Hub management device and hub management method

Abstract

To provide a hub management device and a hub management method which allow electric power from renewable energy to be used more efficiently.SOLUTION: A workload migration planning program 2470 migrates an execution hub where a calculation load is executed, in accordance with a migration plan made by using a hub power management table 2410. Further, the workload migration planning program 2470 acquires error information indicating an amount of deviation and a deviation occurrence time being a time when the deviation occurs, the amount of deviation being the deviation between a first predictive value being a value shown by the hub power management table 2410 used for making a migration plan of an object amount being at least one of a power demand and a renewable energy supply and a second predictive value predicted after the first predictive value of the object amount. Then, the workload migration planning program 2470 makes a corrected migration plan obtained by correcting the migration plan on the basis of the error information and load related information (2420 to 2460).SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a base management device and a base management method. [Background technology]

[0002] The use of renewable energy is being promoted as a measure against climate change. In the past, the general method of procuring renewable energy was to purchase certificates such as non-fossil certificates. However, in recent years, with the increase in the amount of renewable energy in circulation, an increasing number of electric power companies are purchasing renewable energy directly from power generation companies through PPAs (Power Purchase Agreements) and acquiring environmental value through the transfer of certificates associated with the purchase. This trend is also seen in consumers who consume large amounts of electricity, and electricity procurement through PPAs is becoming widespread, mainly in Europe and the United States, even among data centers (DCs), which are large consumers of electricity. In addition, advanced electric power companies in Europe and the United States are working to increase the utilization rate of renewable energy on an hourly basis, instead of the currently mainstream annual basis, with a view to increasing the utilization rate of renewable energy throughout society in the future.

[0003] On the other hand, DCs, which are large-scale consumers, require large amounts of electricity throughout the day, so the efficient use of renewable energy, which is unstable and has geographical restrictions, is a challenge. In response to this, a technology that selects a DC to execute a computational load (WL: Workload) from multiple DCs located in different locations according to the area and time period where renewable energy is abundant is attracting attention. In this technology, the computational load is reallocated according to a reallocation plan that is formulated based on the predicted value of the supply and demand of electricity from renewable energy sources.

[0004] For example, Patent Document 1 discloses a technology that automatically relocates applications based on the predicted amount of electricity (including electricity from renewable energy sources) supplied to each of multiple hosting sites that run the applications, and the predicted cost of that supply, and the available amount of computing resources at each hosting site. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] US Patent Application Publication No. 2011 / 0282982 Summary of the Invention [Problem to be solved by the invention]

[0006] In the technology described in Patent Document 1, application reallocation is performed at a predetermined interval, but the predicted results of the power supply amount and the like may change over time, so the predicted values ​​of the power supply amount and the like may change during the interval. In this case, the reallocation plan may deviate from the appropriate value, resulting in a decrease in the amount of power used from renewable energy sources and CO2 emissions. 2 There is a risk of increased emissions and increased costs.

[0007] An object of the present invention is to provide a site management device and a site management method that are capable of using electricity generated from renewable energy sources more efficiently. [Means for solving the problem]

[0008] A base management device according to one aspect of the present disclosure is a base management device that manages a plurality of bases capable of executing a computation load that performs computation processing, and includes a processor and a memory, and the memory stores power management information indicating, for each base, a predicted value of a target amount, which is at least one of a supply amount, which is an amount of power from renewable energy supplied to the base, and a demand amount, which is an amount of power consumed at the base, a power consumption amount, which is an amount of power consumed by the computation load, an execution base, which is the base that executes the computation load, and an estimated transition time, which is an estimated value of a transition time required to transition from the execution base to another base. and load-related information indicated for each of the computational loads, and the processor migrates the execution base of each computational load in accordance with a migration plan for migrating the execution base of each computational load, created using the power management information, acquires error information indicating the amount of deviation, which is the magnitude of deviation between a first predicted value, which is the predicted value indicated in the power management information used to create the migration plan, and a second predicted value predicted after the first predicted value of the target quantity, and the deviation occurrence time, which is the time at which the deviation occurs, and creates a corrected migration plan by correcting the migration plan based on the error information and the load-related information. Effect of the Invention

[0009] According to the present invention, it becomes possible to use electricity generated from renewable energy sources more efficiently. [Brief description of the drawings]

[0010] [Figure 1] 1 is a diagram illustrating an overall configuration of a multi-site data center management system according to an embodiment of the present disclosure. [Diagram 2] FIG. 2 is a diagram illustrating a configuration of a multi-site management server. [Diagram 3] FIG. 11 illustrates an example of a base power management table. [Figure 4] FIG. 13 is a diagram illustrating an example of an arrangement rule table. [Diagram 5] FIG. 13 illustrates an example of a calculation load management table. [Figure 6] FIG. 13 illustrates an example of a storage management table. [Figure 7] FIG. 13 illustrates an example of an application management table. [Figure 8] FIG. 13 is a diagram illustrating an example of a transition time table. [Figure 9] 13 is a flowchart illustrating an example of a migration plan correction process. [Figure 10] FIG. 13 is a diagram showing an example of an operation result screen. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. Note that the embodiments described below do not limit the scope of the invention, and not all of the elements and combinations thereof described in the embodiments are necessarily essential to the solution of the invention.

[0012] In the following description, various information may be described using the expression "aaa table", but the various information may be expressed in a data structure other than a table. In order to show that it does not depend on the data structure, the "aaa table" may be called "aaa information". In the following description, the "program" may be used as the subject, but the program may be used as the subject because the program is executed by a processor (e.g., a central processing unit (CPU) or a graphics processing unit (GPU)) and performs a process determined by the execution of the program while appropriately using a storage resource (e.g., a memory) and an interface device (e.g., a communication device). Similarly, the subject of the process performed by executing the program may be a controller, device, system, computer, node, storage device, server, client, or host having a processor. In addition, a part or all of the program may be processed using a specific hardware circuit. In addition, the various programs may be installed in each computer by a program distribution server or a storage medium. In the following description, two or more programs may be realized as one program, and conversely, one program may be realized as two or more programs.

[0013] In the following description, an ID is used as identification information for an element, but other types of identification information may be used instead of or in addition to the ID.

[0014] 1 is a diagram showing an overall configuration of a multi-site data center management system, which is a computer system according to an embodiment of the present disclosure. The multi-site data center management system shown in FIG. 1 includes a plurality of sites (Site Systems) 1000, a multi-site management server (Multi-Site Management Server) 2000, and a client server (Client Server) 3000.

[0015] Each base 1000 is a location where a DC (data center) is located. The DC is a group of devices for storing data and executing applications for performing computational load (computational processing) using the data. Each base 1000 is installed, for example, in a geographically separated location. Also, one base 1000 may have multiple DCs. Each base 1000 or each DC may build different cloud environments such as on-premise, public cloud, and private cloud. In other words, the present disclosure is applicable to a so-called hybrid cloud having different cloud environments.

[0016] The multi-site management server 2000 is a site management device (management computer) that manages the execution of applications by each DC and the settings of each device based on information about the DCs arranged at each site 1000. The multi-site management server 2000 is communicably connected to the DCs present at each site 1000 via a wide area network 1800.

[0017] The client server 3000 is a client computer used by a user of the DC, and is communicably connected to the DC present at each base 1000 via the client network 6000. The client server 3000 transmits various instructions, such as an instruction to execute an application, to the DC, and receives a response from the DC according to the instructions.

[0018] Each base 1000 has the following components that make up a DC: a DC Facility 1050, an IT Asset 1100, an Application Management Server 1200, an IT Asset Management Server 1300, and an Energy Management Server 1400, which are connected to each other so that they can communicate with each other via a management network 5000.

[0019] The DC equipment 1050 is equipment used to operate the DC, and is, for example, air conditioning equipment and standby power equipment such as a UPS (Uninterruptible Power Supply). The IT device 1100 is a device constituting the DC main body, and is a processing computer for storing data and executing a calculation load using the data. In the example of Fig. 1, the IT device 1100 has a storage 1101 for storing data, a network switch 1102 for managing communication of data and the like, and a server 1103 for executing an application for executing a calculation load using the data.

[0020] The application management server 1200 is a management computer that manages applications executed on the server 1103 of the IT device 1100, and obtains and stores application management information relating to the applications.

[0021] The IT equipment management server 1300 is a management computer that manages the IT equipment 1100, and acquires and stores IT equipment information related to the IT equipment 1100. The IT equipment information includes, for example, configuration information indicating the connection relationships of the IT equipment 1100, and performance information indicating the usage status of each resource (CPU, network card, disk, etc.) that constitutes the IT equipment 1100.

[0022] The power management server 1400 is a management computer that manages the power consumption consumed by the DC, and acquires and stores power information related to the power consumption. The power consumption of the DC includes not only the power consumed by the IT equipment 1100, but also the power consumed by the DC facility 1050.

[0023] In this embodiment, the management computers (application management server 1200, IT device management server 1300, and power management server 1400) are distributed, but any one of the management computers may be integrated with another management computer. Also, in this embodiment, the multi-site management server 2000 is disposed independently of each site 1000, but may be disposed at any one of the sites 1000.

[0024] Fig. 2 is a diagram showing the configuration of the multi-site management server 2000. As shown in Fig. 2, the multi-site management server 2000 includes a management network interface (I / F) 2100, a processor 2200, an input / output (IO) device 2300, a local disk 2400, and a memory 2500.

[0025] The management network I / F 2100 is communicably connected to devices in each base 1000 shown in Fig. 1. The input / output device 2300 is a user interface, such as a monitor, keyboard, mouse, etc., for inputting and outputting information to and from a user.

[0026] Processor 2200 is a processing unit that performs various processes by loading and executing programs into memory 2500. Local disk 2400 is a storage device that stores programs that define the operation of processor 2200 and various information used or generated in the processing by those programs. Memory 2500 is a main storage device that temporarily stores various information and is used as a work area for programs.

[0027] In this embodiment, the local disk 2400 stores, as information, a site energy management table 2410, a placement policy table 2420, a workload management table 2430, a storage management table 2440, an application management table 2450, and a migration time table 2460. The local disk 2400 also stores, as programs, a workload replacement planning program 2470 and a workload replacement program 2480.

[0028] The base power management table 2410 is power management information configured with power status information related to the power status at each time in each base 1000. The multi-base management server 2000 acquires at least a part of the power information stored in the power management server 1400 and manages it as the base power management table 2410. The power status indicates, for example, at least one of a renewable energy supply amount, which is the amount of power from renewable energy supplied to the base 1000, and a power demand amount, which is the amount of power consumed at the base. The power status information includes a predicted value related to the power status, and may also include an actual measurement value related to the power status and electricity price information. The method of creating the base power management table 2410 is not particularly limited. For example, a predicted value related to the power status may be acquired from the power management server 1400, or may be calculated by the multi-base management server 2000 from an actual value or the like.

[0029] The allocation rule table 2420, the computation load management table 2430, the storage management table 2440, the application management table 2450, and the transition time table 2460 constitute load-related information regarding the computation load executed at each base 1000 (each DC).

[0030] The placement rule table 2420 is configured with placement rule information indicating placement rules that are rules regarding the transfer (relocation) of the computational load. The transfer of the computational load is performed by transferring (changing) the execution site, which is the site 1000 (DC) that executes the computational load, to another site. There is no particular limitation on the method of creating the placement rule table 2420. For example, the placement rule table 2420 may be created by the multi-site management server 2000, or may be created by another computer such as the IT device management server 1300.

[0031] The calculation load management table 2430 is configured with calculation load information, which is information regarding the calculation load constituting the application executed by the server 1103 of the IT device 1100. The multi-site management server 2000 acquires at least a part of the application management information stored in the application management server 1200 and manages it as the calculation load management table 2430.

[0032] The storage management table 2440 is configured with storage information related to the storage 1101 that stores data used by the computational loads that constitute the applications. The storage information is configured with information acquired by the multi-site management server 2000 from the IT device management server 1300, but may also be created from data acquired from other management computers.

[0033] The application management table 2450 is configured with application information, which is information related to applications executed by the server 1103 of the IT equipment 1100. The application information is configured with information acquired by the multi-site management server 2000 from the application management server 1200, but may also be created from data acquired from other management computers.

[0034] The migration time table 2460 is configured with migration-related information related to the migration of the calculation load executed by the server 1103 of the IT equipment 1100. The migration-related information is configured with information acquired by the multi-site management server 2000 from the application management information stored in the application management server 1200, but may also be created from data acquired from other management computers.

[0035] The workload migration plan program 2470 is a program for creating a migration plan for migrating the execution base of a computational load to another base. The migration plan indicates, for example, a migration target load, which is the computational load to be migrated, and the migration source and destination execution bases that execute the migration target load.

[0036] The workload migration planning program 2470 creates a migration plan based on each piece of information stored in the local disk 2400. For example, the workload migration planning program 2470 creates a migration plan so as to maximize the use of the renewable energy supply amount, which is the amount of power from renewable energy supplied to the base 1000. The migration plan indicates, for example, the computational load to be migrated, the base to which the data is to be migrated, and the migration type for each predetermined interval (e.g., 30 minutes).

[0037] Furthermore, the workload shift planning program 2470 creates a corrected shift plan by correcting the shift plan based on error information indicating an error between a first predicted value, which is a value indicated in the base power management table 2410 used in creating a shift plan for a target quantity, which is at least one of a renewable energy supply amount and a power demand amount, and a second predicted value, which is a value predicted after the first predicted value of the target quantity. The error information indicates, for example, a deviation amount, which is the magnitude of deviation between the first predicted value and the second predicted value, and a deviation occurrence time at which the deviation occurs. Note that, when the target quantity is both a renewable energy supply amount and a power demand amount, the deviation amount indicated by the error information may be, for example, a sum of the deviation amounts of the renewable energy supply amount and the power demand amount.

[0038] The workload migration program 2480 migrates the execution base of the computational load in accordance with the migration plan and the corrected migration plan created by the workload migration planning program 2470 .

[0039] 3 is a diagram showing an example of the base power management table 2410. The base power management table 2410 shown in FIG.

[0040] Column 2411 stores a site ID, which is identification information for identifying the base 1000. Column 2412 stores a timestamp indicating a time. Column 2413 stores a renewable energy supply forecast, which is a forecast value predicting the amount of power supplied from renewable energy at a given base (base 1000 with a site ID in the same record) at a given time (the time indicated by the timestamp in the same record). Column 2414 stores a power demand forecast, which is a forecast value predicting the amount of power demand at a given base at a given time.

[0041] Each of the records 241A to 241D of the base power management table 2410 indicates time-series power status information of each base 1000. For example, the record 241A indicates that the renewable energy supply forecast value is 3.0 MW and the power demand forecast value is 10.1 MW as of "2018 / 4 / 1 / 10:00" for the base with site ID "01".

[0042] 4 is a diagram showing an example of the arrangement rule table 2420. The arrangement rule table 2420 shown in FIG.

[0043] Column 2421 stores a policy ID for identifying a placement rule. Column 2422 stores the site ID of the base 1000 where an application can be placed as the content of the placement rule (placement rule of the policy ID of the same record).

[0044] Each of records 242A to 242E in the placement rule table 2420 indicates placement rule information for an application. For example, in record 242A, the placement rule for policy ID "01" indicates that the application can be placed in the bases 1000 with site IDs "01" and "02". Also, in the case of records 242C and 242E where there is only one site ID, the placement rule for that record indicates that the application cannot be relocated.

[0045] 5 is a diagram showing an example of the calculation load management table 2430. The calculation load management table 2430 shown in FIG.

[0046] Column 2431 stores a workload ID, which is identification information for identifying a computation load. Column 2432 stores an application ID for identifying an application configured by the computation load (the computation load of the workload ID of the same record). Column 2433 stores an estimated power consumption that estimates the amount of power consumed by the computation load. Column 2434 stores storage configuration information that indicates the configuration of the storage 1101 used by the computation load.

[0047] The storage configuration information indicates whether a volume is assigned to an application configured by a computation load (i.e., whether there is data used by the application) and whether data is copied. Specifically, the storage configuration information indicates "No Volume" when no volume is assigned to the application, and indicates "Allocated" when a volume is assigned to the application. Furthermore, the storage configuration information indicates "Allocated and Copied for Migration" when a volume is assigned to the application and data is copied for application migration, and indicates "Allocated and Copied for DR" when a volume is assigned to the application and data is copied for application disaster recovery. Note that copying of data may be performed for other purposes such as backup. Furthermore, the purpose of copying (application migration and disaster recovery) may not be required.

[0048] Each record 243A to 243D of the computational load management table 2430 indicates computational load information for each computational load. For example, record 243A indicates that the computational load with workload ID "01" constitutes an application with application ID "01", its power consumption is 30W, and a volume that is a storage area is not assigned.

[0049] 6 is a diagram showing an example of the storage management table 2440. The storage management table 2440 shown in FIG.

[0050] Column 2441 stores a volume ID, which is identification information for identifying a volume. Column 2442 stores a workload ID of a computation load constituting an application to which the volume (volume with a volume ID included in the same record) is assigned. Column 2443 stores a copy flag, which indicates whether or not the volume is in a copy state in which data of the volume is copied to another volume. In this embodiment, the copy flag indicates "True" in the case of a copy state, and indicates "False" in the case of no copy state. Column 2444 stores a copy pair ID, which is identification information for identifying a copy pair of the volume. A copy pair is a combination of a copy source and a copy destination of a volume in a copy state, and when a volume is not in a copy state, the copy pair ID indicates "n / a". Column 2445 stores a type of the copy pair of the volume. When the volume is a primary volume that is a copy source, the type indicates "Primary", and when the volume is a secondary volume that is a copy destination, the type indicates "Secondary". Column 2446 stores a capacity of the volume. Column 2447 stores the used capacity of the volume. The sum of the used capacities of the volumes allocated to an application is the amount of data used by the application. Column 2448 stores the device type, which is the type of device (storage 1101) to which the volume is allocated. Column 2449 stores the site ID of the base 1000 that has the volume.

[0051] Each of the records 244A to 244E of the storage management table 2440 indicates storage information. For example, the record 244A indicates that a volume with a volume ID of "01" is assigned to a computation load with a workload ID of "02", the volume is in a copy state, has a copy pair ID of "01", is a primary volume, has a capacity of "500 GB" of which "110 GB" has been used, has a device type of "A", and is provided at the base 1000 with a site ID of "01". The volumes corresponding to the records 244A and 244B respectively form a copy pair with each other. The record 244E indicates a state in which a copy pair is not formed, and data is shared between the sites by the computation load with a workload ID of "04" and the computation load with a workload ID of "08". In this configuration, it is possible to accommodate an access path switching, which is one of the migration types described later.

[0052] 7 is a diagram showing an example of the application management table 2450. The application management table 2450 shown in FIG.

[0053] Column 2451 stores an application ID. Column 2452 stores a migration type, which is a migration method permitted for the application (an application having an application ID included in the same record). Column 2453 stores a policy ID of a placement rule set for the application. Column 2454 stores a site ID of a base where the application is executed.

[0054] In this embodiment, the types of migration include data copy, access path switching, and duplication (Sync (Synchronization): synchronization destination migration). Data copy is a migration method in which, when migrating a computational load, data used in the computational load is copied from the migration source site 1000 to the migration destination site 1000, and the copied data is used after the migration. Access path switching is a migration method in which data used in the computational load is not migrated to the migration destination site 1000, and the migration source data is used even after the migration. Duplication is a migration method in which the execution site is migrated to the site 1000 where the data used in the computational load has been synchronously copied in advance, and the synchronously copied data is used after the migration.

[0055] Each of the records 245A to 245E in the application management table 2450 indicates application information. For example, the record 245A indicates that the types of migration permitted for the application with application ID "01" are "duplication", "data copy" and "access path switching", that the placement rule with policy ID "01" is set for the application, and that the application is executed at the site with site ID "01".

[0056] 8 is a diagram showing an example of the transition time table 2460. The transition time table 2460 shown in FIG.

[0057] Column 2461 stores a workload ID. Column 2462 stores one of the migration types permitted for the computation load (the computation load of the workload ID included in the same record). Column 2463 stores an estimated migration time that is an estimate of the migration time required when the computation load is migrated with the migration type (the migration type of the same record).

[0058] Each of records 246A-246E in the migration time table 2460 indicates migration-related information. For example, record 246A indicates that the estimated migration time due to access path switching for the computational load with workload ID "01" is "10 seconds". Record 246C indicates that the estimated migration time due to "data copy" for the computational load with workload ID "03" is "10 seconds", and record 246D indicates that the estimated migration time due to "access path switching" for the computational load with workload ID "03" is "10 seconds".

[0059] The migration time of the computational load differs depending on the time required to stop the computational load executed at the migration source site 1000, the time required to start the computational load executed at the migration destination site 1000, and the time required to copy the data used in the computational load to the migration destination site 1000. For this reason, the evaluation migration time is calculated based on, for example, the migration type of the computational load, the type of storage that stores the data used in the computational load at the sites 1000 before and after migration, and the amount of data. The multi-site management server 2000 may calculate the evaluation migration time based on, for example, the computational load management table 2430 and the storage management table 2440, and add it to the migration time table 2460.

[0060] For example, as shown in record 244A of the storage management table 2440, the volume with volume ID "01" is a storage area allocated to the computational load with workload ID "02", and the copy flag is "True", so 110 GB of used data has been copied to the volume with volume ID "02" at the site with site ID "02" that has the same copy pair ID "01". Therefore, the multi-site management server 2000 estimates that the time required to switch the computational load is the time required to migrate the computational load, calculates this time as an evaluated migration time, and adds it to the migration time table 2460.

[0061] Also, as shown in record 244E of storage management table 2440, the copy flag of the volume with volume ID "05" is "False", so data copying has not been performed. Therefore, the multi-site management server 2000 identifies that the time required for the migration of the computational load includes the time required for copying the used 450 GB of data to the migration destination, and further identifies the throughput (MB / s) performance of the storage medium from the device type information in storage management table 2440 and the network transfer performance (bps) between the migration source and migration destination sites from the site ID information, calculates the time required for migration based on this information as an evaluated migration time, and adds it to the migration time table 2460. The method of calculating the migration time is not limited to the above-mentioned method, and may be a method of creating and using a machine learning model using data on past migration configurations and migration times, and the effect of the present disclosure does not depend on the calculation method of the migration time.

[0062] 9 is a flowchart for explaining an example of migration plan correction processing for creating a corrected migration plan by the workload migration plan program 2470. Note that, in the following, a case where the computational load of a target base, which is a base to which the computational load is transferred, is reduced will be explained as an example, but the following explanation also applies to a case where the computational load of the target base is increased. For example, a case where the computational load of a target base is increased is equivalent to a case where the computational load of another base is reduced, and the following explanation can be applied to the other base.

[0063] First, the workload migration planning program 2470 acquires error information indicating the error between a first forecast value, which is at least one of the renewable energy supply forecast value and the power demand forecast value of each base 1000 used to create the migration plan, and a second forecast value predicted after the first forecast value (step S1).

[0064] The method of acquiring the error information is not particularly limited. For example, the workload migration planning program 2470 may receive the error information from the power management server 1400 of each base 1000, or may receive information indicating the second predicted value from the power management server 1400 of each base 1000, and calculate the error information based on the received information and the base power management table 2410 stored in the local disk 2400.

[0065] Furthermore, the workload migration planning program 2470 may acquire error information for all bases 1000, or may acquire error information for a specific base 1000. For example, the workload migration planning program 2470 may receive error information for a target base, which is a base to which a computational load is to be migrated, together with information indicating the target base from the power management server 1400 of the target base. The information indicating the target base is, for example, the site ID of the target base.

[0066] When the error information is acquired, the workload migration planning program 2470 identifies the migration type permitted for the application executed at the target site for each computational load based on the computational load management table 2430 and the application management table 2450 (step S2). Specifically, the workload migration planning program 2470 identifies the application executed at the target site based on the site ID in column 2454 of the application management table 2450 and the application ID in column 2451. The workload migration planning program 2470 identifies the migration type permitted for the identified application as the migration type of the computational load based on the workload ID in column 2431 of the computational load management table 2430, the application ID in column 2432 of the computational load management table 2430, and the migration type in column 2452 of the application management table 2450. The target site may be selected based on the site ID received together with the error information, or a site that meets a specific condition, such as a site with the largest error among the error information of each site, may be selected.

[0067] Furthermore, the workload migration planning program 2470 selects a target computation load that can be migrated from among the computation loads executed at the target base based on the placement rule table 2420, the computation load management table 2430, the storage management table 2440, and the application management table 2450 (step S3). Specifically, the workload migration planning program 2470 identifies the policy ID of the placement rule set for the application from the column 2453 of the application management table 2450. The workload migration planning program 2470 identifies the site ID of the base 1000 where the application can be placed based on the identified policy ID and the placement rule table 2420. When selecting a computation load that can be migrated, the workload migration planning program 2470 selects only the computation load that can be migrated based on the information of the computation load management table 2430 and the storage management table 2440 in addition to the information of the migration type permitted for the application identified in step S2. For example, record 244E of storage management table 2440 shows an example of a configuration that can accommodate access path switching as described above, and even if the migration type of an application configured with the computational load to be migrated includes access path switching, if the storage configuration is not one of the configurations exemplified in record 244E of storage management table 2440, it cannot be migrated and is therefore excluded from the migration target. In this way, storage management table 2440 further shows, in the migration type permitted for the computational load, the execution bases at which the computational load can be executed due to the storage configuration.

[0068] The workload migration planning program 2470 starts a loop process A for repeating the candidate selection process from step S5 to step S13 for each selected computational load (step S4).

[0069] In the candidate selection process, the workload migration planning program 2470 first determines whether or not data copy is included in the migration types permitted for the target computational load (Step S5).

[0070] If the migration type includes data copy (step S5: Yes), the workload migration planning program 2470 judges whether the migration time required to migrate the execution base by copying data of the target computational load is shorter than the remaining time from the current time to the specified time (step S6). As the migration time, the evaluated migration time included in the migration time table 2460 may be used, or the workload migration planning program 2470 may calculate it in a similar manner to the method of calculating the evaluated migration time. The specified time is, for example, the time of occurrence of the deviation included in the error information.

[0071] If the migration time is shorter than the remaining time (step S6: Yes), the workload migration planning program 2470 selects the target computational load as a candidate for a migration target load, which is a computational load to migrate the execution base, and adds the candidate for the migration target load to a migration candidate list in association with its migration type (data copy) (step S7).

[0072] Furthermore, if the migration type does not include data copy (step S5: No) and if the migration time is equal to or greater than the remaining time (step S6: No), the workload migration planning program 2470 determines whether the migration types permitted for the target computational load include access path switching (step S8).

[0073] If the migration type includes access path switching (step S8: Yes), the workload migration planning program 2470 judges whether the target computational load after migration of the execution base satisfies a predetermined performance requirement (step S9). Specifically, the workload migration planning program 2470 calculates a performance estimate value evaluating the performance of the target computational load after migration based on the configuration of the migration destination and post-migration bases 1000, and judges whether the performance estimate value satisfies a predetermined performance requirement. The predetermined performance requirement is specified, for example, by a user who uses an application configured by the computational load. The performance estimate value may be calculated from an actual measurement value of network performance between the bases 1000, or may be calculated by another method. The performance estimate value may be calculated in advance and stored in the local disk 2400, etc.

[0074] If the target computational load satisfies the performance requirements (step S9: Yes), the workload migration planning program 2470 selects the target computational load as a candidate for migration target load, associates the candidate for migration target load with its migration type (access path switching), and adds it to a migration candidate list (step S10).

[0075] Furthermore, if the migration type does not include access path switching (step S8: No) and if the target computational load does not satisfy the performance requirements (step S9: No), the workload migration planning program 2470 determines whether the migration types permitted for the target computational load include duplication (step S11).

[0076] If the migration type includes duplication (step S11: Yes), the workload migration planning program 2470 judges whether the migration time required for migrating the execution base due to duplication of the target computing load is shorter than the remaining time from the current time to the specified time (step S12). The specific method of judgment is the same as the process of step S6.

[0077] If the migration time is shorter than the remaining time (step S12: Yes), the workload migration planning program 2470 selects the target computing load as a candidate for migration target load, associates the candidate for migration target load with its migration type (duplication), and adds it to the migration candidate list (step S13).

[0078] Furthermore, if duplication is not included in the migration type (step S11: No), if the migration time is equal to or greater than the remaining time (step S12: No), or if the processing of step S13 is completed, the candidate selection processing for the target computational load is terminated. Then, when the candidate selection processing of steps S5 to S13 is completed for all the target computational loads, the workload migration planning program 2470 terminates loop processing A (step S14).

[0079] Then, the workload migration planning program 2470 selects a migration target load, which is a computational load to which the execution base is actually migrated, from the migration candidate list based on the error information and the computational load management table 2430, creates a migration plan for migrating the migration target load as a corrected migration plan (step S15), and ends the process. In selecting the migration target load, the workload migration planning program 2470 selects a computational load as the migration target load, for example, based on the estimated power consumption of the computational load in the computational load management table 2430 and the deviation amount of the error information, so that the sum of the estimated power amounts corresponds to the deviation amount. Here, the sum of the estimated power amounts corresponds to the deviation amount means that the sum of the estimated power amounts is equal or approximately equal to the deviation amount.

[0080] The above-described migration plan correction process is merely an example, and is not limited thereto. For example, in the loop process, the workload migration plan program 2470 may end the loop process when the computation load is selected until the total estimated power consumption corresponds to the deviation amount, instead of repeating the candidate selection process for all the target computation loads, or may end the loop process when another end condition is satisfied.

