JP6939327B2 - Programs and information processing equipment - Google Patents

Description

The present invention relates to a program and an information processing device.

In Patent Document 1, the migration source server, the migration target virtual machine, and the migration destination server are determined based on the load information of the virtual machines between a plurality of servers, and the migration source server and migration after the preparation for dynamic migration is completed. A technique for determining whether or not dynamic migration can be executed based on the load information of the destination server is disclosed.

Japanese Patent No. 5549189

In the cloud service, a virtual machine called an instance that can execute a plurality of various services (image processing service, etc.) used by the user at the same time is provided, and may be charged according to the number of instances in use. For example, if the instance that is the execution subject of the service is fixed, if even one service that is running remains, the use of the instance cannot be terminated unless the service is terminated, so reduce the number of instances. I can't.
Therefore, an object of the present invention is to reduce the number of virtual machines that execute a service as compared with the case where the execution subject of the service is fixed.

The program according to claim 1 of the present invention provides a computer with free resources capable of executing one or more services from a plurality of virtual machines each executing a plurality of services using a first resource assigned to the computer. The detection unit that detects the first virtual machine and the detected resource usage rate of the first virtual machine of the second virtual machine that is executing a service that can be executed by the free resource among the plurality of virtual machines. When the value is equal to or higher than the threshold value determined from the resource usage rate, the first virtual machine is made to function as a determination unit for determining the transfer destination of the service.

In the configuration according to claim 1, the program according to claim 2 of the present invention has the determination unit, in the case where there are a plurality of the second virtual machines that are candidates for the transfer source of the service, the resource after the transfer of the service. It is characterized in that the transfer destination of the service whose transfer source is the second virtual machine whose usage rate is smaller than that of other candidates for the transfer source is determined.

In the configuration according to claim 1 or 2, the program according to claim 3 of the present invention is a service in which, when there are a plurality of transferable services, the processing being executed is completed earlier than other services. It is characterized by determining the relocation destination of.

In the configuration according to any one of claims 1 to 3, the program according to claim 4 of the present invention has a service in which the service that can be executed by the free resource is different from that of the second virtual machine. When the service is made redundant by using a virtual machine, the transfer destination is not determined for the service.

The program according to claim 5 of the present invention has the configuration according to any one of claims 1 to 4, wherein the computer is assigned to one or more running services. The first calculation unit that calculates each of the plurality of virtual machines with the first ratio to the resource usage rate as the resource usage rate, and the third resource allocated to the running service excluding the service scheduled to be transferred to another virtual machine. The second ratio to the first resource is used as the resource usage rate to function as a second calculation unit for calculating each of the plurality of virtual machines, and the determination unit is determined from the second ratio of the second virtual machine. It is characterized in that the first virtual machine in which the second ratio equal to or higher than the threshold value is calculated is determined as the transfer destination of the service that can be executed by the free resource.

In the configuration according to claim 5, the program according to claim 6 of the present invention has the determination unit, in the case where there are a plurality of the second virtual machines that are candidates for the transfer source of the service, the first after the transfer of the service. It is characterized in that the transfer destination of the service running in the second virtual machine whose ratio is smaller than the other candidates of the transfer source is determined.

In the program according to claim 7 of the present invention, in the configuration according to claim 5 or 6, the determination unit transfers the service to the free resource when there are a plurality of the first virtual machines as candidates for the transfer destination. It is characterized in that the first virtual machine having a later second ratio larger than that of other candidates for the relocation destination is determined as the relocation destination.

The program according to claim 8 of the present invention causes the computer to function as a storage unit for storing the transfer history of services in the plurality of virtual machines in the configuration according to any one of claims 1 to 7. The determination unit is characterized in that the threshold value is changed based on the stored transfer history.

In the program according to claim 9 of the present invention, in the configuration according to claim 8, the determination unit determines the threshold value as the frequency of transfer or the number of transfers indicated by the stored transfer history of the first virtual machine decreases. It is characterized by making the size smaller.

In the configuration according to claim 8, the program according to claim 10 of the present invention makes the threshold value smaller as the latest transfer time indicated by the stored transfer history of the first virtual machine is older. It is characterized by that.

In the configuration according to any one of claims 1 to 10, the program according to claim 11 of the present invention determines the transfer destination by using any one of a plurality of methods, and the plurality of determination units determine the transfer destination. Among the methods described in the above, the method in which the number of services executed by the second virtual machine becomes 0 by transferring the service to the determined transfer destination or the method having a large number of times is preferentially used. ..

In the configuration according to any one of claims 1 to 11, the program according to claim 12 of the present invention has the determination unit when the process executed by the service that can be executed by the free resource is completed. When the resource usage rate of the first virtual machine is less than the threshold value, another virtual machine is determined as the transfer destination of the service.

The program according to claim 13 of the present invention stores the computer as a storage unit for storing the transfer history of services in the plurality of virtual machines in the configuration according to any one of claims 1 to 12. It is characterized in that it functions as a history output unit that outputs the transfer history.

The program according to claim 14 of the present invention is a state output unit that outputs information indicating an execution state of a service in the plurality of virtual machines to the computer in the configuration according to any one of claims 1 to 13. It is characterized by functioning as.

The information processing device according to claim 15 of the present invention has a free resource capable of executing one or more services from a plurality of virtual machines, each of which executes a plurality of services using a first resource assigned to the device. The resource usage of the detection unit that detects the first virtual machine and the detected resource usage rate of the first virtual machine are the resources of the second virtual machine that is executing the service that can be executed by the free resource among the plurality of virtual machines. When the value is equal to or higher than the threshold value determined from the usage rate, the first virtual machine is provided with a determination unit for determining the transfer destination of the service.

According to the inventions according to claims 1, 5 and 15, the number of virtual machines that execute the service can be reduced as compared with the case where the execution subject of the service is fixed.
According to the inventions of claims 2 and 6, the free resource can be biased to a specific virtual machine as compared with the case where another virtual machine is used as the transfer source.
According to the invention of claim 3, the transfer time of the service can be advanced as compared with the case where the service is transferred regardless of the end time of the processing.
According to the invention of claim 4, it is possible to reduce the recovery error of the service as compared with the case of determining the transfer destination of the redundant service.
According to the invention of claim 7, the service can be biased to a specific virtual machine as compared with the case where another virtual machine is used as the transfer destination.
According to the invention of claim 8, it is possible to make it easier for the transferred service to settle in the transfer destination as compared with the case where the threshold value is fixed.
According to the invention of claim 9, it is possible to make it easier for the transferred service to settle in the transfer destination as compared with the case where the virtual machine having a high frequency or frequency of transfer is set as the transfer destination.
According to the invention of claim 10, it is possible to make it easier for the transferred service to settle in the relocation destination as compared with the case where the latest relocation time is the new virtual machine as the relocation destination.
According to the invention of claim 11, the number of virtual machines that execute the service can be reduced as compared with the case where the method of determining the transfer destination is fixed.
According to the invention of claim 12, the transfer of the service to the instance that is no longer suitable as the transfer destination can be suppressed as compared with the case where the transfer destination is not reviewed.
According to the invention of claim 13, the history of past relocation in each virtual machine can be grasped.
According to the invention of claim 14, the transfer status can be grasped from the execution state of the service.

Diagram showing the overall configuration of the resource provision system according to the embodiment Diagram showing the hardware configuration of the server device Diagram showing the hardware configuration of the resource management device Diagram showing the functional configuration realized by the resource provision system Diagram showing the functional configuration realized by the resource provision system Diagram showing an example of determining the relocation destination Diagram showing an example of determining the relocation destination Diagram showing examples of multiple transfer source candidates Diagram showing examples of multiple transfer destination candidates The figure which shows an example of the operation procedure of each device in the transfer process. Diagram showing an example of a threshold table Diagram showing an example of the decision result of the relocation destination Diagram showing another example of a threshold table Diagram showing another example of a threshold table The figure which shows an example of the operation procedure of each apparatus in the transfer process of a modification Diagram showing a decision example of the service to be transferred

[1] Example FIG. 1 shows the overall configuration of the resource providing system 1 according to the embodiment. The resource providing system 1 is a system for implementing a resource providing service that allows a user to use resources used for executing processing (information processing). The resources referred to here include a CPU (Central Processing Unit), memory, storage, a physical device including them, and a virtual device (so-called virtual machine) realized by them.

