermined device May be performed.

  Further, in the virtual machine operation support method according to the present embodiment, the information processing system holds, in the storage device, at least history information of past resource usage as information on the physical server, and the continuity is In the determination, the history information of the past resource usage held in the storage device is applied to a predetermined algorithm, and the prediction of the future resource usage after the predetermined time is executed for each physical server, and Based on the obtained predicted value of the future resource usage of each physical server and the information of the resource required for the predetermined virtual machine at the predetermined time, for each physical server, the future for the predetermined virtual machine from the predetermined time for the predetermined virtual machine The continuity of the time during which the corresponding resource can be secured may be determined by a predetermined algorithm.

Further, in the virtual machine operation support method of the present embodiment, the information processing system in the storage device has already assigned history information of the past resource usage as information on the physical server and each virtual machine for each time. And at least the reservation information of the resource, and when determining the continuity, applying the history information of the past resource usage amount held by the storage device to a predetermined algorithm, the future after the predetermined time The resource usage forecast of each physical server is executed for each virtual server, and each virtual server at the corresponding time of the corresponding physical server indicated by the reservation information with respect to the forecast value of the future resource usage of each physical server obtained by the forecast The correction is performed by applying the information on the use of the resource by the machine, and the predicted value after the correction and the predetermined virtual machine Even if the continuity of the time during which the resource can be secured for the predetermined virtual machine from the predetermined time to the future can be determined for each physical server based on the information of the resource required for the predetermined time using a predetermined algorithm Good.

  Further, in the virtual machine operation support method according to the present embodiment, the information processing system holds, in the storage device, at least a display object indicating an arrangement position of each physical server in a predetermined space as the information on the physical server. In the process of displaying the continuity information on a predetermined device, the display data is read from the storage device, and the continuity information on the physical server indicates the area for each physical server in the display object. The display mode may be set according to the degree of continuity, and the display object after the setting may be output to the predetermined device.

  Further, in the virtual machine operation support method according to the present embodiment, the information processing system in the storage device includes, as information related to the physical server, reservation information of resources already allocated for each virtual machine in a predetermined space. It holds at least a display object indicating the arrangement position of each physical server, and when determining the continuity, information of the resource required by the predetermined virtual machine at a predetermined time, and each physical value held by the storage device Based on the reservation information for each time of the server, the physical server further executes processing of determining the margin of the resource to be secured for the predetermined virtual machine by a predetermined algorithm, and the continuity information is determined by the predetermined device During the display process, the display object is read from the storage device, and the display object is displayed. The display form is set according to each of the continuity level indicated by the continuity information about the physical server and the margin level indicated by the margin information for the area for each physical server in the event. The display object may be output to the predetermined device.

1 operation management target system 10 network 20, 30 racks 21-23, 31-33 hosts 25-27, 35-37 virtual machine 100 operation management server (virtual machine operation support system)
101 storage device 102 program 103 memory 104 CPU (arithmetic unit)
105 communication device 110 resource reservation processing unit 111 resource usage prediction unit 112 migration destination candidate evaluation unit 125 VM management information DB
126 Host Management Information DB
127 Allocated resource amount DB
128 overcommit rate upper limit DB
129 Operation history DB
130 Resource Usage Forecast DB
200 operation terminal

Claims (10)