[0081] Furthermore, even if the migration type includes access path switching (step S8: Yes), the workload migration planning program 2470 may perform processing to determine whether the migration time is shorter than the time until the specified time, similar to steps S6 and S12.

[0082] Furthermore, if there are multiple discrepancy occurrence times in the error information acquired in step S1, the workload migration planning program 2470 may create multiple time slots based on the multiple discrepancy occurrence times, and perform the processing of steps S2 to S15 for each of the time slots. For example, if the discrepancy occurrence times are "2018 / 4 / 1 10:00" and "2018 / 4 / 1 10:30", the workload migration planning program 2470 creates a 30-minute time slot from "2018 / 4 / 1 10:00" to "2018 / 4 / 1 10:30" and a 30-minute time slot from "2018 / 4 / 1 10:30" to "2018 / 4 / 1 11:00", and performs the processing of steps S2 to S15 for the two time slots.

[0083] When the workload migration plan program 2470 creates a corrected migration plan, the workload migration program 2480 migrates the execution site of the computational load according to the corrected migration plan. At this time, if the migration type of the computational load is duplex, the synchronization state of the data after migration is maintained. In this case, even when the computational load is migrated back to the original site 1000, there is no need to copy the data, and it is possible to swiftly migrate the execution site.

[0084] FIG. 10 is a diagram showing a GUI (Graphical User Interface) that is an example of a display screen that presents the operation results and operation improvement proposals of a DC that uses a migration plan generated by the workload migration plan program 2470.

[0085] The GUI 9000 shown in FIG. 10 is an example of a screen that the workload migration planning program 2470 presents to a user (for example, a user at a data center), and is divided into an upper portion 9100, a middle portion 9200, and a lower portion 9300.

[0086] The upper 9100 is the current arrangement of the DC CO 2 of CO2 emissions reduction effect per year 2 Reduction (Annual CO 2 Reduction) 9110 and the change in placement rules for DC CO 2 Estimated optimal CO emission reduction effect 2 Reduction (Estimated Optimized CO 2 Reduction) 9120. Note that a detailed explanation of the change in the arrangement rule will be omitted.

[0087] The central section 9200 includes a table 9210 showing the result of migration of the computational load (Workload migration result). The table 9210 is information showing the content of a corrected migration plan as a migration plan for resolving the power shortage caused by renewable energy, and shows, for each computational load ID, the application configured by the workload, the migration source base and DC, the migration destination base and DC, the size of the data used, the power consumption, and the like.

[0088] The lower part 9300 includes a table 9310 showing optimization recommendations that propose changes to placement rules as information aimed at improving the migration operation of computational loads. For each computational load ID, the table 9310 shows the applications that make up the workload, the current placement rules, the proposed changes to the placement rules, and the CO2 reduction caused by the changes. 2 For example, the record in the first row of table 9310 indicates that the current placement rule is "01" and that changing the rule to permit execution of the application at site "03" will reduce the amount of CO 2 This indicates that the weight can be reduced by 100 kg.

[0089] In the above description, a computational load is used as a migration unit for migrating an execution base, but an application consisting of one or more computational loads may be used as a migration unit.

[0090] As described above, according to this embodiment, the workload migration planning program 2470 migrates the execution site that executes the computational load according to the migration plan created using the site power management table 2410. The workload migration planning program 2470 also acquires error information indicating the amount of deviation between the first predicted value, which is the value indicated in the site power management table 2410 used to create the migration plan for the target amount, which is at least one of the power demand and the renewable energy supply, and the second predicted value predicted after the first predicted value of the target amount, and the deviation occurrence time, which is the time when the deviation occurs. Then, the workload migration planning program 2470 creates a corrected migration plan by correcting the migration plan based on the error information and the load-related information (2420 to 2460). Therefore, since the migration plan can be corrected according to the error of the predicted value of the power supply amount or the power demand amount from renewable energy, it becomes possible to use the power from renewable energy more efficiently. This reduces CO 2 It is possible to suppress increases in emissions or costs, etc.

[0091] In this embodiment, the load-related information indicates, for each computational load, a migration type, which is a change method for changing the execution base permitted for the computational load. In particular, the migration types include data copy, access path switching, and duplication. Therefore, it is possible to create an appropriate corrected migration plan that takes the migration type into consideration.

[0092] Furthermore, in this embodiment, the workload migration planning program 2470 calculates an estimated migration time based on the migration type and the type of storage that stores the data at the site. Therefore, the estimated migration time can be calculated appropriately.

[0093] Furthermore, in this embodiment, the workload migration planning program 2470 selects a computation load to migrate the execution base from among migration candidate loads, which are computation loads whose estimated migration time is shorter than the time from the current time to the time of the deviation, based on the power consumption and deviation amount of each migration candidate load. Therefore, it is possible to migrate the execution base of a computation load whose migration is completed before an error occurs, and therefore it is possible to create an appropriate corrected migration plan that can use electricity from renewable energy more efficiently.

[0094] In this embodiment, the workload migration planning program 2470 selects a computation load to be migrated so that the sum of the power consumption of the computation loads to be migrated corresponds to the deviation amount. In this case, an appropriate corrected migration plan that can use electricity from renewable energy more efficiently can be created.

[0095] Furthermore, in this embodiment, for a computational load whose migration type is access path switching, the workload migration planning program 2470 selects a computational load to which the execution base is to be migrated from among computational loads that satisfy a predetermined performance requirement after migration. Therefore, it is possible to guarantee the performance requirement after migration.

[0096] The above-described embodiments of the present disclosure are illustrative examples of the present disclosure, and are not intended to limit the scope of the present disclosure to only these embodiments. A person skilled in the art can implement the present disclosure in various other forms without departing from the scope of the present disclosure. [Explanation of symbols]

[0097] 1000: Base 1050: Data Center facility 1100: IT equipment 1101: Storage 1103: Server 1200: Application management server 1300: IT equipment management server 1400: Power management server 1800: Wide area network 2000: Multi-base management server 2100: Management network I / F 2200: Processor 2300: Input / output device 2400: Local disk

Claims

1. A site management device that manages a plurality of sites capable of executing a calculation load that performs calculation processing, A processor and a memory, The memory includes: Power management information indicating, for each of the bases, a predicted value of a target amount, which is at least one of a supply amount, which is the amount of electricity from renewable energy supplied to the base, and a demand amount, which is the amount of electricity consumed at the base; storing load-related information indicating, for each of the computational loads, a power consumption amount that is the amount of power consumed by the computational load, an execution base that is the base that executes the computational load, and an estimated transition time that is an estimated transition time required to transition the execution base to another base; The processor, migrating the execution base of each computing load according to a migration plan for migrating the execution base of each computing load, the migration plan being created using the power management information; obtain error information indicating a deviation amount, which is a magnitude of deviation between a first predicted value, which is the predicted value indicated in the power management information used to create the migration plan, and a second predicted value predicted after the first predicted value of the target amount, and a deviation occurrence time, which is a time at which the deviation occurs; The site management device creates a corrected transition plan by correcting the transition plan based on the error information and the load-related information.

2. The site management device according to claim 1 , wherein the load-related information further indicates, for each of the computational loads, a migration type that is a change method for changing the execution site permitted for the computational load.

3. The site management device according to claim 2 , wherein the load-related information further indicates the execution site capable of executing the computational load.

4. The base management device of claim 2, wherein the migration types include a data copy in which data used in the computational load is copied from the source of the execution base to the destination of the execution base when migrating the execution base, an access path switching in which data used in the computational load is not copied, and a synchronization destination migration in which the execution base is migrated to a base to which the data used in the computational load is synchronously copied.

5. The site management device according to claim 2 , wherein the processor calculates the evaluated migration time based on the migration type and a type of storage that stores data at the site.

6. The base management device of claim 1, wherein the processor creates the corrected migration plan by selecting a computational load to migrate the execution base from among migration candidate loads, which are computational loads whose evaluation migration time is shorter than the time from the current time to the time when the deviation occurs, based on the power consumption of each of the migration candidate loads and the deviation amount.

7. The base management device according to claim 6, wherein the processor creates the corrected migration plan by selecting, from the migration candidate loads, a computational load to which the execution base is to be migrated such that a sum of the power consumption of the computational loads to which the execution base is to be migrated corresponds to the deviation amount.

8. 5. The base management device according to claim 4, wherein the processor creates the corrected migration plan for the computational loads whose migration type is the access path switching, by selecting computational loads to which the execution base is migrated from among the computational loads whose migration type satisfies specified performance requirements based on the power consumption of each of the computational loads and the deviation amount.

9. The base management device of claim 1, wherein the processor receives information indicating a target base, which is the base to which the computational load is to be transferred, and creates the corrected transfer plan by selecting a computational load to be transferred to the execution base from among the computational loads being executed at the target base.

10. A site management method by a site management device that manages a plurality of sites capable of executing a calculation load that performs calculation processing, comprising: The base management device includes a processor and a memory. The memory includes: Power management information indicating, for each of the bases, a predicted value of a target amount, which is at least one of a supply amount, which is the amount of electricity from renewable energy supplied to the base, and a demand amount, which is the amount of electricity consumed at the base; storing load-related information indicating, for each of the computational loads, a power consumption amount that is the amount of power consumed by the computational load, an execution base that is the base that executes the computational load, and an estimated transition time that is an estimated transition time required to transition the execution base to another base; The processor, migrating the execution base of each computing load according to a migration plan for migrating the execution base of each computing load, the migration plan being created using the power management information; obtain error information indicating a deviation amount, which is a magnitude of deviation between a first predicted value, which is the predicted value indicated in the power management information used to create the migration plan, and a second predicted value predicted after the first predicted value of the target amount, and a deviation occurrence time, which is a time at which the deviation occurs; A site management method, comprising: creating a corrected transition plan by correcting the transition plan based on the error information and the load-related information.

Images