Further, in the resource providing system 1, a virtual machine called an instance capable of simultaneously executing a plurality of various services (OCR (Optical Character Recognition) service, translation service for translating sentences, etc.) used by the user is provided as a resource. .. In this embodiment, the user is charged according to the number of instances in use.

The resource providing system 1 includes a communication line 2, a user terminal 3, a cloud system 4, and a resource management device 20. The communication line 2 includes, for example, the Internet, a mobile communication network, a telephone line, and the like, and mediates communication between devices connected to the own line. The user terminal 3 and the cloud system 4 are connected to the communication line 2.

The user terminal 3 is a terminal used by a user who uses the resource providing service described above. The user includes, for example, a person who uses the resource providing service as a business of a company and a person who uses the resource providing service for personal activities (for example, playing an online game). Although only one user terminal 3 is shown in FIG. 1, it is assumed that a plurality of user terminals 3 are normally used.

The user terminal 3 requests the cloud system 4 to execute the service described above by the user's operation, receives the execution result of the requested service, and displays, for example, a screen corresponding to the execution result. Execution of a service by the cloud system 4 means executing one or more processes corresponding to each service (in the case of an OCR service, a character area extraction process, a character cutout process, a character recognition process, etc.).

The cloud system 4 includes a plurality of server devices 10 that are resources for executing processing, and uses those resources to execute the service requested by the user terminal 3. Each server device 10 realizes the above-mentioned instance. The resource management device 20 is a device for the administrator of the resource providing system 1 to manage the operating status of the resource, and in this embodiment, the execution status of the service in each instance and its history are displayed.

FIG. 2 shows the hardware configuration of the server device 10. The server device 10 is a computer including a CPU 11, a RAM (Random Access Memory) 12, a ROM (Read Only Memory) 13, a communication unit 14, and a storage 15. The CPU 11 controls the operation of each part by executing a program stored in the ROM 13 or the storage 15 using the RAM 12 which is a memory as a work area. The communication unit 14 has a communication circuit or the like, and communicates with an external device via the communication line 2. The storage 15 is a storage means such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), and a flash memory, and stores data and programs used by the CPU 11 for control.

FIG. 3 shows the hardware configuration of the resource management device 20. The resource management device 20 is a computer including a CPU 21, a RAM 22, a ROM 23, a communication unit 24, a storage 25, and a UI unit 26 (User Interface). The CPU 21 to the storage 25 are the same hardware as the hardware of the same name shown in FIG. The UI unit 26 includes display means such as a liquid crystal display, and displays a menu screen or the like for operating the own device. In addition, the UI unit 26 includes input means such as a keyboard and a mouse.

The functions described below are realized by the CPU of each device included in the resource providing system 1 executing a program to control each unit.
4 and 5 show the functional configuration realized by the resource providing system 1. Although only one server device 10 is shown in FIG. 4, each server device 10 has a virtual machine realization unit 30, an instance group 31, a first autoscale control unit 32, and a second autoscale control. A unit 33, a request / response queue 34, a service definition storage unit 35, a service management information storage unit 36, and a transfer control unit 40 are provided.

As shown in FIG. 5, the transfer control unit 40 includes a service status management unit 401, a service status storage unit 402, a resource usage rate calculation unit 403, a service transfer determination unit 404, a transfer target selection unit 405, and a service. It includes a stop determination unit 406 and a service transfer information storage unit 407. The resource management device 20 includes a service management information display unit 201. The instance group 31 includes a plurality of instances such as instance 50-1, instance 50-2, and instance 50-3. Hereinafter, when they are not distinguished, they are referred to as "instance 50".

Instance 50-1 includes a container control request receiving unit 51-1, a container execution unit 52-1 and a container control unit 53-1. Instance 50-2 includes a container control request receiving unit 51-2, a container execution unit 52-2, and a container control unit 53-2. In addition, each instance after instance 50-3 has a function common to instance 50-1 and instance 50-2. When the functions of each instance 50 are not distinguished, they are referred to as "container control request receiving unit 51", "container execution unit 52", and "container control unit 53", respectively.

In this embodiment, the case where the transfer destination of the service executed in the instance 50-1 is the instance 50-2 will be mainly described (however, the case where the instance 50-2 is not the transfer destination will also be described). When instance 50-1 is the transfer source of the service 60 and instance 50-2 is the transfer destination, the instance 50-2 is an example of the "first virtual machine" of the present invention, and the instance 50-1 is the "first virtual machine" of the present invention. This is an example of a "second virtual machine".

The virtual machine realization unit 30 is a function of realizing a virtual machine, and in this embodiment, a virtual machine using a container-type virtualization technology is realized. The virtual machine realization unit 30 is realized by, for example, Docker (registered trademark), but is not limited to this, and may be realized by other container-type virtualization programs. In the example of FIG. 4, the virtual machine realization unit 30 also realizes the first autoscale control unit 32, the transfer control unit 40, and the like, which are functions other than the instances, in addition to each instance such as the instance 50-1.

The container control request receiving unit 51 receives a control request for the container from the outside of its own instance, and transmits the request content to the container control unit 53. The container execution unit 52 executes a plurality of services such as the OCR service described above by using the resources allocated to the own instance. Allocating resources to an instance means allocating resources for that instance. Within the instance, resources are allocated (allocated) to each service in the same way.

In the example of FIG. 5, the container execution unit 52-1 executes three services, service 61-1, service 61-2, and service 61-3, and the container execution unit 52-2 of the instance 50-2 , Service 62-1, service 62-2, and service 62-3. In the following, when each service is not distinguished, it is referred to as "service 60".

In the resource providing system 1, each instance 50 and each service 60 are identified by their respective identification information (instance ID (Identification), service ID). In the following, it is assumed that the information regarding each instance 50 and each service 60 is always given the identification information thereof, so that the information of which instance 50 and service 60 can be known.

The request / response queue 34 handles a request to the service 60 from the user terminal 3 shown in FIG. 1 and a response from the service 60 to the request as a queue (first-in first-out data). The service 60 receives a message indicating a request from the request / response queue 34, executes the process specified by the message, and stores the message indicating the process result as a response in the request / response queue 34.

The container control unit 53 controls the execution of the service 60 by the container execution unit 52. The container control unit 53 allocates the service 60 to the container execution unit 52, stops the service 60 being executed by the container execution unit 52, allocates resources necessary for executing the service 60, and allocates more resources than the allocated resources. It limits the use of the service 60 and manages the execution status of the service 60.

The arrangement of the service 60 means to secure the resource of the container execution unit 52 and create the execution environment of the service 60. The service is arranged based on the service definition information stored in the service definition storage unit 35. The service definition information is information indicating programs, data, parameters, resources used, and the like used in the service 60.

The transfer control unit 40 is a function of controlling the operation related to the transfer of the service 60, and operates independently of each instance 50. In the instance group 31, the service 60 is transferred between the instances. In the present embodiment, the transfer of the service 60 means that the service 60 executed in the transfer source instance is executed in the transfer destination instance, and the service 60 executed in the transfer source instance is stopped.

The service status management unit 401 of the transfer control unit 40 acquires and manages the status of the service 60 in the container execution unit 52 of each instance 50. The state of service 60 includes three states of being created, being executed, and being prepared for relocation. “Creating” means that the container execution unit 52 is in the process of creating the execution environment of the service 60, and the service 60 being created cannot execute the process.

"Running" is a state in which the creation of the execution environment of the service 60 has been completed and the container execution unit 52 can execute the process. Even if there is no request and the process is not executed, it will be in the running state if the creation of the execution environment is completed. When the state of the service 60 at the transfer destination changes from being created to being executed, the transfer control unit 40 stops the service 60 at the transfer source as will be described later. In this way, the transfer is completed when the service 60 of the transfer source is stopped.

"Preparing for relocation" is a state in which the transfer is scheduled to be performed on another instance, and the same service 60 is being created in the transfer destination instance. In this embodiment, if there is a service 60 that is being prepared for relocation, that is, a service 60 that has been decided to be relocated to another instance, the container execution unit 52 waits until the execution environment of the service 60 is completed in the relocation destination instance. (That is, until the state of the service 60 becomes "running"), the service 60 is made to execute a new process.

In the above case, the container execution unit 52 may prevent the service 60 from executing a new process even if the execution environment of the service 60 is not completed in the transfer destination instance. When a new process is executed, the progress of the process is faster than when the new process is not executed, and when the new process is not executed, the service is transferred compared to the case where the new process is executed. Will be hastened.

When the state of the service 60 is changed, the service 60 is stopped, the service 60 in the "running" state starts processing, or the processing ends, the container execution unit 52 may perform the processing. The service status management unit 401 is notified to that effect via the container control unit 53 and the container control request reception unit 51. Upon receiving this notification, the service status management unit 401 stores the status of the service 60 in the service status storage unit 402, and updates the status if the status of the service 60 is already stored. In this way, the service state storage unit 402 stores the state of the service 60 of each instance 50.

The resource usage rate calculation unit 403 calculates the resource usage rate of each instance 50. The resource usage rate of the instance 50 means the ratio of the resources allocated to the service 60 running on the instance 50 among the resources allocated to the instance 50. The resource usage rate calculation unit 403 calculates the current usage rate and the usage rate after the transfer as the resource usage rate.

The current utilization is the resource utilization for the resources currently assigned to the service 60 currently running on instance 50. The service 60 currently being executed is the service 60 in the above-mentioned three states (creating, executing, and preparing for relocation). The service 60 being created is also included as being in the process of execution. The resource usage rate calculation unit 403 acquires information indicating the status of the service 60 in each instance 50 from the service status management unit 401.

The resource usage rate calculation unit 403 determines the ratio (hereinafter referred to as “first resource”) of the resources allocated to the service 60 in these three states (hereinafter referred to as “second resource”) to the resources allocated to the instance 50 (hereinafter referred to as “first resource”). (Hereinafter referred to as "first ratio") is used as the current usage rate, and is calculated for each of the plurality of instances 50. The resource usage rate calculation unit 403 for calculating the first ratio is an example of the “first calculation unit” of the present invention.

The usage rate of the instance 50 after the transfer is the service 60 running on the instance 50 minus the service 60 scheduled to be transferred to another instance 50, that is, the service 60 scheduled to be transferred is executed after the service 60 is transferred. The resource usage rate for the expected service 60. The service 60 that is expected to be executed after the relocation is the service 60 that is being created and is being executed, excluding the service 60 that is being prepared for relocation, out of the three states.

The resource usage rate calculation unit 403 determines the ratio (hereinafter referred to as “third resource”) of the resource allocated to the service 60 that is expected to be executed after the transfer to the first resource (resource allocated to the instance). (Hereinafter referred to as "second ratio") is used as the usage rate after the transfer, and is calculated for each of the plurality of instances 50. The resource usage rate calculation unit 403 for calculating the second ratio is an example of the “second calculation unit” of the present invention.

When the service transfer determination unit 404 instructs the resource usage rate calculation unit 403 to calculate the resource usage rate, the resource usage rate calculation unit 403 reads the service status of each instance 50 from the service status storage unit 402 and calculates the resource usage rate. The resource usage rate calculation unit 403 supplies the calculated resource usage rate of the instance 50 to the service transfer determination unit 404.

The service transfer determination unit 404 makes a determination regarding the transfer of the service 60. In this embodiment, the service transfer determination unit 404 makes this determination at a predetermined time interval. When this time interval elapses, the service transfer determination unit 404 instructs the resource usage rate calculation unit 403 to calculate the resource usage rate as described above, and the resource usage rate of each instance 50 (current usage rate and after transfer). Usage rate).

Based on the acquired resource usage rate, the service transfer determination unit 404 first, from a plurality of instances 50 (a plurality of virtual machines each executing a plurality of services using the resources assigned to itself), one or more. Instance 50 having free resources that can execute the service is detected as a transfer destination candidate. The service transfer determination unit 404 is an example of the "detection unit" of the present invention.

In the following, for the sake of clarity, it is assumed that the resources assigned to each instance 50 are equal and the resources assigned to each service 60 are also equal and all are one-third of the first resource. do. That is, a maximum of three services 60 are executed in any instance 50. It should be noted that these assumptions are for convenience of explanation only, and the present invention is not limited thereto.

The service transfer determination unit 404 uses, for example, at least one service 60 when the acquired current usage rate is the usage rate when executing two or less services 60 (that is, 66.6% or less). Since there are enough free resources to execute, the instance 50 is detected as a transfer destination candidate.

Further, the service transfer determination unit 404 acquires the status of the service 60 of each instance 50 from the service status storage unit 402, and based on the acquired status, the service 60 that is a candidate for transfer (the service that is actually transferred). 60 candidates) are detected. In this embodiment, the service transfer determination unit 404 selects the service 60 that has not executed the process (that is, the service 60 that has finished the process) among the services 60 that are in the “executing” state. Detect as a candidate for transfer.

Further, the service transfer determination unit 404 detects the instance 50 executing the detected service 60 of the transfer target candidate as the transfer source candidate. The service transfer determination unit 404 uses the resource usage rate (current usage rate and post-relocation usage rate) obtained from the identification information indicating the transfer target candidate, transfer source candidate, and transfer destination candidate detected in this way as the transfer target selection unit. Supply to 405. The transfer target selection unit 405 selects the transfer target candidate, the transfer source candidate, and the transfer destination candidate indicated by the supplied identification information to be actually transferred.

The transfer target selection unit 405 is executing the service 60 (transfer target candidate) whose resource usage rate of the detected instance 50 (transfer destination candidate) can be executed by the free resource of the transfer destination candidate among the plurality of instances 50. If it is equal to or more than the threshold value determined from the resource usage rate of the instance 50 (transfer source candidate) of the above, the transfer destination candidate is determined as the transfer destination of the transfer target candidate. The transfer target selection unit 405 is an example of the “decision unit” of the present invention.

In this embodiment, the service transfer determination unit 404 uses the usage rate after transfer as the resource usage rate to be compared, and uses the usage rate after transfer of the transfer source candidate as the threshold value. That is, the service transfer determination unit 404 determines the transfer destination candidate for which the usage rate after the transfer (threshold value in this embodiment) or more after the transfer of the transfer source candidate is calculated is calculated as the transfer destination. The method of determining the relocation destination will be described with reference to FIGS. 6A and 6B.

6A and 6B show an example of determining the relocation destination. In the example of FIG. 6, for the sake of simplicity, it is assumed that only instances 50-1 and 50-2 are realized. In FIG. 6A (a), services 61-1 and 61-2 are being executed and processing is being executed (processing) in instance 50-1, and services 61-3 are being executed. However, the processing is completed (processing is completed). Further, in instance 50-2, services 62-1, 62-2, and 62-3 are in the state of being executed and being processed.

In this example, the service transfer determination unit 404 detects the service 61-3 in the processed end state as a transfer target candidate, and detects the instance 50-1 that executes the service 61-3 as a transfer source candidate, but the instance 50. -2 does not detect instance 50-2 as a transfer destination candidate because there is no free resource capable of executing service 61-3. Therefore, service 61-3 is not transferred to instance 50-2.

In FIG. 6A (b), service 61-1 and service 62-1 are stopped from the state of FIG. 6A (a) to generate free resources. In this case, the service transfer determination unit 404 detects as a transfer destination candidate because the instance 50-2 has a free resource capable of executing the service 61-3. In the transfer target selection unit 405, the usage rate (66.6.%) After the transfer of the transfer destination candidate (instance 50-2) is the post-relocation usage rate (66.6) of the transfer source candidate (instance 50-1). ..%) or more, so instance 50-2 is decided as the transfer destination.

As described above, in this embodiment, the service is sent to either of the instances having the same usage rate after the transfer. By biasing the service execution subject to a specific instance 50 in this way, on the contrary, the number of instances 50 that do not become the service execution subject, that is, the instances 50 that may be stopped increases, and those instances 50 are stopped and used. Billing is reduced according to the number of instances.

In FIG. 6A (c), service 62-2 is changing from the state of FIG. 6A (b) during preparation for relocation. In this case, the service transfer determination unit 404 detects the instance 50-2 as a transfer destination candidate because the current usage rate of the instance 50-2 indicates a vacancy for executing the service 61-3. However, the post-relocation usage rate (33.3%) of the transfer destination candidate (instance 50-2) is less than the post-relocation usage rate (66.6 ...%) of the transfer source candidate (instance 50-1). Therefore, the transfer target selection unit 405 does not determine the instance 50-2 as the transfer destination.

In FIG. 6A (d), services 61-1 and 61-2 are changing from the state of FIG. 6A (c) during preparation for relocation. In this case, the current usage rate of the relocation candidate (66.6%) is less than the current usage rate of the relocation source candidate (100%), but the post-relocation usage rate of the relocation candidate (33.3 ...%). %) Is equal to or higher than the post-relocation usage rate of the transfer source candidate (33.3 ..%), so the transfer target selection unit 405 determines the instance 50-2, which is the transfer destination candidate, as the transfer destination.

In FIG. 6B (e), services 61-2 and 62-3 are changing from the state of FIG. 6A (c) during preparation for relocation. In this case, the current usage rate of the relocation candidate (66.6.%) Is higher than the current usage rate of the relocation source candidate (66.6%), but the usage rate of the relocation candidate after relocation (0). %) Is less than the post-relocation usage rate (33.3 ..%) of the transfer source candidate, so the transfer target selection unit 405 does not determine the transfer destination candidate instance 50-2 as the transfer destination.

In FIG. 6B (f), services 61-2 and 62-1 are changing from the state of FIG. 6A (c) during preparation for relocation. In this case, the post-relocation usage rate of the relocation candidate (33.3%) is higher than the post-relocation usage rate of the relocation source candidate (33.3%), but at the moment it is in instance 50-2. Since there are no free resources, the service transfer determination unit 404 does not detect the instance 50-2 as a transfer destination candidate.

As in the case of FIG. 6B (f), the service transfer determination unit 404 provides an instance 50-2 capable of executing the service 60 of the transfer target candidate with the free resources indicated by the usage rate after the transfer. It may be detected as a relocation destination candidate. Then, even in the case of FIG. 6B (f), the service transfer determination unit 404 detects the instance 50-2 as the transfer destination candidate, and the transfer target selection unit 405 determines the instance 50-2 as the transfer destination. ..

Multiple transfer source candidates and transfer destination candidates may be detected. The method of determining the relocation destination in such cases will be described. First, in the transfer target selection unit 405, when there are a plurality of instances 50 as transfer source candidates, the resource usage rate after the transfer target service 60 is transferred is smaller than that of other transfer source candidates (that is, the resource usage rate). The instance 50 (which is the smallest) is determined as the transfer source, and the transfer destination of the service 60 to be transferred with the instance 50 as the transfer source is determined.

FIG. 7 shows an example of a plurality of transfer source candidates. In FIG. 7A, service 61-2 is being executed and being processed in instance 50-1, and service 61-3 is being executed and being processed. Further, in the instance 50-2, the services 62-1 and 62-2 are in the state of being executed and being processed, and the service 62-3 is in the state of being executed and the processing is completed. In this case, both instances 50-1 and 50-2 are detected as transfer source targets.

The transfer target selection unit 405 determines the value obtained by subtracting the resource usage rate by the transfer target service 60 from the post-relocation usage rate calculated by the resource usage rate calculation unit 403, and the resource after the transfer target service 60 is transferred. Used as usage rate. In the example of FIG. 7A, the resource usage rate of the instance 50-1 is 33.3 ...%, and the resource usage rate of the instance 50-2 is 66.6 ...%. Therefore, the transfer target selection unit 405 Determines the instance 50-1 with the smaller resource utilization as the transfer source.

In FIG. 7 (b), the states of services 62-1 and 62-2 are changed from FIG. 7 (a) during execution and preparation for relocation. In this case, the resource usage rate of instance 50-1 remains 33.3 ..%, but the resource usage rate of instance 50-2 is 0%, so that the transfer target selection unit 405 has a higher resource usage rate. Determine the smaller instance 50-2 as the transfer source. By determining the transfer source in this way, the free space of resources is biased to the specific instance 50 as compared with the case where another instance 50 is determined as the transfer source. As a result, the instance 50 in which the service to be executed becomes 0 is likely to occur, and the instance 50 is likely to be stopped.

When there are a plurality of instances 50 that have free resources and are candidates for the transfer destination, the transfer target selection unit 405 determines that the resource usage rate after the service 60 to be transferred is transferred to the free resources is another candidate for the transfer destination. Instance 50, which is larger than (that is, has the highest resource utilization), is determined as the transfer destination.
FIG. 8 shows an example of a plurality of transfer destination candidates. In FIG. 8A, the service 61-3 is in the running state in the instance 50-1, and the services 62-2 and 62-3 are in the running state in the instance 50-2. In this case, both instances 50-1 and 50-2 are detected as transfer destination targets.

The transfer target selection unit 405 adds the value obtained by adding the resource usage rate by the transfer target service 60 to the post-relocation usage rate calculated by the resource usage rate calculation unit 403, and the resource after the transfer target service 60 is transferred. Used as usage rate. In the example of FIG. 8A, the resource usage rate of instance 50-1 is 33.3 ...% (utilization rate after relocation) + 33.3 ...% (resource usage rate by service 60) = 66.6.・ ・%, And the resource usage rate of instance 50-2 is 66.6 ・ ・% (utilization rate after relocation) + 33.3 ・ ・% (resource usage rate by service 60) = 100%, so relocation The target selection unit 405 determines the instance 50-2 having the larger resource usage rate as the transfer destination.

In FIG. 8 (b), the states of services 62-2 and 62-3 are changed from FIG. 8 (a) during execution and preparation for relocation. In this case, the resource usage rate of instance 50-1 is 66.6%, which is the same as the example of FIG. 8A, but the resource usage rate of instance 50-2 is 0% (use after transfer). Rate) + 33.3 ...% (Resource usage rate by service 60) = 33.3 ..%, so the transfer target selection unit 405 uses the instance 50-1 with the larger resource usage rate as the transfer destination. decide.

By determining the relocation destination in this way, the number of services 60 that gather at the relocation destination instance 50 after the relocation increases, and the services 60 are concentrated on a specific instance 50, as compared with the case where another instance 50 is determined as the relocation destination. Will be done. Then, the service 60 does not gather in the other instances 50, and the service 60 can be easily moved out from the instance 50 in which the number of the service 60 is reduced. As a result, the instance 50 can be easily stopped. ..

When the transfer target selection unit 405 determines the transfer destination of the service 60, the transfer target selection unit 405 requests the placement of the service 60 to be transferred to the container control request reception unit 51 of the instance of the transfer destination. By this request, the execution environment of the same service 60 as the service 60 to be transferred is created in the instance 50 of the transfer destination as described above (for example, if the transfer target is an OCR service, the execution environment of the OCR service is also created at the transfer destination). Further, the transfer target selection unit 405 supplies the identification information of the determined transfer destination instance 50, the transfer source instance 50, and the transfer target service 60 to the service stop determination unit 406.

The service stop determination unit 406 inquires the container control request reception unit 51 of the transfer destination instance 50 indicated by the received identification information about the status of the service 60 to be transferred. When the status of the service 60 to be transferred changes from being created to being executed, the service stop determination unit 406 sends the service 60 to be transferred to the container control request receiving unit 51 of the instance 50 of the transfer source indicated by the received identification information. Request to stop. When the container control unit 53 of the transfer source stops the service 60 based on this request, the transfer is completed.

When the transfer is completed, the service suspension determination unit 406 provides information regarding the transfer of the service 60 (service transfer information), for example, the supplied instance 50 of the transfer destination, the instance 50 of the transfer source, and the identification information of the service 60 to be transferred. , The time information indicating the time when the transfer is completed is supplied to the service transfer information storage unit 407. The service transfer information storage unit 407 stores the service transfer information supplied from the service stop determination unit 406. The service transfer information storage unit 407 supplies the stored service transfer information to the service management information storage unit 36.

The service management information storage unit 36 accumulates the supplied service transfer information and stores it as a transfer history of the service 60 in the plurality of instances 50. Further, the service management information storage unit 36 acquires and stores execution status information indicating the execution status (creating, executing, relocation preparation) of the service 60 in the plurality of instances 50 from the service status storage unit 402.

Further, the service management information storage unit 36 acquires and stores usage rate information indicating the resource usage rate (current usage rate and usage rate after transfer) in the plurality of instances 50 from the resource usage rate calculation unit 403. .. The service management information storage unit 36 is an example of the "storage unit" of the present invention. The transfer history, execution status information, and usage rate information to the service management information storage unit 36 may be supplied by the pull type requested by the service management information storage unit 36, or the supply source actively supplies the information. It may be done by the push type that comes.

The service management information display unit 201 of the resource management device 20 displays the service management information stored in the service management information storage unit 36. For example, when the administrator of the resource providing system 1 performs an operation to display the transfer history of the service 60, the service management information display unit 201 acquires and displays the transfer history of the service 60 from the service management information storage unit 36. In this case, the service management information storage unit 36 is a function of outputting the transfer history of the service 60 in the plurality of stored instances 50 to the service management information display unit 201, which is a display means, and is the "history output unit" of the present invention. Is an example. By outputting the service transfer history in a plurality of instances in this way, the history of past transfer in each instance can be grasped.

Further, when the administrator performs an operation to display the execution status information (information indicating the execution status of the service 60) in all or a part of the instances 50, the service management information display unit 201 is designated by the service management information storage unit 36. Acquires and displays the execution status information of the instance 50. In this case, the service management information storage unit 36 is a function of outputting the execution status information of the plurality of stored instances 50 to the service management information display unit 201 which is a display means, and is a function of the "state output unit" of the present invention. This is an example.

By outputting the execution status information, the transfer status can be grasped from the execution status of the service. For example, it can be understood that an instance whose service status does not change from the running state has few services that come and go by relocation, and an instance that has many states in preparation for relocation has many services that come and go by relocation.

Further, when the administrator performs an operation to display the resource usage rate (current usage rate and usage rate after transfer) in all or some of the instances 50, the service management information display unit 201 changes to the service management information storage unit. The usage rate information of the instance 50 specified from 36 is acquired, and the resource usage rate indicated by the usage rate information is displayed. As a result, the transfer status can be grasped from the resource usage rate. For example, it can be seen that an instance whose resource usage rate is maintained at a high value has few services that go in and out due to relocation, and an instance whose resource usage rate changes at a low value has many services that go in and out due to relocation.

The first autoscale control unit 32 controls the scale-in and scale-out of the instance according to the defined rule. Scale-in means reducing the number of virtual machines, and scale-out means increasing the number of virtual machines. In this embodiment, for example, when the service stop determination unit 406 of the instance 50-1 stops all the services 60 executed by the container execution unit 52-1, the first autoscale control unit 32 is notified to that effect.

The first autoscale control unit 32 determines that the instance 50 in which the number of services 60 executed by the container execution unit 52-1 has become 0 (all the services 60 that have been executed have been stopped) is the target of scale-in. And control to stop. When the instance 50 is stopped in this way, it is no longer subject to billing (billing according to the number of instances in use) in this embodiment.

In addition to this, the first autoscale control unit 32 may determine scale-in and scale-out based on, for example, the status of the queue stored in the request / response queue 34. The second autoscale control unit 33 controls the scale-in and scale-out of the service 60 according to a predetermined rule. The second autoscale control unit 33 requests the container control request reception unit 51 to stop the service 60 when performing scale-in, and requests the placement of service 60 when performing scale-out.

Each device included in the resource providing system 1 performs a transfer process for transferring services between instances based on the above configuration.
FIG. 9 shows an example of the operation procedure of each device in the transfer process. In FIG. 9, only instances 50-1 and 50-2 are shown, but it is assumed that there are other instances 50. First, when the instance 50 changes the service state or stops the service (steps S11 and S12), the transfer control unit 40 is notified to that effect (steps S13 and S14), and the first autoscale control unit 32 is also notified. Notify (steps S15, S16).

The transfer control unit 40 (resource usage rate calculation unit 403) calculates the resource usage rate (current usage rate and usage rate after transfer) of each instance 50 based on the notified information (step S21). Next, the transfer control unit 40 (service transfer determination unit 404) detects the transfer destination candidate instance 50 based on the calculated resource usage rate (step S22). Subsequently, the transfer control unit 40 (service transfer determination unit 404) detects the service 60 of the transfer target candidate and the instance 50 of the transfer source candidate that executes the service 60 based on the calculated resource usage rate (step S23). ).

Then, the transfer control unit 40 (transfer target selection unit 405) transfers the service 60 to be actually transferred from the detected transfer destination candidate, transfer target, and transfer source candidate, the transfer source instance 50, and the transfer destination. Instance 50 is determined (step S24). In the example of FIG. 9, it is assumed that the instance 50-1 is determined as the transfer destination and the instance 50-2 is determined as the transfer source. The transfer control unit 40 (transfer target selection unit 405) requests the instance 50-1 determined as the transfer source to allocate the service 60 to be transferred (step S31).

Instance 50-1 creates an execution environment for the requested service 60 (step S32). The transfer control unit 40 (service stop determination unit 406) inquires of the instance 50-1 about the status of the service 60 to be transferred (step S33). Instance 50-1 notifies the transfer control unit 40 of the status of the inquired service (step S34). The transfer control unit 40 (service stop determination unit 406) determines whether or not the execution environment is completed (step S35), and if it determines that the execution environment is not completed (NO), returns to step S33 (inquiry) and operates. I do.

When the transfer control unit 40 (service stop determination unit 406) determines that the execution environment is completed (YES) in step S35, the transfer control unit 40 (service stop determination unit 406) tells the transfer source instance 50-2 to stop the service 60 to be transferred. Request (step S36). Instance 50-2 stops the requested service (step S37), and notifies the first autoscale control unit 32 to that effect (step S41) in the same manner as in step S16.

The first autoscale control unit 32 determines whether or not the service executed in the instance 50-2 has become 0 based on the notification contents in steps S16 and S41 (step S42), and has not (NO). If it is determined, this operation procedure is terminated. When the first autoscale control unit 32 determines that the service has become 0 (YES) in step S42, the first autoscale control unit 32 instructs the instance 50-2 to scale in, that is, to stop (step S43). Upon receiving this instruction, the instance 50-2 stops its own instance (step S44) and ends this operation procedure.

In this embodiment, as described above, the instance that is the execution subject of the service changes depending on the status of the resource usage rate (in the example of FIG. 9, the execution subject of the service 60 to be transferred is from instance 50-2 to instance 50-1. Has changed to). As a result, for example, if an instance with only one service continues to execute the service, the number of instances does not decrease, but by moving the service to another instance, the execution subject (instance) of the service is fixed. Compared to the case where the service is executed, the service execution subject is biased to a specific instance, and as a result, the number of instances that execute the service is reduced, and the above-mentioned charge for each number of instances is reduced.

[2] Modifications The above-mentioned examples are merely examples of the implementation of the present invention, and may be modified as follows. Further, the examples and the respective modifications may be carried out in combination as necessary.

[2-1] Threshold value of resource usage rate The transfer target selection unit 405 may use a value different from that of the embodiment as a threshold value determined from the resource usage rate of the instance 50 of the transfer source candidate. The transfer target selection unit 405 changes the threshold value based on, for example, the service transfer history in a plurality of instances stored in the service transfer information storage unit 407. The service transfer information storage unit 407 in this case is an example of the "storage unit" of the present invention.

For example, the transfer target selection unit 405 sets a threshold value to be used when determining whether or not to determine the instance 50 as the transfer destination as the frequency of the transfer indicated by the transfer history of the stored transfer destination candidate instance 50 decreases. Make it smaller. Decreasing this threshold means that if the resource usage rate of the instance of the transfer destination candidate is equal to or higher than the threshold value, the instance is determined as the transfer destination, so that the service can be easily transferred to the transfer destination candidate. means. The transfer target selection unit 405 determines the threshold value using, for example, a threshold value table in which the frequency of transfer and the threshold value are associated with each other.

FIG. 10 shows an example of a threshold table. In the example of FIG. 10, “R1 × 0.5”, “R1 × 1.0” and “R1 × 2” are included in the range of transfer frequency of “less than F1”, “F1 or more and less than F2” and “F2 or more”. A threshold value of ".0" is associated with it. R1 represents the resource usage rate of the instance of the transfer destination candidate. The result of determining the transfer destination when this threshold table is used will be described with reference to FIG.

FIG. 11 shows an example of the determination result of the relocation destination. In FIG. 11, the case where the relocation destination candidate is determined as the relocation destination is indicated by “◯”, and the case where the relocation destination candidate is not determined is indicated by “x”. In FIG. 11A, services 61-1, 61-2, 61-3 are running in instance 50-1, and services 62-2, 62-3 are running in instance 50-2. Our service 61-3 is subject to relocation. That is, the usage rate R1 after the transfer of the instance 50-1 is 100%, and the usage rate R2 after the transfer of the instance 50-2 is 66.6 ...%.

In this case, when the frequency of transfer of the instance 50-2 is less than F1, the transfer target selection unit 405 is 100% (R1) × 0.5 = 50% <66.6 ...% (R2), so the instance 50 -2 is decided as the relocation destination. In addition, when the frequency of transfer of the instance 50-2 is F1 or more and less than F2, the transfer target selection unit 405 becomes 100% (R1) × 1.0 = 100%> 66.6% (R2), and the instance. If the frequency of transfer of 50-2 is F2 or higher, it will be 100% (R1) x 2.0 = 200%> 66.6 ...% (R2), so in both cases, instance 50-2 will be transferred to the destination. Do not decide as.

In FIG. 11B, the service 61-1 of the instance 50-1 is stopped and becomes empty from the state of FIG. 11A, and the usage rate R1 after the transfer of the instance 50-1 is 66.6.%. It has changed to. In this case, the transfer target selection unit 405 transfers 66.6.% (R1) x 0.5 = 33.3.% <66.6. When the frequency of transfer of the instance 50-2 is less than F1.・% (R2), and if the frequency of instance 50-2 transfer is F1 or more and less than F2, 66.6 ・ ・% (R1) × 1.0 = 66.6 ・ ・% = 66.6 ・ ・% Since it becomes (R2), instance 50-2 is determined as the transfer destination in both cases. In addition, the transfer target selection unit 405 transfers 66.6.% (R1) x 2.0 = 133.3.%> 66.6 when the frequency of transfer of the instance 50-2 is F2 or more. % (R2), so instance 50-2 is not determined as the transfer destination.

In FIG. 11 (c), the service 61-2 of the instance 50-1 is stopped and becomes empty from the state of FIG. 11 (b), and the usage rate R1 after the transfer of the instance 50-1 is 33.3.%. It has changed to. In this case, the transfer target selection unit 405 transfers 33.3 ..% (R1) x 0.5 = 16.6 ..% <66.6. When the frequency of transfer of the instance 50-2 is less than F1.・% (R2), and if the frequency of instance 50-2 transfer is F1 or more and less than F2, 33.3 ・ ・% (R1) × 1.0 = 33.3 ・ ・% <66.6 ・ ・% (R2), and if the frequency of instance 50-2 transfer is F2 or higher, 33.3% (R1) x 2.0 = 66.6% = 66.6% (R2). Therefore, in either case, instance 50-2 is determined as the transfer destination.

As shown in the example of FIG. 11, in this modification, the less frequently the transfer history of the instance 50-2, which is a candidate for the transfer destination, indicates, the easier it is for the instance 50-2 to be determined as the transfer destination. As in the example of FIG. 11C, if R2 is sufficiently larger than R1, it may always be determined as the transfer destination regardless of the frequency of transfer, and conversely, if R2 is too small compared to R1. , No matter how infrequent the relocation is, it may not be decided as the relocation destination.

Further, the transfer target selection unit 405 may use the transfer history of other services. For example, the transfer target selection unit 405 sets a threshold value to be used when determining whether or not to determine the instance 50 as the transfer destination as the number of transfers indicated by the transfer history of the stored transfer destination candidate instance 50 decreases. Make it smaller. In this case, the transfer target selection unit 405 determines the threshold value using the threshold value table in which the number of transfers and the threshold value are associated with each other.

FIG. 12 shows another example of the threshold table. In the example of FIG. 12, “R1 × 0.5”, “R1 × 1.0” and “R1 × 2” are included in the range of the number of transfers of “less than C1”, “C1 or more and less than C2” and “C2 or more”. A threshold value of ".0" is associated with it. By using this threshold table, as in the example of the frequency of transfer shown in FIG. 10, the smaller the number of transfers, the smaller the threshold is used.

Among the services, there are services in which scale-out and scale-in by the second autoscale control unit 33 are frequently performed (services in which usage is temporarily concentrated) and services in which they are rarely performed (changes in the number of usage). There are few services). Instances running the former service tend to have lower usage rates after relocation than instances running the latter service, and are more likely to relocate, and the frequency and frequency of relocations are more likely to increase. Therefore, when a service is transferred to an instance with a low frequency and frequency of transfer, the transferred service is more likely to settle in the instance as compared with the case where the instance with a high frequency and frequency of transfer is set as the transfer destination.

In addition, the transfer target selection unit 405 determines whether or not to determine the instance 50 as the transfer destination as the latest transfer time indicated by the transfer history of the stored transfer destination candidate instance 50 is older. The threshold used may be reduced. The latest relocation time here means the time when the service was last relocated in the instance retroactively from the present time. In this case, the transfer target selection unit 405 determines the threshold value using the threshold value table in which the latest transfer time and the threshold value are associated with each other.

FIG. 13 shows another example of the threshold table. In the example of FIG. 13, “R1 × 0.5”, “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” and “R1 × 1.0” A threshold value of "R1 x 2.0" is associated with it. By using this threshold table, as in the case of the transfer frequency example shown in FIG. 10, the older the latest transfer time, the smaller the threshold value will be used.

Instances running services that are frequently scaled out and scaled in are more up-to-date with lower post-relocation utilization and more prone to relocation than instances running services that are less frequent. The relocation time is likely to be new. Therefore, when a service is transferred to an instance with the latest transfer time, the transferred service is more likely to settle in the instance than when the instance with the latest transfer time is the transfer destination.

As described above, in the present modification, by changing the threshold value based on the service transfer history, the transferred service is more likely to settle in the instance than in the case where the threshold value is fixed. As a result, it becomes easier to maintain the bias of the service execution subject described above, and conversely, the number of instances that do not become the service execution subject tends to increase, and those instances are stopped by scale-in, so that the number of instances in use The billing will be reduced accordingly.

[2-2] Review of relocation destination Even if the relocation destination is decided and the service is in the state of being prepared for relocation, it will not be stopped until the execution environment is completed in the instance of the relocation destination, so it will be relocated in the meantime. The resource utilization of the previous instance may change. In that case, the relocation destination once decided by the relocation target selection unit may be reviewed.

In this modification, if the transfer target selection unit 405 determines that the resource usage rate of the transfer destination instance is less than the threshold value when the processing executed by the transfer target service is completed, the other instance is used as the service. Will be decided as the relocation destination. The operation procedure of the transfer control unit 40 in this case will be described with reference to FIG.

FIG. 14 shows an example of the operation procedure of each device in the transfer process of this modified example. In this operation procedure, steps S11 (change of service state / stop of service) shown in FIG. 9 to step S35 (determination of completion of execution environment) are performed. Next, the transfer control unit 40 (transfer target selection unit 405) inquires about the service status of the transfer destination instance (instance 50-1 in this example) (step S51).

Upon receiving this inquiry, the instance 50-1 notifies the transfer control unit 40 of the service status of its own instance (step S52). The transfer control unit 40 (resource usage rate calculation unit 403) calculates the resource usage rate by reflecting the notified service status (step S53). Next, the transfer control unit 40 (transfer target selection unit 405) determines whether or not the resource usage rate of the instance 50-1, that is, the transfer destination is less than the threshold value (step S54).

If it is determined in step S54 that the resource usage rate of the transfer destination is not less than the threshold value (NO), the transfer control unit 40 (service stop determination unit 406) performs the operation of step S36 (service stop request). In FIG. 14, the operations after step S41 in FIG. 7 are omitted. If it is determined in step S54 that the resource usage rate of the transfer destination is less than the threshold value (YES), the transfer control unit 40 (resource usage rate calculation unit 403) returns to step S21 (calculation of resource usage rate) and operates. I do. By performing the operation from step S21 again, the relocation destination is reviewed.

Since there is a time lag between the decision of the relocation destination and the actual completion of the relocation, for example, services are scaled in and executed less services before the relocation is completed, making it unsuitable as a relocation destination. There is. In this modification, by reviewing the relocation destination as described above, the transfer of services to the instance that is no longer suitable as the relocation destination is suppressed as compared with the case where the relocation destination is not reviewed.

[2-3] Allotted Resources In the embodiment, in order to make the explanation easy to understand, the resources assigned to each instance (first resource) are equal, and all the resources assigned to each service are also 3 of the first resource. It is assumed to be one-third, but it is not limited to this. For example, the first resource may be different for each instance, or the resource allocated to each service may be different. Also, the number of services executed in each instance may be different.

[2-4] Transfer of a plurality of services A transfer request may be sent from a plurality of instances to an instance whose resources are free. If the resources of each service are equal as in the embodiment, the resource usage rate of the own instance does not change regardless of which transfer request is accepted, but if the resources of each service are different as in the above modification, which transfer request is received. The resource usage rate of your instance changes depending on whether you accept.

Therefore, in this modification, when the transfer target selection unit 405 of the transfer control unit 40 transfers services from a plurality of transfer source candidate instances to the transfer destination candidate instance, the resource usage rate of the transfer destination candidate is increased. Determine larger services for relocation.
FIG. 15 shows an example of determining the service to be transferred. In FIG. 15A, instance 50-1 with 50% free space is detected as a transfer destination candidate, and service A using 40% of resources and service B using 30% of resources are candidates for transfer. Detected as a service.

In this case, the transfer target selection unit 405 transfers the service A whose resource usage rate is 90% rather than the service B whose resource usage rate after the transfer of the transfer destination candidate (instance 50-1) is 80%. Determined as a service. In FIG. 15B, instance 50-1 with 50% free space is detected as a transfer destination candidate, and in addition to services A and B, service C using 20% of resources is detected as a transfer target candidate service. ing.

In this case, the transfer target selection unit 405 determines the services B and C whose resource usage rate after the transfer of the transfer destination candidate is 100% as the transfer target services. By accepting the transfer request in this way, the resource usage rate of the transfer destination candidate is increased as compared with the case where the transfer target service is determined regardless of the resource usage rate after the transfer of the transfer destination candidate.

The transfer target selection unit 405 may determine the transfer target service by another method. When there are a plurality of services that can be transferred (services that are candidates for transfer), the transfer target selection unit 405 determines the transfer destination of the service whose processing being executed finishes earlier than other services, that is, the service is transferred. Determined as a service. In this case, the container execution unit 52 of the instance 50 notifies the service state management unit 401 of the estimated end time of the process being executed by the service 60 as the state of the service 60, so that the service state storage unit 402 performs the status. Memorize the expected end time.

The transfer target selection unit 405 reads out each estimated end time of the services of the plurality of transfer target candidates, and determines the service having the earliest time as the transfer target service. As a result, the service transfer time is earlier than in the case of transferring the service regardless of the processing end time, so that the bias of the service executing subject to the specific instance 50 progresses faster and the instance may be stopped. Will occur sooner.

[2-5] Service Redundancy Some of the services that are candidates for transfer are made redundant by using other instances (hot standby type redundancy, cold standby type redundancy, replication by replication, etc.). May be. In that case, the transfer target selection unit 405 does not have to determine the transfer destination for such redundant services. In this modification, the container execution unit 52 of the instance 50 notifies the service state management unit 401 of the presence or absence of redundancy of the service 60 as the state of the service 60, so that the service state storage unit 402 performs the redundancy. Remember the presence or absence.

The transfer target selection unit 405 reads out whether or not the services of the plurality of transfer target candidates are redundant, and for the services that have been made redundant, the transfer destination is not determined by excluding them from the transfer target candidates, and the services are made redundant. Decide where to relocate for services that are not available. As a result, even if the service is redundant, the recovery error of the redundant service is reduced as compared with the case where the relocation destination is determined and relocated.

[2-6] Resource usage rate to be compared In the example, the transfer target selection unit 405 used the usage rate after the transfer as the resource usage rate to be compared between the transfer source candidate and the transfer destination candidate, but the current usage rate. May be used. For example, in the example of FIG. 6A (c), the usage rate after relocation was smaller for instance 50-2 than for instance 50-1, so the relocation was not performed, but at the current usage rate, both instances. Since both will be the same value, the relocation will be carried out. Even in this case, the number of instances that execute the service is reduced and the charge for each number of instances is reduced as compared with the case where the execution subject (instance) of the service is fixed.

[2-7] Selection of Effective Determination Method The transfer target selection unit 405 uses the plurality of determination methods described in the examples and modifications as a method of determining a transfer destination to actually transfer from a plurality of transfer destination candidates. Either may be used. The plurality of determination methods are, for example, a first method of determining an instance whose resource usage rate after the transfer target service described in FIG. 7 is smaller than that of other transfer source candidates as a transfer source, and its resources. The second method of using the current usage rate instead of using the usage rate after relocation to obtain the usage rate, and the second method of determining the relocation destination by the first method, and then reviewing the relocation destination described in FIG. This is the third method or the fourth method of reviewing the relocation destination after determining the relocation destination by the first method.

In that case, the transfer target selection unit 405 determines the frequency or number of times that the number of services executed by the transfer source instance becomes 0 by transferring the service to the determined transfer destination among the plurality of determination methods. The method with more is may be used with priority. In this modification, the container execution unit 52 of the instance 50 notifies the service status management unit 401 of the number of services being executed when the service to be transferred is stopped as the service status of the own instance. , The service status storage unit 402 stores the number of remaining services at the time of the service transfer.

For example, the transfer target selection unit 405 determines the transfer destination by evenly using the above-mentioned plurality of determination methods for a certain period of time, and notifies the service status management unit 401 of the determination method used at that time to notify the service status. The storage unit 402 stores the notified determination method in association with the number of remaining services described above. After the period has passed, the transfer target selection unit 405 refers to the service state storage unit 402 when determining the transfer destination, and calculates the frequency with which the number of remaining services becomes 0 for each determination method.