A storage device holding at least reservation information of resources allocated for each virtual machine at each time as information on each physical server to which the virtual machine is to be allocated ;
For each physical server, the information for each physical server is obtained for the predetermined virtual machine based on the information on the resource required for the predetermined virtual machine from the predetermined device regarding the predetermined virtual machine and the information on each physical server held by the storage device. a process of determining at a predetermined algorithm the continuity of time can be secured to the appropriate resources from a predetermined timing, a predetermined virtual machine resources required for the predetermined time information, for each timing for each physical server that the storage device is held A process of determining, with a predetermined algorithm, an allowance of resources to be secured for the predetermined virtual machine in each physical server based on the reservation information, the allowance and the continuity obtained for each physical server by each determination A process of displaying each piece of information on a predetermined device ;
A virtual machine operation support system comprising: The storage device is
At least history information on past resource usage is held as the information on the physical server,
The arithmetic device is
In determining the continuity, history information of past resource usage held in the storage device is applied to a predetermined algorithm, and prediction of future resource usage after the predetermined time is executed for each physical server. Each physical server for the predetermined virtual machine based on the predicted value of the future resource usage of each physical server obtained by the prediction and the information of the resource required for the predetermined virtual machine at the predetermined time The continuity of the time during which the corresponding resource can be secured from the predetermined time to the future is determined by a predetermined algorithm,
The virtual machine operation support system according to claim 1, characterized in that: The storage device is
As information on the physical server, at least history information on resource usage in the past and reservation information on resources already allocated for each virtual machine at each time are held,
The arithmetic device is
In determining the continuity, history information of past resource usage held in the storage device is applied to a predetermined algorithm, and prediction of future resource usage after the predetermined time is executed for each physical server. Correcting the predicted value of the future resource usage of each physical server obtained by the prediction by applying the information on the use / non-use of the resource by each virtual machine at the corresponding time of the corresponding physical server indicated by the reservation information For each physical server based on the predicted value after the correction and the resource information required by the predetermined virtual machine for the predetermined time from the predetermined time to the future for the predetermined virtual machine The continuity of the time that can be secured is determined by a predetermined algorithm,
The virtual machine operation support system according to claim 1, characterized in that: The storage device is
As information on the physical server, at least a display object indicating the arrangement position of each physical server in a predetermined space is held,
The arithmetic device is
In the process of displaying the continuity information on a predetermined device, the display object is read from the storage device, and the continuity indicated by the continuity information on the physical server for the area for each physical server in the display object is indicated. The display mode is set according to the height, and the display object after the setting is output to the predetermined device.
The virtual machine operation support system according to claim 1, characterized in that: The storage device is
As information on the physical server, it holds at least reservation information of resources already allocated for each virtual machine with respect to each virtual machine, and a display object indicating an arrangement position of each physical server in a predetermined space,
The arithmetic device is
In the determination of the continuity, the predetermined virtual in each physical server based on information of the resource required for the predetermined virtual machine at a predetermined time and the reservation information for each physical server held by the storage device for each time Execute further processing to determine the degree of margin of resources to be secured for machines by a predetermined algorithm,
In the process of displaying the continuity information on a predetermined device, the display object is read from the storage device, and the continuity indicated by the continuity information on the physical server for the area for each physical server in the display object is indicated. The display form is set according to each of the high and low and the degree of margin indicated by the information of the degree of margin, and the display object after the setting is outputted to the predetermined device.
The virtual machine operation support system according to claim 1, characterized in that: An information processing system comprising a storage device holding at least reservation information of resources allocated for each virtual machine at each time as information on each physical server to which the virtual machine is to be allocated ,
For each physical server, the information for each physical server is obtained for the predetermined virtual machine based on the information on the resource required for the predetermined virtual machine from the predetermined device regarding the predetermined virtual machine and the information on each physical server held by the storage device. a process of determining at a predetermined algorithm the continuity of time can be secured to the appropriate resources from a predetermined timing, a predetermined virtual machine resources required for the predetermined time information, for each timing for each physical server that the storage device is held A process of determining, with a predetermined algorithm, an allowance of resources to be secured for the predetermined virtual machine in each physical server based on the reservation information, the allowance and the continuity obtained for each physical server by each determination Displaying each piece of information on a predetermined device;
A virtual machine operation support method characterized by The information processing system
The storage device holds at least history information of past resource usage as the information on the physical server,
In determining the continuity, history information of past resource usage held in the storage device is applied to a predetermined algorithm, and prediction of future resource usage after the predetermined time is executed for each physical server. Each physical server for the predetermined virtual machine based on the predicted value of the future resource usage of each physical server obtained by the prediction and the information of the resource required for the predetermined virtual machine at the predetermined time Determine the continuity of the time during which the corresponding resource can be secured from a predetermined time to the future by a predetermined algorithm,
The virtual machine operation support method according to claim 6, characterized in that: The information processing system
The storage device holds, as information on the physical server, at least history information on past resource usage and reservation information on resources already allocated for each virtual machine at each time,
In determining the continuity, history information of past resource usage held in the storage device is applied to a predetermined algorithm, and prediction of future resource usage after the predetermined time is executed for each physical server. Correcting the predicted value of the future resource usage of each physical server obtained by the prediction by applying the information on the use / non-use of the resource by each virtual machine at the corresponding time of the corresponding physical server indicated by the reservation information For each physical server based on the predicted value after the correction and the resource information required by the predetermined virtual machine for the predetermined time from the predetermined time to the future for the predetermined virtual machine Determine the continuity of the time that can be secured by a predetermined algorithm,
The virtual machine operation support method according to claim 6, characterized in that: The information processing system
In the storage device, at least a display object indicating an arrangement position of each physical server in a predetermined space is held as the information on the physical server,
In the process of displaying the continuity information on a predetermined device, the display object is read from the storage device, and the continuity indicated by the continuity information on the physical server for the area for each physical server in the display object is indicated. The display mode is set according to the height, and the display object after the setting is output to the predetermined device.
The virtual machine operation support method according to claim 6, characterized in that: The information processing system
The storage device holds, as information related to the physical server, at least reservation information of resources allocated for each virtual machine at each time and a display object indicating an arrangement position of each physical server in a predetermined space. ,
In the determination of the continuity, the predetermined virtual in each physical server based on information of the resource required for the predetermined virtual machine at a predetermined time and the reservation information for each physical server held by the storage device for each time Execute further processing to determine the degree of margin of resources to be secured for machines by a predetermined algorithm,
In the process of displaying the continuity information on a predetermined device, the display object is read from the storage device, and the continuity indicated by the continuity information on the physical server for the area for each physical server in the display object is indicated. The display form is set according to each of the high and low and the degree of margin indicated by the information of the degree of margin, and the display object after the setting is outputted to the predetermined device.
The virtual machine operation support method according to claim 6, characterized in that: