Data center system, inter-base workload control method, and inter-base workload control system

JP2024169095APending Publication Date: 2024-12-05HITACHI VANTARA LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2023086288
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-25
Publication Date
2024-12-05

AI Technical Summary

Benefits of technology

【0016】 本発明によれば、ワークロードの移行不可能時間帯を考慮しつつワークロードのDC拠点間の空間制御とDC拠点内での時間制御を実行することで、DC拠点全体において再エネ率を向上するようにDCの電力需要を調整することができる。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024169095000001_ABST
    Figure 2024169095000001_ABST
Patent Text Reader

Abstract

To adjust power demand at each base so as to improve a renewable energy utilization ratio at all bases.SOLUTION: An inter-base workload control system manages an amount of excess power obtained by subtracting an amount of renewable energy power supplied from an amount of power consumed in association with execution of a workload in a future time range at each base, spatial migratable time range information on a spatial migratable time range where the workload in a future time range at the bases can be spatially migrated to another base, temporal migratable time range information on temporal migration of delaying execution of the workload in the future time range at the bases within the same base and migrating the workload to another time range, and a predicted amount of power to be consumed by execution of the workload in the future time range at the bases. The inter-base workload control system determines a power adjustment amount that moves through the spatial migration and the temporal migration of each workload in the future time range at the bases so that the sum total of excess power amounts in the future time range at the plurality of bases becomes smaller, on the basis of the above mentioned information.SELECTED DRAWING: Figure 9
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a data center system, an inter-DC workload control method, and an inter-DC workload control system. [Background technology]

[0002] So-called decarbonization, which aims to move away from fossil fuels in order to prevent the emission of greenhouse gases such as carbon dioxide, which causes global warming, has been attracting attention. In this regard, a data center (DC) is equipped with a large number of information processing devices and communication devices to execute a certain processing load (workload), and these devices require a large amount of electricity to operate. Therefore, attempts are being made to achieve decarbonization by covering such electricity with renewable energy.

[0003] In this case, it is important to maintain the ratio of renewable energy usage to power consumption (renewable energy utilization rate: renewable energy rate), but in order to realize a cleaner DC, it is preferable to maintain this renewable energy rate at a finer time granularity (for example, hourly rather than daily).

[0004] Another problem with renewable energy is that its supply is highly unstable as it depends on factors such as the weather. As dependence on renewable energy increases, power suppliers will need to have more adjustment capacity (the amount of power needed to match electrical demand and supply) in order to ensure a stable power supply, which places a burden on them.

[0005] Therefore, there are attempts to control the power consumed by a DC by controlling the workload executed at the DC either temporally or spatially between multiple DC sites (a collection of DCs in the same area), thereby improving the renewable energy rate of the DC and using the DC as a regulating force for the power grid.

[0006] Here, controlling the workload in terms of time means temporarily delaying the execution of the workload within the same data center. This allows, for example, delaying a large number of workloads from a time period when there is a shortage of renewable energy supply to a time period when there is an excess of renewable energy supply, thereby increasing the renewable energy ratio.

[0007] Spatial control of workloads means moving the execution location of the workload from one DC site to another. As with temporal control, this allows the renewable energy ratio to be increased by moving the workload from a DC site with a scarce renewable energy supply to a DC site with an abundant supply.

[0008] Furthermore, workloads executed at a DC site can generally be broadly divided into two types: interactive workloads and batch jobs. Interactive workloads are those that require real-time processing, such as web applications, and are basically always in a running state. Batch jobs are those that do not necessarily require immediate processing, such as image processing and machine learning learning processing, and only need to be executed by a set time, and processing can be delayed within that range.

[0009] Here, although interactive workloads can be controlled spatially, they are not suitable for temporal control because processing cannot be delayed. On the other hand, batch jobs can be controlled temporally because processing can be delayed. Therefore, for ease of understanding, we will refer to the former as a workload for spatial migration and the latter as a workload for temporal migration.

[0010] Patent Document 1 discloses a technology that efficiently utilizes renewable energy generated at each DC location by transferring workload execution processing from a DC location where a renewable energy supply is predicted to be insufficient to a DC where a renewable energy supply is predicted to be in surplus. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] Patent Publication No. 2021-189845 Summary of the Invention [Problem to be solved by the invention]

[0012] By using the technology of Patent Document 1, it is possible to spatially control the workload according to the amount of renewable energy supply at each DC site and adjust the ratio of the workload executed at each DC site. This makes it possible to improve the renewable energy rate at each DC site.

[0013] However, workloads executed at DC sites cannot always be freely migrated between DC sites. When a workload is migrated between DC sites, the workload will experience some downtime, so the time period during which the user who executes the workload can migrate between DC sites may be restricted. For example, if the workload is something like a web application, downtime during times when users access it frequently would cause problems, so it may be set to not be spatially migrated. During such times, the workload cannot be spatially controlled, so the power demand of the DC may not be adjusted.

[0014] The present invention has been made in consideration of the above background, and has an object to provide a data center system, an inter-DC site workload control method, and an inter-DC site workload control system that are capable of adjusting DC power demand so as to improve the renewable energy rate across all DC sites by performing spatial control between DC sites and time control within a DC site of workload while taking into account time periods when workload migration is not possible. [Means for solving the problem]

[0015] As one aspect for solving the above problem, there is provided a data center system having a plurality of bases each having a computational resource and a renewable energy power source for executing a workload, and an inter-base workload control system for controlling the migration of the workload between the plurality of bases and within the same base, wherein a processor of the inter-base workload control system calculates a positive excess power amount obtained by subtracting a predicted value of the power supply amount of the renewable energy power source from a predicted value of the power consumption amount of the computational resource associated with the execution of the workload in a future time period at each of the bases, spatial migration possible time period information regarding time periods during which the spatial migration is possible and impossible for the workload to be executed in the future time period at each of the bases to another base, and a spatial migration possible time period information regarding the time periods during which the spatial migration is possible and impossible for the workload to be executed in the future time period at each of the bases. The method manages time transfer available time information regarding time transfer in which execution of a load is delayed within the same location and transferred to another time zone, and power consumption predicted values ​​which are predicted values ​​of the power consumption of the computing resources associated with the execution of the workloads scheduled to be executed in the future time zones at each of the locations, and based on the spatial transfer available time zone information, the time transfer available time information, and the power consumption predicted values, determines a power adjustment amount indicating the predicted power consumption value to be moved by the spatial transfer and the time transfer of each of the workloads scheduled to be executed in the future time zones at each of the locations so that the total amount of excess power in the future time zones at the multiple locations is smaller, and performs the spatial transfer and the time transfer of the workloads based on the power adjustment amount. Effect of the Invention

[0016] According to the present invention, by performing spatial control between DC sites and time control within a DC site of workloads while taking into account time periods when workload migration is not possible, it is possible to adjust the power demand of DCs so as to improve the renewable energy rate across all DC sites. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a diagram illustrating an example of a configuration of a data center system according to an embodiment of the present invention. [Diagram 2]FIG. 2 is a diagram illustrating an example of the configuration of a site management computer. [Diagram 3] FIG. 1 is a diagram illustrating an example of a configuration of an inter-site workload control system. [Figure 4] FIG. 13 is a diagram showing an example of a renewable energy power table. [Diagram 5] FIG. 4 is a diagram illustrating an example of an electricity demand table. [Figure 6] FIG. 13 illustrates an example of a workload table. [Figure 7] FIG. 13 is a diagram illustrating an example of a spatial power transfer table. [Figure 8] FIG. 13 is a diagram illustrating an example of a supply and demand adjustment amount ratio table. [Figure 9] FIG. 11 is a flow diagram for explaining an overview of a workload control planning process. [Figure 10] FIG. 11 is a flow diagram illustrating details of each base data update process. [Figure 11] FIG. 11 is a flow diagram illustrating a workload control execution process. [Figure 12] FIG. 11 is a flow diagram illustrating the details of a workload reallocation planning process. [Figure 13] FIG. 11 is a flow diagram illustrating details of a time-shifted power determination process. [Figure 14] FIG. 13 is a diagram illustrating an example of a workload input screen. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. This embodiment is merely an example for realizing the present invention, and does not limit the technical scope of the present invention. Not all of the elements and combinations thereof described in this embodiment are necessarily essential to the solution of the invention.

[0019] In the following description, the same components are generally assigned the same reference numerals, and repeated description is omitted. The number of each component in the following description is not limited unless otherwise noted. In the following description, the same components are described with distinction by using reference numerals including subscripts, and are described without distinction by using reference numerals excluding subscripts.

[0020] In the following explanation, the processing performed by a program may be described. A computer performs processing defined by a program using a processor (e.g., a CPU (Central Processing Unit), a GPU (Graphics Processing Unit)) while using memory in a main storage device, etc. Therefore, the processor may be the entity that performs processing by executing a program. The processor executes a program to realize a functional unit that performs processing.

[0021] Similarly, the subject of processing performed by executing a program may be a controller, device, system, computer, or node having a processor. The subject of processing performed by executing a program may be a computing unit, and may include a dedicated circuit that performs specific processing. The dedicated circuit is, for example, a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC).

[0022] In the following description, a program may be installed in a computer from a program source. The program source may be, for example, a program distribution server or a non-transitory storage medium readable by a computer. When the program source is a program distribution server, the program distribution server may include a processor and a storage resource (storage) for storing a program to be distributed, and the processor of the program distribution server may distribute the program to be distributed to other computers. In addition, in an embodiment, two or more programs may be realized as one program, and one program may be realized as two or more programs.

[0023] In the following description, various data will be described in a table format, but the data format is not limited to the table format and may be other data formats such as a queue, a list, or CSV (Comma Separated Value).

[0024] In the following description, an ID (identifier), a number, and the like are used as information for identifying an object, but the present invention is not limited to these, and various types of identification information may be used.

[0025] FIG. 1 is a diagram showing an example of the configuration of a data center system S according to this embodiment.

[0026] The data center system S includes one or more DC (Data Center) locations 1000, and an inter-location workload control system 10000. The DC locations 1000 are connected to each other via a wide area network 9000 so as to be able to communicate with each other.

[0027] The DC base 1000 includes a base management computer 2000, one or more servers 3000 used by an administrator or a user (user) of the DC base 1000, and one or more storages 4000 used by the administrator or the user of the DC base 1000. The server 3000 and the storage 4000 are communicatively connected to each other via a data network 6000. In addition, the base management computer 2000, the server 3000, and the storage 4000 are communicatively connected to each other via a management network 5000.

[0028] The management network 5000, data network 6000, and wide area network 9000 are, for example, the Internet, a Local Area Network (LAN), a Wide Area Network (WAN), or a wired or wireless communication network such as a dedicated line.

[0029] The power to drive the site management computer 2000, the server 3000, and the storage 4000 is supplied from a power supply bus 8000 in the DC site 1000, and can be supplied by a renewable energy power source 7000 in the DC site 1000 or an external system power source 11000. The renewable energy power source 7000 is a renewable energy power generation facility such as solar power or wind power installed in the DC site 1000, and the power generated in the DC site 1000 can be used within the DC site 1000.

[0030] The server 3000 and the storage 4000 execute various types of processing, such as the above-mentioned workload for spatial migration and workload for time migration.

[0031] The inter-site workload control system 10000 allocates the processing load (workload) of the system instructed by the user to each DC site 1000. In addition, it appropriately migrates the workload already being executed at each DC site to another DC site.

[0032] The local management computer 2000 manages the workload allocated to each DC site 1000 by the inter-site workload control system 10000 in units of a predetermined time interval (time slot), and controls the timing of executing the workload within each DC site.

[0033] In this embodiment, the workload may refer to the job (process) itself. In addition, hereinafter, "renewable energy" may be abbreviated to "re-energy." In addition, in the following, the time slot interval is set to one hour, but the time slot interval is not limited to this interval and can be set to any time interval.

[0034] FIG. 2 is a diagram illustrating an example of the configuration of the local management computer 2000. As shown in FIG.

[0035] The local management computer 2000 includes a storage device 12000 such as a hard disk drive (HDD) or a solid state drive (SSD). The local management computer 2000 also includes a processing device 13000 (processor) such as a central processing unit (CPU), a digital signal processor (DSP), a graphics processing unit (GPU), or an FPGA. The local management computer 2000 also includes a main storage device 14000 (memory) such as a read only memory (ROM) or a random access memory (RAM). The local management computer 2000 also includes a communication device 15000 configured with a network interface card (NIC), a wireless communication module, a universal serial bus (registered trademark), a USB (USB, the same applies below) module, or a serial communication module. The local management computer 2000 also includes an input device 16000 configured with a mouse, a keyboard, or the like, and an output device 17000 configured with a liquid crystal display or an organic electroluminescence (EL) display, or the like.

[0036] The base management computer 2000 also stores an intra-base workload control program 12500, a power demand forecasting program 12600, and a renewable energy forecasting program 12700.

[0037] The intra-site workload control program 12500 calculates a target value for the amount of power consumption in future time slots so as to maximize the renewable energy rate, based on the predicted value of the amount of power consumption in future time slots in the DC site 1000. Also, based on the determined target value of the amount of power consumption, it determines the execution timing of each workload in the future time slots, and executes each workload at the determined timing.

[0038] The power demand forecasting program 12600 calculates a forecast value of the amount of power to be consumed by the DC site 1000 in a future time slot.

[0039] The renewable energy prediction program 12700 calculates a predicted value of the amount of power of renewable energy to be supplied to the DC site 1000 in a future time slot. The renewable energy to be supplied to the DC site 1000 may be generated by a renewable energy source 7000 or may be supplied through a PPA (Power Purchase Agreement) or the like.

[0040] Furthermore, the site management computer 2000 stores the following databases: a renewable energy power table 12100, a power demand table 12200, a workload table 12300, and a spatial power transfer table 12400.

[0041] The renewable energy power table 12100 stores the predicted value of the supply amount of renewable energy predicted by the renewable energy prediction program 12700 in each time slot, and the actual value thereof.

[0042] The power demand table 12200 stores predicted and actual values ​​of the amount of power consumed by the DC site 1000 in each time slot, and a power consumption target value.

[0043] The workload table 12300 stores and accumulates information regarding predicted and actual power consumption values, execution schedules, time when spatial migration is possible, time when spatial migration is not possible, and the like, of each workload executed in the DC base 1000. The server 3000 and the storage 4000 execute each workload in accordance with this workload table 12300.

[0044] The spatial power migration table 12400 stores information on the amount of power of a workload for spatial migration that is to be relocated from one DC site 1000 to another DC site 1000 in each time slot.

[0045] FIG. 3 is a diagram illustrating an example of the configuration of an inter-site workload control system.

[0046] The inter-site workload control system 10000 includes a storage device 18000 such as an HDD or SSD, a processing device 19000 (processor) such as a CPU, DSP, GPU, or FPGA, and a main storage device 20000 (memory) such as a ROM or RAM. The inter-site workload control system 10000 also includes a communication device 21000 configured with a NIC, a wireless communication module, a USB module, a serial communication module, or the like. The inter-site workload control system 10000 also includes an input device 22000 configured with a mouse, keyboard, or the like, and an output device 23000 configured with a liquid crystal display, an organic EL display, or the like.

[0047] The inter-site workload control system 10000 also stores an inter-site initial placement program 18300, an inter-site relocation program 18400, and an inter-site data collection program 18500.

[0048] The inter-site initial placement program 18300 calculates the ratio of the workload to be initially placed at each DC site in the future time slot based on the renewable energy supply amount and power demand in the future time slot at each DC site 1000. Then, at that time, control is performed to allocate the workload to each DC site 1000 based on the calculated ratio.

[0049] The inter-site reallocation program 18400 determines how much power of a workload for spatial migration already being executed at a DC site should be reallocated to another DC site in a future time slot. This power amount is determined based on information on the renewable energy supply amount, power demand, time period when spatial migration is possible, and time period when spatial migration is not possible in the future time slot at each DC site 1000. Then, the workload is reallocated at the time of the future time slot.

[0050] The inter-site data collection program 18500 collects information on the power demand and renewable energy supply amount predicted by the power demand prediction program 12600 and the renewable energy prediction program 12700 at each DC site 1000.

[0051] Furthermore, the inter-site workload control system 10000 stores the databases of site data 18100 and a supply and demand adjustment amount ratio table 18200 .

[0052] The base data 18100 stores information such as power demand and renewable energy supply collected from each DC base.

[0053] The supply and demand adjustment amount ratio table 18200 stores the supply and demand adjustment amount ratio indicating the ratio at which the initially allocated workload is allocated to which DC base for each time slot.

[0054] Next, specific examples of each database will be described.

[0055] (Renewable Energy Electricity Table) 4 is a diagram showing an example of a renewable energy power table 12100 for each DC site 1000. The renewable energy power table 12100 has data items of a time slot ID 12110 in which identification information of a time slot is set, and a time 12120 in which the start time of the time slot is set. The renewable energy power table 12100 also has a data item of a renewable energy supply forecast 12130 in which a forecast value of a renewable energy supply amount at a target time is set. The renewable energy power table 12100 also has a data item of a renewable energy supply actual measurement 12140 in which an actual measurement value of a renewable energy supply amount actually measured at a target time is set. The renewable energy power table 12100 is composed of one or more records having these data items.

[0056] The renewable energy supply forecast 12130 is a forecast value of the amount of power supplied by the renewable energy power source (renewable energy power source 7000) in a future time slot at each DC site 1000, and the amount of power supplied / purchased from outside through PPA or the like.

[0057] The predicted values ​​and actual measured values ​​in the renewable energy power table 12100 may be input by the user, or may be automatically obtained from a predetermined database.

[0058] (Electricity demand table) FIG. 5 is a diagram showing an example of a power demand table 12200 for each DC base 1000. The power demand table 12200 has data items of a time slot ID 12210 in which identification information of a time slot is set, and a time 12220 in which a start time of the time slot is set. The power demand table 12200 also has a data item of a DC power consumption prediction 12230 in which a predicted value of the power consumption of the DC base 1000 at a target time is set. The power demand table 12200 also has a data item of a DC power consumption actual measurement 12240 in which an actual measured value of the power consumption of the DC base 1000 actually measured at a target time is set. The power demand table 12200 also has a data item of a power consumption target value 12250 indicating the amount of power to be targeted when the intra-base workload control program 12500 actually deploys the workload at the target time. The power demand table 12200 is composed of one or more records having these data items.

[0059] The DC power consumption prediction 12230 is a predicted value of the power consumption including the computational resources (servers 3000, storage 4000) and network devices and air conditioning (not shown) associated with the execution of the workload in a future time slot at each DC site 1000.

[0060] The predicted values ​​and actual measured values ​​in the power demand table 12200 may be input by the user, or may be automatically obtained from a predetermined database.

[0061] (Workload table) FIG. 6 is a diagram showing an example of the workload table 12300. The workload table describes information on the workload allocated through the inter-site workload control system 10000 in each DC site 1000. The workload table 12300 has data items of a workload ID 12305 in which identification information of the workload is set, and a power consumption prediction 12310 in which a predicted value of the power consumption of the workload is set. The workload table 12300 also has data items of an actual power consumption 12315 in which an actual measured value of the power consumption of the workload is set, and an input time 12320 in which the time when the information of the workload is input to the site management computer 2000 is set. In the case of a workload for time shift, the workload table 12300 also has a data item of an execution schedule 12325 indicating the time at which the execution of the workload may be started. In addition, the workload table 12300 has a data item of a changed execution schedule 12330 in which the execution timing changed (delayed) from the execution schedule 12325 by the intra-site workload control program 12500 is set for the workload. The workload table 12300 also has data items of a delayable time 12335 indicating the time that the workload can be delayed when the workload is a workload for time migration, and an executable site 12340 in which the DC site 1000 in which the workload can be executed is set. For a workload for time migration, if the delayable time 12335 is 0 or if the intra-site workload control program 12500 judges that there is no need to delay, the post-change execution schedule 12330 becomes n / a, and the workload is executed at the time of the execution schedule 12325. If the delayable time 12335 is set sufficiently long and the intra-site workload control program 12500 judges that the workload should be delayed, the time is written in the post-change execution schedule 12330 and the workload is executed at that time. The workload table 12300 also has data items of a spatial migration possible time zone 12345 in which a time zone in which the workload can be spatially migrated when the workload is a workload for spatial migration is set.The workload table 12300 also has a data item of spatial migration impossible time period 12350, which sets a time period during which spatial migration is impossible when the workload is a workload for spatial migration. The workload table 12300 also has a data item of execution status 12355, which sets the execution status of the workload. The workload table 12300 is made up of one or more records having these data items.

[0062] The possible delay time 12335 is time shift possible time information regarding time shifting, in which a workload scheduled to be executed in a future time slot at each DC site is delayed in execution within the same DC site 1000 and shifted to another time slot (time slot).

[0063] The spatial migration possible time slot 12345 is information regarding a time slot during which spatial migration is possible, in which a workload scheduled to be executed in a future time slot (time slot) at each DC site 1000 can be migrated to another DC site 1000. The spatial migration impossible time slot 12350 is information regarding a time slot during which spatial migration is impossible, in which a workload scheduled to be executed in a future time slot (time slot) at each DC site 1000 can be migrated to another DC site 1000. The spatial migration possible time slot 12345 and the spatial migration impossible time slot 12350 are spatial migration possible time slot information.

[0064] In the case of a workload for time migration, the user inputs the workload by describing the execution schedule 12325, the allowable delay time 12335 indicating the maximum time that may be delayed from the execution schedule 12325, and the executable location 12340. In the case of a workload for spatial migration, the user inputs the workload by describing the spatial migration possible time period 12345 and the spatial migration impossible time period 12350. In the case of a workload for spatial migration, the workload is executed immediately at the time it is input, so there is no information on the execution schedule 12325 and the allowable delay time 12335.

[0065] (Spatial power transfer table) 7 is a diagram showing an example of a spatial power transfer table 12400. The spatial power transfer table 12400 has data items of a time slot ID 12410 in which identification information of a time slot is set, and a time 12420 in which a start time of the time slot is set. The spatial power transfer table 12400 also has a data item of a DC site (N) 12430 indicating the amount of power of a workload for spatial transfer to be relocated to a DC site 1000 other than itself at a target time. The spatial power transfer table 12400 is composed of one or more records having these data items.

[0066] The number of columns stored in the DC site (N) 12430 is equal to the number of other DC sites 1000 excluding the DC site 1000 itself in which this table is stored. For example, if the number of DC sites 1000 is M, the number of columns stored in the DC site (N) 12430 is M-1.

[0067] (Supply and demand adjustment ratio table) 8 is a diagram showing an example of a supply and demand adjustment amount ratio table 18200. The supply and demand adjustment amount ratio table 18200 has data items of a time slot ID 18210 in which identification information of a time slot is set, and a time 18220 in which the start time of the time slot is set. The supply and demand adjustment amount ratio table 18200 also has a data item of a DC site (N) 18230 indicating at what ratio the workload input to the inter-site workload control system 10000 is to be allocated to each DC site 1000 at the target time. The supply and demand adjustment amount ratio table 18200 is made up of one or more records having these data items.

[0068] The number of columns stored in the DC site (N) 18230 is equal to the number of DC sites 1000. For example, if the number of DC sites 1000 is M, the number of columns stored in the DC site (N) 18230 is M.

[0069] If each of the programs described above is executed by the site management computer 2000, it is executed by the processing device 13000 reading out (the program stored in the main memory device 14000 or the memory device 12000). Similarly, if each of the programs is executed by the inter-site workload control system 10000, it is executed by the processing device 19000 reading out (the program stored in the main memory device 20000 or the memory device 18000). Each of the programs can be distributed, for example, by recording it on a recording medium.

[0070] The local management computer 2000 and the inter-site workload control system 10000 may be realized, in whole or in part, by using virtual information processing resources provided by using virtualization technology, process space separation technology, or the like. For example, they may be virtual servers provided by a cloud system. Also, all or in part of the functions provided by the local management computer 2000 and the inter-site workload control system 10000 may be realized, for example, by a service provided by the cloud system via an API (Application Programming Interface) or the like.

[0071] Next, the processing executed by the site management computer 2000 and the inter-site workload control system 10000 will be described.

[0072] <Workload control planning process> 9 is a flow diagram for explaining an overview of a workload control planning process, which is a process for planning in advance how to control the workload in a future time slot in the DC base 1000. Workload control is executed by repeating two processes: the workload control planning process (FIG. 9) and a workload control execution process (FIG. 11) described later.

[0073] First, the workload control planning process is repeatedly executed at a predetermined time (e.g., once a day), at a predetermined time interval (e.g., a predetermined time before the start of each time slot), or at a predetermined timing (a time designated by a user) before actually controlling the workload. For example, on the day before actually controlling the workload, a plan is made on how to control the workload on the next day. However, on the day before, it is not possible to know exactly what kind of workload will be input on the day of control. Therefore, instead of making a control plan for each individual workload, a rough policy is made on how much or what proportion of the amount of power predicted to be consumed by the workload in each future time slot should be controlled. The control is a temporal or spatial transition.

[0074] The inter-site workload control system 10000 calculates a supply and demand adjustment ratio that indicates the ratio at which the input workload should be allocated in each future time slot. It also creates a plan for how much of the power consumed by the space control workload that has been executed at a certain DC site 1000 should be transferred to another DC site 1000.

[0075] The site management computer 2000 makes a plan as to how to control the amount of power consumed by the time-shifting workload in the DC site 1000 in terms of time in each future time slot.

[0076] Then, in the workload control execution process, at the actual time slot, control of individual workloads is performed based on the spatial control and time control policies (how much power to shift temporally or spatially) planned in the workload control planning process.

[0077] First, in step S1000, the base management computer 2000 executes a base data update process for predicting the amount of renewable energy supply and the amount of DC power consumption in each DC base 1000 and storing the past data thereof. The details of the base data update process (S1000) will be described later.

[0078] Next, in step S2000, the inter-site data collection program 18500 in the inter-site workload control system 10000 collects update data and stores it in each site data 18100. The update data is the predicted and actual values ​​of the renewable energy supply and DC power consumption at all DC sites 1000 updated in step S1000.

[0079] Next, in step S3000, the inter-site initial arrangement program 18300 in the inter-site workload control system 10000 calculates a supply and demand adjustment amount ratio in a future time slot based on the information of each DC site 1000 collected in step S2000. Then, the supply and demand adjustment amount ratio is stored in the DC site (N) 18230 in the supply and demand adjustment amount ratio table 18200. The supply and demand adjustment amount ratio may be calculated based on the supply and demand balance of the renewable energy supply amount and the power consumption amount at each DC site 1000, as described in Patent Document 1, for example, or another method may be used.

[0080] Furthermore, in each DC base 1000, the base management computer 2000 calculates the target power consumption for a future time slot based on the supply and demand adjustment amount ratio determined by the inter-base workload control system 10000, the renewable energy supply amount at the base, and the predicted value of the DC power consumption. Then, the target power consumption is stored in the power consumption target value 12250 in the power demand table 12200. Any method may be used to calculate the target power consumption as long as it is a method for determining the amount of power that improves the renewable energy rate.

[0081] Next, in step S4000, the inter-site relocation program 18400 in the inter-site workload control system 10000 executes a workload relocation planning process. In the workload relocation planning process, a plan is made to relocate workloads that have already been executed in each DC site 1000 in a future time slot to another DC site 1000. The workload relocation planning process (S4000) will be described in detail later.

[0082] Next, in step S5000, the control parameters of each DC site 1000 updated by the inter-site workload control system 10000 in step S4000 are synchronized with each DC site 1000. The above processing is executed repeatedly.

[0083] <Data update process for each location> Next, the details of the each base data update process (S1000) will be described. Fig. 10 is a flow diagram illustrating the details of the each base data update process (S1000). This process is executed in all DC bases 1000.

[0084] First, in step S1100, the renewable energy prediction program 12700 predicts the supply amount of renewable energy in each time slot after the present. Specifically, for example, the renewable energy prediction program 12700 acquires each value of the time 12120 and the actual renewable energy supply amount 12140 of each record of the renewable energy power table 12100. Then, based on a predetermined algorithm (for example, performing a time series analysis or creating a prediction model by machine learning) for the acquired value, the renewable energy supply amount in each time slot after the present is predicted. The renewable energy prediction program 12700 stores each predicted supply amount in the renewable energy supply amount prediction 12130 of the record of each time slot of the renewable energy power table 12100.

[0085] The data used by the renewable energy prediction program 12700 to predict the amount of renewable energy supply is not limited to the actual measured value of the amount of renewable energy supply shown here. In other words, other information such as weather data at the DC base 1000 may be used in combination.

[0086] Furthermore, the renewable energy prediction program 12700 acquires the amount of renewable energy supply in past time slots from a predetermined device (for example, an external database or server). Then, the acquired supply amount is stored as an actual value in the renewable energy supply amount actual measurement 12140 of the record related to the time slot in the renewable energy power table 12100.

[0087] Next, in step S1200, the power demand prediction program 12600 predicts the amount of power consumption in each time slot after the present. Specifically, for example, the power demand prediction program 12600 acquires each value of the time 12220 and the DC power consumption actual measurement 12240 of each record of the power demand table 12200. Then, based on a predetermined algorithm (for example, executing a time series analysis, creating a prediction model by machine learning) for the acquired values, it predicts the amount of power consumption of the entire DC site in each time slot after the present. The power demand prediction program 12600 stores each predicted amount of power consumption in the DC power consumption prediction 12230 of the record of each time slot of the power demand table 12200.

[0088] The data used by the power demand forecasting program 12600 to forecast the amount of power consumption is not limited to the actual measured values ​​of the amount of power consumption shown here, and may be used in combination with other information such as weather data at the DC base 1000.

[0089] The power demand forecasting program 12600 also acquires the power consumption of the entire DC site in the DC site 1000 in the past time slot from a predetermined device (for example, an external database or server). Then, the acquired power consumption is stored as an actual value in the DC power consumption actual measurement 12240 of the record related to the time slot in the power demand table 12200.

[0090] <Workload control execution process> The workload control execution process is a process that determines when and at which DC site 1000 an individual workload will be executed, based on the control policy for each time slot planned in the workload control planning process, and then executes the determined result. The workload control execution process consists of two processes: a workload initial placement control process and a workload reallocation control process. The workload initial placement control process is a control process for a workload before execution, and the workload reallocation control process is a control process for a workload that is already being executed at one of the DC sites 1000.

[0091] FIG. 11 is a flow diagram illustrating the workload control execution process.

[0092] First, in step S6000, the inter-site initial placement program 18300 and the intra-site workload control program 12500 perform workload initial placement control processing.

[0093] When a user has a workload that he or she wishes to execute at one of the DC sites 1000, the user issues an instruction to submit the workload to the inter-site workload control system 10000. At this time, the user takes into consideration the characteristics of the workload to be executed and specifies the DC sites 1000 that are permitted to execute the workload. For example, since only DC sites 1 and 2 satisfy SLA (Service Level Agreement) such as latency, it may be permitted to place the workload at DC sites 1 and 2, but not at other DC sites 1000. In addition, if the workload is a time-shifted workload, information such as the permitted delay from submission to execution is specified.

[0094] The inter-site initial placement program 18300 determines to which DC site 1000 the workload should be placed, based on the supply and demand adjustment amount ratio 18230 calculated in step S3000 of the workload control planning process and information on which DC site 1000 the workload can be placed.

[0095] A workload allocated to one of the DC sites 1000 by the inter-site initial placement program 18300 is executed immediately if it is a workload for spatial control. If it is a workload for time control, the intra-site workload control program 12500 determines an appropriate execution time for the workload for time control based on the workload power consumption prediction 12310 and stores it in the changed execution schedule 12330. At that time, the information of the allowable delay time 12335 is taken into consideration so that the power consumption in each time slot is as close as possible to the power consumption target value 12250. Then, when the time set in the changed execution schedule 12330 arrives, the workload is executed.

[0096] The placement control to the DC site 1000 performed by the inter-site initial placement program 18300 and the time control performed by the intra-site workload control program 12500 are collectively called a workload initial placement control process.

[0097] Next, in step S7000, the inter-site relocation program 18400 performs workload relocation control processing. The inter-site relocation program 18400 determines the workload to be migrated to each DC site 1000 so as to approximate the power amount stored in the DC site (N) 12430 of the space power transfer table 12400 based on the workload power consumption prediction 12310. Then, the inter-site relocation program 18400 actually performs the migration.

[0098] <Workload relocation planning process> FIG. 12 is a flow diagram illustrating the details of the workload reallocation planning process (S4000).

[0099] First, in step S4110, the inter-site relocation program 18400 selects the first one of the DC sites 1000. Any DC site 1000 may be selected.

[0100] Next, in step S4120, the inter-base reallocation program 18400 selects the nearest time slot. Specifically, the inter-base reallocation program 18400 refers to the renewable energy power table 12100, and selects a record in which the time 12120 indicates the closest future time to the current time.

[0101] Next, in step S4130, the inter-base reallocation program 18400 determines whether or not there is a power excess in the currently targeted time slot. The power excess (excess power amount) is a positive value obtained by subtracting the predicted value of the power supply amount of the renewable energy power source (renewable energy power source 7000) from the predicted value of the power consumption amount including the computational resources and the network equipment and air conditioning (not shown) accompanying the execution of the workload in the future time slot (time slot) at each DC base 1000. Here, the computational resources are the server 3000 and the storage 4000.

[0102] Specifically, in the target time slot, it is determined whether the power consumption target value 12250-renewable energy supply forecast 12130 is equal to or greater than 0. If the power consumption target value 12250-renewable energy supply forecast 12130 is equal to or greater than 0, it is determined that there is a power excess, and the inter-base relocation program 18400 proceeds to step S4140. If not, it is determined that there is no power excess, and the program proceeds to step S4160.

[0103] In step S4140, the inter-site relocation program 18400 determines the amount of power to be spatially relocated in the currently targeted time slot so that the total sum of the power excesses in all DC sites 1000 in the entire data center system S is smaller. Preferably, the total sum is minimized. The inter-site relocation program 18400 determines the amount of power to be migrated from the currently targeted DC site 1000 to another DC site 1000 based on the renewable energy supply amount and power consumption data of each DC site 1000 collected in step S2000 in the workload control planning process, and the spatial migration possible time zone 12345 and the spatial migration impossible time zone 12350 of the workload for spatial migration. The migration destination DC site 1000 does not need to be one, and may be multiple.

[0104] In step S4150, the inter-base relocation program 18400 decreases the DC power consumption forecast 12230 and the power consumption target value 12250 in the power demand table 12200 in the currently targeted DC base 1000 by the amount of power transferred in step S4140. Also, in the migration destination DC base 1000, the DC power consumption forecast 12230 and the power consumption target value 12250 in the power demand table 12200 are increased by the amount of the transfer.

[0105] In step S4160, the inter-base relocation program 18400 determines whether the currently targeted time slot is the last time slot. If this is the last time slot, the inter-base relocation program 18400 proceeds to step S4210, and if not, proceeds to step S4170.

[0106] In the workload control planning process, it is assumed that the user has decided in advance how many future time slots to include in the plan and this is stored in the storage device 18000 of the inter-site workload control system 10000. For example, when creating a workload control plan for the next day, the last time slot is the time slot 24 hours from now.

[0107] In step S4170, the inter-base relocation program 18400 again determines whether or not there is a power excess in the currently targeted timeslot, similar to step S4130. If it is determined that there is a power excess, the inter-base relocation program 18400 proceeds to step S4180. If not, it proceeds to step S4200.

[0108] In step S4180, the inter-base relocation program 18400 judges whether or not there is a workload that is not spatially migrateable in the currently targeted timeslot. Specifically, the inter-base relocation program 18400 judges whether or not there is a workload whose execution status 12355 in the workload table 12300 is "RUNNING" and whose time set in the spatially migrateable time zone 12350 includes the time of the currently targeted timeslot. If such a workload is present, the inter-base relocation program 18400 proceeds to step S4190, and if not, proceeds to step S4200.

[0109] In step S4200, the inter-site relocation program 18400 selects the time slot next to the currently selected time slot as the target time slot, and proceeds to step S4130.

[0110] In step S4210, the inter-site relocation program 18400 judges whether or not all DC sites 1000 have been targeted. If all DC sites 1000 have been targeted, the workload relocation planning process ends. If there is still a DC site 1000 that has not been targeted, the process proceeds to step S4220.

[0111] In step S4220, the inter-base relocation program 18400 appropriately selects one from the DC bases 1000 that have not yet been targeted, and sets it as the next target. After that, the inter-base relocation program 18400 proceeds to step S4120.

[0112] <Time transition power amount determination process> FIG. 13 is a flow diagram illustrating the time transition power amount determination process (S4190) in detail.

[0113] First, in step S4310, the inter-site relocation program 18400 targets the next time slot of the currently targeted time slot. The time slot targeted in the time shift power amount determination process (S4190) is held independently of the time slot targeted in the workload relocation plan process (S4000). For example, assume that a certain time slot is targeted in the workload relocation plan process (S4000), and the process of step S4310 is executed for the first time in the time shift power amount determination process (S4190). In this case, the inter-site relocation program 18400 targets the next time slot of the time of the time slot targeted in the workload relocation plan process (S4000).

[0114] However, in the workload relocation planning process (S4000), this does not change the time slot itself held by the inter-base relocation program 18400. Also, in the time transition power amount determination process (S4190), when step S4310 is executed for the second or subsequent time, the inter-base relocation program 18400 targets the time slot of the time next to the time targeted previously.

[0115] In the explanation of Fig. 13, the "current target time slot" refers to the time slot targeted in the time transition power amount determination process (S4190), i.e., different from the time slot targeted in the workload relocation plan process (S4000).

[0116] Next, in step S4320, the inter-base relocation program 18400 acquires a workload that will be changed from a time that is not spatially transferable to a time that is spatially transferable in the currently targeted time slot. Specifically, the inter-base relocation program 18400 refers to the execution status 12355, the spatial transferable time slot 12345, and the spatial transferable time slot 12350 in the workload table 12300. Then, it acquires a record of a workload for spatial transfer where the execution status 12355 is "RUNNING", the currently targeted time slot is included in the spatial transferable time slot 12345, and the previous time slot is included in the spatial transferable time slot 12350.

[0117] In step S4330, the inter-base relocation program 18400 judges whether or not the workload for spatial migration acquired in step S4320 can be migrated to another DC base 1000. The inter-base relocation program 18400 judges whether or not the workload for spatial migration acquired in step S4320 has a DC base 1000 as a migration destination from the renewable energy supply amount, power consumption amount, target power consumption amount, etc. of the DC base 1000 stored in each base data 18100. If there is a migration destination for even one workload, proceed to step S4340, and if not, proceed to step S4360.

[0118] In step S4340, the inter-site relocation program 18400 performs time shifting from the time slot targeted by the inter-site relocation program 18400 in the workload relocation planning process (S4000) to that time slot. The amount of power shifted here is an amount of power with the power amount of the workload for spatial migration that is spatially migrated to another DC site 1000 in step S4330 as the upper limit. Then, the power consumption target value 12250 of the power demand table 12200 in the time slot targeted by the inter-site relocation program 18400 in the workload relocation planning process (S4000) is subtracted by the amount of power.

[0119] That is, if the power excess (excess power amount) still found to exist in step S4170 is equal to or less than the second power adjustment amount to be moved by the spatial movement of the workload acquired in step S4320, a first power adjustment amount to be delayed is determined in step S4340. By delaying the first power adjustment amount, a part or all of the power excess in the relevant time period is moved to the next time period, with the power excess as an upper limit.

[0120] If the power excess (excess power amount) still found to exist in step S4170 exceeds the second power adjustment amount moved by the spatial movement of the workload acquired in step S4320, a first power adjustment amount to be delayed is determined in step S4340. By delaying the first power adjustment amount, a part of the power excess in the corresponding time period is moved to the next time period with the second power adjustment amount as an upper limit.

[0121] In the process of step S4340, in the time slot targeted in the workload reallocation plan process (S4000), the workload for time migration is migrated from that time to the time slot currently targeted in the time migration power amount determination process (S4180). In addition, in the time slot currently targeted in the time migration power amount determination process (S4180), the workload for space migration having the same power amount as the time-migrated power amount is migrated to another DC site 1000. Therefore, in the time slot targeted in the workload reallocation plan process (S4000), the power amount decreases, and in the time slot currently targeted in the time migration power amount determination process (S4180), there is no increase or decrease in the power amount. This is reflected in the power consumption target value 12250 of the power demand table 12200.

[0122] In step S4350, the inter-site relocation program 18400 judges whether or not there is a power excess in the time slot targeted in the workload relocation planning process (S4000). This judgment is performed in the same manner as in step S4130 in the workload relocation planning process. If there is a power excess, the inter-site relocation program 18400 proceeds to step S4360, and if not, ends the process.

[0123] In step S4360, the inter-site relocation program 18400 judges whether the currently targeted time slot is the last time slot. This judgment is made in the same manner as in step S4190 in the workload relocation planning process. If the currently targeted time slot is the last time slot, the inter-site relocation program 18400 ends the time shift power amount determination process because time control is not possible after this time slot. On the other hand, if the currently targeted time slot is not the last time slot, the inter-site relocation program 18400 proceeds to step S4310.

[0124] <Workload input screen> 14 is a diagram showing an example of a workload input screen 24000. The workload input screen 24000 is a screen used when a user inputs a new workload. The workload input screen 24000 includes an available DC site input form 24100 for specifying a DC site 1000 to which the workload to be input can be allocated.

[0125] The workload input screen 24000 also includes a delayable time input form 24200 for the user to input a delayable time when the workload to be input is a time-shift workload. The workload input screen 24000 also includes a space-shiftable time input form 24300 for the user to input a time period when space-shift is possible when the workload is for space-shift. The workload input screen 24000 also includes a space-transfer-prohibited time input form 24400 for the user to input a time period when space-transfer is not possible when the workload is for space-transfer. The workload input screen 24000 also includes a source code input form 24500 for specifying the source code of the workload to be executed at the DC base 1000, and a decision button 24600 for performing workload input.

[0126] The user inputs at least one DC site in the available DC site input form 24100. The content input here is input to the executable site 12340 of the workload table 12300 in the DC site 1000 to which this workload is allocated.

[0127] In the delayable time input form 24200, if the workload to be submitted is a workload for time shifting, the user inputs the delayable time. If the execution of the workload cannot be delayed or it is not a workload for time shifting, 0 or n / a is input, or the field is left blank. The content input here is input to the delayable time 12335 in the workload table 12300 in the DC base 1000 where this workload is placed.

[0128] In the spatial migration available time input form 24300, if the workload to be input is a workload for spatial migration, the user inputs the time when spatial migration is possible. If the workload is for spatial migration but cannot be migrated at any time, or if the workload is not for spatial migration, leave this field blank or "n / a." The content entered here is entered into the spatial migration available time period 12345 of the workload table 12300 in the DC base 1000 where this workload is placed.

[0129] In the spatial migration possible time input form 24300, if the workload to be input is a workload for spatial migration, the user inputs the time when spatial migration is not possible. If the workload is for spatial migration but can be migrated at any time, or if the workload is not for spatial migration, leave this field blank or "n / a." The content entered here is entered into the spatial migration impossible time period 12350 in the workload table 12300 in the DC base 1000 where this workload is placed.

[0130] In the source code input form 24500, the user specifies the path of the source code of the workload to be executed at the DC base 1000. This may be done by specifying the URL of the source code uploaded to the cloud, or by specifying the path of the source code stored in a local storage area. Also, instead of specifying a path or URL, the source code may be uploaded directly.

[0131] When input is complete, the user presses the decision button 24600 to start the deployment of the workload.

[0132] (Effects of the embodiment) In the above embodiment, the time when workloads can be migrated between bases, the time when workloads cannot be migrated, and the time when execution time can be delayed within the same base are managed, and a combination of spatial migration for relocating workloads between bases and time migration for relocating workloads within a base is performed. In this way, by performing spatial migration of workloads between bases and time migration within a base while taking into account the time when workloads cannot be migrated, it is possible to efficiently and more precisely adjust the power demand of each base so that the renewable energy utilization rate in the entire data center system is improved. Furthermore, it is possible to achieve a reduction in the power consumption of the entire data center system and contribute to environmental protection. In addition, by adjusting the power demand in the entire data center system, it is possible to utilize it in an initiative for local production and consumption of renewable energy, which efficiently utilizes renewable energy generated in the vicinity.

[0133] In the above embodiment, after determining the amount of power of the excess power to be transferred in space, the amount of power to be transferred in time is determined so that the excess power that cannot be adjusted by the spatial transfer is adjusted by the time transfer. This makes it possible to more precisely adjust the power demand of each base.

[0134] In the above embodiment, only when there is a workload that can be spatially migrated in the next time slot, the amount of excess power in the target time slot is shifted in time, with the power movement due to the spatial migration of this workload as the upper limit. This makes it possible to prevent the excess power from expanding in the destination time slot due to the time migration, thereby preventing the adjustment of power demand from becoming worse.

[0135] In the above embodiment, when the excess power to be time-shifted is equal to or less than the power adjustment amount due to the spatial shift of the workload in the next time slot, a part or all of the excess power is time-shifted. In the above embodiment, when the excess power to be time-shifted exceeds the power adjustment amount due to the spatial shift of the workload in the next time slot, a part of the excess power is time-shifted with this power adjustment amount as the upper limit. This makes it possible to prevent the power excess from expanding in the time slot to which the power shift is to be shifted, resulting in a deterioration in the adjustment of power demand.

[0136] In the above embodiment, the amount of power to be transferred in space is first determined, and then the workload is selected and transferred in space based on the amount of power at the time of actual transfer. In the above embodiment, the amount of power to be transferred in time is first determined, and then the workload is selected and transferred in time based on the amount of power at the time of actual transfer. Therefore, by creating an overall transfer plan based on the amount of power and selecting individual workloads at the time of actual transfer, it is possible to flexibly and steadily adjust the power demand.

[0137] The present invention is not limited to the above-described embodiment, and can be implemented using any components without departing from the scope of the present invention. The above-described embodiments and modifications are merely examples, and the present invention is not limited to these contents as long as the features of the invention are not impaired. In addition, although various embodiments and modifications have been described above, the present invention is not limited to these contents. Other aspects conceivable within the scope of the technical idea of ​​the present invention are also included in the scope of the present invention.

[0138] For example, some of the functions provided in each device of the present embodiment may be provided in another device, and functions provided in another device may be provided in the same device.

[0139] Furthermore, the program configuration described in the present embodiment is merely an example. For example, a part of a program may be incorporated into another program, or multiple programs may be configured as a single program.

[0140] Furthermore, in this embodiment, the executable time is expressed as the delayable time, but the executable time may be specifically designated.

[0141] Furthermore, in this embodiment, the workload is a workload in a data center, but the present invention is also applicable to information processing executed in other facilities or networks.

[0142] In addition, in this embodiment, the purpose of controlling the workload is mainly to improve the renewable energy rate, but as described in the background art, this technology can also be used as a means of providing adjustment power. [Explanation of symbols]

[0143] S...data center system, 1000...DC site, 2000...site management computer, 7000...renewable energy power source, 10000...inter-site workload control system, 12100...renewable energy power table, 12200...power demand table, 12300...workload table, 12400...spatial power transfer table.

Claims

1. A data center system including a plurality of bases each having a computing resource and a renewable energy power source for executing a workload, and an inter-base workload control system for controlling migration of the workload between the plurality of bases and within the same base, The processor of the inter-site workload control system includes: A positive excess power amount obtained by subtracting a predicted value of the power supply amount of the renewable energy power source from a predicted value of the power consumption amount of the computing resource associated with the execution of the workload in a future time period at each of the bases; and spatial migration available time zone information regarding time zones during which spatial migration is possible and time zones during which spatial migration is not possible for migrating the workload scheduled to be executed in the future time zone at each of the bases to another base; time-shiftable time information regarding a time shift in which the workload scheduled to be executed in the future time period at each of the bases is shifted to another time period by delaying the execution within the same base; and a predicted power consumption value which is a predicted value of the power consumption of the computing resource associated with the execution of the workload scheduled to be executed in the future time period at each of the bases; Manage and determining a power adjustment amount indicating the predicted power consumption value to be moved by the spatial transition and the time transition of each of the workloads scheduled to be executed in the future time slots at each of the bases, based on the spatial transition available time period information, the temporal transition available time information, and the predicted power consumption value, so that a total of the excess power amounts in the future time slots at the multiple bases becomes smaller; Performing the spatial and temporal migration of the workload based on the power adjustment amount A data center system comprising:

2. The data center system according to claim 1 , The processor, determining the amount of power adjustment to be moved by the spatial migration of the workload to be executed in a specific time period at a specific base where the excess power amount exists in the future time period at each of the bases, so that the total sum becomes smaller; If there is still an excess amount of power in the specific time period at the specific base after the power adjustment amount is determined, a first power adjustment amount is determined, the first power adjustment amount indicating the predicted power consumption value that is to be shifted due to the time shift of a first workload scheduled to be executed in the specific time period at the specific base, so that the total sum becomes smaller. A data center system comprising:

3. The data center system according to claim 2, The processor, determining whether there is a second workload scheduled to be executed in the time period next to the specific time period; When the second workload capable of the spatial transition is present in the time period next to the specific time period, a second power adjustment amount indicating the predicted power consumption value to be shifted by the spatial transition of the second workload is determined as the first power adjustment amount. A data center system comprising:

4. The data center system according to claim 3, The processor, When the amount of excess power in the specific time period is equal to or less than the second power adjustment amount, the first power adjustment amount is determined so that the amount of excess power in the specific time period is set as an upper limit and a part or all of the amount of excess power in the specific time period is moved to the next time period. A data center system comprising:

5. The data center system according to claim 3, The processor, When the excess power amount in the specific time period exceeds the second power adjustment amount, the first power adjustment amount is determined so that a part of the excess power amount in the specific time period is moved to the next time period with the second power adjustment amount as an upper limit. A data center system comprising:

6. The data center system according to claim 2, determining the workload to be subject to the spatial transition during the future time period during which the spatial transition is possible based on the power adjustment amount; Performing the spatial transition on the determined workload. A data center system comprising:

7. The data center system according to claim 2, determining, based on the first power adjustment amount, the workload to be subject to the time shift during the future time period during which the spatial shift is not possible; performing the time migration on the determined workload; A data center system comprising:

8. An inter-site workload control method executed by an inter-site workload control system that controls migration of a workload between a plurality of sites having computing resources for executing the workload and a renewable energy power source and within the same site, comprising: A processor of the inter-site workload control system, A positive excess power amount obtained by subtracting a predicted value of the power supply amount of the renewable energy power source from a predicted value of the power consumption amount of the computing resource associated with the execution of the workload in a future time period at each of the bases; and spatial migration available time zone information regarding time zones during which spatial migration is possible and time zones during which spatial migration is not possible for migrating the workload scheduled to be executed in the future time zone at each of the bases to another base; time-shiftable time information regarding a time shift in which the workload scheduled to be executed in the future time period at each of the bases is shifted to another time period by delaying the execution within the same base; and a predicted power consumption value which is a predicted value of the power consumption of the computing resource associated with the execution of the workload scheduled to be executed in the future time period at each of the bases; Manage and determining a power adjustment amount indicating the predicted power consumption value to be moved by the spatial transition and the temporal transition of each of the workloads scheduled to be executed in the future time slots at each of the bases, based on the spatial transition available time period information, the temporal transition available time information, and the predicted power consumption value, so that a total of the excess power amounts in the future time slots at the multiple bases becomes smaller; Performing the spatial and temporal migration of the workload based on the power adjustment amount A method for controlling workload between bases, comprising the steps of:

9. An inter-site workload control system for controlling migration of a workload between a plurality of sites having computing resources for executing the workload and a renewable energy power source and within the same site, comprising: The processor of the inter-site workload control system includes: A positive excess power amount obtained by subtracting a predicted value of the power supply amount of the renewable energy power source from a predicted value of the power consumption amount of the computing resource associated with the execution of the workload in a future time period at each of the bases; and spatial migration available time zone information regarding time zones during which spatial migration is possible and time zones during which spatial migration is not possible for migrating the workload scheduled to be executed in the future time zone at each of the bases to another base; time-shiftable time information regarding a time shift in which the workload scheduled to be executed in the future time period at each of the bases is shifted to another time period by delaying the execution within the same base; and a predicted power consumption value which is a predicted value of the power consumption of the computing resource associated with the execution of the workload scheduled to be executed in the future time period at each of the bases; Manage and determining a power adjustment amount indicating the predicted power consumption value to be moved by the spatial transition and the time transition of each of the workloads scheduled to be executed in the future time slots at each of the bases, based on the spatial transition available time period information, the temporal transition available time information, and the predicted power consumption value, so that a total of the excess power amounts in the future time slots at the multiple bases becomes smaller; Performing the spatial and temporal migration of the workload based on the power adjustment amount The inter-site workload control system is characterized by:

Citation Information

Patent Citations

  • Data center system and operation method thereof, as well as apparatus and method for supporting data center system hub design

    JP2021189845A