The transfer target selection unit 405 determines the transfer destination by preferentially using the calculated determination method with higher frequency. Specifically, for example, if the frequency ratio in the first, second, third, and fourth methods is 3: 1: 4: 2, the first method is used with a probability of three-tenths. It is in good condition. The same may be applied when the number of times is used instead of the frequency. As a result, the number of services remaining in the transfer source instance is likely to be 0, as compared with the case where the transfer destination determination method is fixed, so that the number of virtual machines executing the services is likely to decrease.

As a plurality of determination methods, a plurality of determination methods for determining the service to be transferred may be used. In that case, the transfer target selection unit 405 determines the frequency or number of times that the number of services executed by the transfer source instance becomes 0 by transferring the determined transfer target service among the plurality of determination methods. The method with more is may be used with priority. Further, as a plurality of determination methods, a plurality of determination methods for determining the transfer source instance may be used. In that case, the transfer target selection unit 405 transfers the services of the determined transfer source instance among the plurality of determination methods, so that the number of services executed by the transfer source instance becomes 0. The method with more frequency or rotation may be used with priority.

Among those plurality of determination methods, the transfer target selection unit 405 has a high frequency or number of times that the number of services executed by the transfer source instance becomes 0 by transferring the services of the determined transfer source instance. The method may be used with priority. In either case, compared to the case where the method of determining the transfer destination is fixed, the number of services remaining in the transfer source instance is likely to be 0, so that the number of virtual machines executing the services is likely to decrease.

[2-8] Timing of Transfer Judgment In the embodiment, the service transfer determination unit 404 makes a determination regarding the transfer of the service 60 at a predetermined time interval, but the present invention is not limited to this, and for example, the service 60 executes the process. You may make a decision regarding the transfer of the service 60 when the service 60 is terminated, or when there is an instance 50 whose CPU usage rate or memory usage rate is below the threshold value, the service 60 running on the instance 50 is transferred. You may make a judgment about.

[2-9] Virtualization Method In the embodiment, a virtual machine is realized by a container-type virtualization technology, but the virtual machine is not limited to this, and is virtualized by a hypervisor-type virtualization technology such as VMware (registered trademark). The machine may be realized. In short, any technology may be used as long as a virtual machine that is logically divided and behaves like one device is realized instead of a physically divided device (physical machine). ..

[2-10] Method for Realizing Each Function In each function shown in FIGS. 4 and 5, two or more functions may be integrated, or one function may be divided into two or more functions. For example, the resource usage rate calculation unit 403 may be divided into a calculation unit that calculates the current usage rate and a calculation unit that calculates the usage rate after the transfer. Further, the service transfer determination unit 404, the transfer target selection unit 405, and the service stop determination unit 406 may be integrated to form a determination unit that collectively determines the service transfer.

Further, in the embodiment, the first autoscale control unit 32 and the like, which are functions other than the instance, are also realized by the virtual machine, but these may be realized by the physical machine. In short, as long as the functions equivalent to the functions shown in FIGS. 4 and 5 are realized in the entire resource providing system 1, the functions may be organized in any way.

[2-11] Category of Invention The present invention can be regarded as an information processing system called a resource providing system including the information processing devices such as the server device 10 and the resource management device 20. Further, the present invention can be regarded as an information processing method for realizing the processing performed by each device, and also as a program for operating a computer that controls each device. This program may be provided in the form of a recording medium such as an optical disk that stores it, or may be downloaded to a computer via a communication line such as the Internet, and installed and made available. May be provided.

1 ... Resource providing system, 10 ... Server device, 20 ... Resource management device, 30 ... Virtual machine realization unit, 31 ... Instance group, 32 ... 1st autoscale control unit, 33 ... 2nd autoscale control unit, 34 ... Request -Response queue, 35 ... Service definition storage unit, 36 ... Service management information storage unit, 40 ... Transfer control unit, 50 ... Instance, 60 ... Service, 201 ... Service management information display unit, 401 ... Service status management unit, 402 ... Service status storage unit, 403 ... resource usage rate calculation unit, 404 ... service transfer determination unit, 405 ... transfer target selection unit, 406 ... service stop determination unit, 407 ... service transfer information storage unit.

Claims (15)

Computer,
A detector that detects a first virtual machine that has free resources capable of executing one or more services from a plurality of virtual machines that each execute a plurality of services using the first resource assigned to itself.
When the detected resource usage rate of the first virtual machine is equal to or greater than a threshold value determined from the resource usage rate of the second virtual machine running a service that can be executed by the free resource among the plurality of virtual machines, the above-mentioned A program that allows the first virtual machine to function as a decision-making unit that determines the transfer destination of the service. When there are a plurality of the second virtual machines that are candidates for the transfer source of the service, the determination unit determines the second virtual machine whose resource usage rate becomes smaller than that of the other candidates for the transfer source after the transfer of the service. The program according to claim 1, which determines the transfer destination of the service to be the transfer source. The program according to claim 1 or 2, wherein the determination unit determines a transfer destination of a service whose processing being executed finishes earlier than other services when there are a plurality of transferable services. Claims 1 to 3 do not determine the transfer destination for the service when the service that can be executed by the free resource is made redundant by using a virtual machine different from the second virtual machine. The program described in any one of the above. The computer
A first calculation unit that calculates for each of the plurality of virtual machines the first ratio of the second resource assigned to one or more running services to the first resource as the resource usage rate.
The second ratio of the third resource allocated to the running service excluding the service scheduled to be transferred to another virtual machine to the first resource is calculated as the resource usage rate for each of the plurality of virtual machines. 2 Make it function as a calculation unit
The determination unit determines the first virtual machine for which the second ratio equal to or higher than the threshold value determined from the second ratio of the second virtual machine is calculated as the transfer destination of the service that can be executed by the free resource. The program according to any one of 1 to 4. When there are a plurality of the second virtual machines that are candidates for the transfer source of the service, the determination unit is the second virtual machine in which the second ratio becomes smaller than the other candidates for the transfer source after the transfer of the service. The program according to claim 5, which determines the transfer destination of the service being executed. When there are a plurality of the first virtual machines that are candidates for the transfer destination, the determination unit determines that the second ratio after the service is transferred to the free resource is larger than that of the other candidates for the transfer destination. The program according to claim 5 or 6, wherein the transfer destination is determined. The computer
It functions as a storage unit for storing the transfer history of services in the plurality of virtual machines.
The program according to any one of claims 1 to 7, wherein the determination unit changes the threshold value based on the stored transfer history. The program according to claim 8, wherein the determination unit reduces the threshold value as the frequency of transfer or the number of transfers indicated by the stored transfer history of the first virtual machine decreases. The program according to claim 8, wherein the determination unit reduces the threshold value as the latest transfer time indicated by the stored transfer history of the first virtual machine is older. The determination unit determines the transfer destination using any of a plurality of methods, and among the plurality of methods, the service executed by the second virtual machine by transferring the service to the determined transfer destination. The program according to any one of claims 1 to 10, which is preferentially used as the method has a frequency of 0 or a large number of times. When the resource usage rate of the first virtual machine is less than the threshold value when the processing executed by the service that can be executed by the free resource is completed, the determination unit determines another virtual machine. The program according to any one of claims 1 to 11, which is determined as the transfer destination of the service. The computer
A storage unit that stores the transfer history of services in the plurality of virtual machines,
The program according to any one of claims 1 to 12, which functions as a history output unit that outputs the stored transfer history. The computer
The program according to any one of claims 1 to 13, which functions as a state output unit that outputs information indicating the execution state of services in the plurality of virtual machines. A detector that detects a first virtual machine that has free resources capable of executing one or more services from a plurality of virtual machines that each execute a plurality of services using the first resource assigned to itself.
When the detected resource usage rate of the first virtual machine is equal to or greater than the threshold value determined from the resource usage rate of the second virtual machine running a service that can be executed by the free resource among the plurality of virtual machines, the above-mentioned An information processing device including a decision unit that determines the first virtual machine as the transfer destination of the service.

Images