System, method executed by the system, program

Abstract

Simple and detailed control is provided in both aspects of controlling the facilities that provide information processing resources and controlling the allocation of workloads to information processing resources. SOLUTION: The system 101 has a data center operation mode plan creation unit 1800 that creates a data center operation mode plan 110, which is a plan for each operation mode 103 of a data center 102, based on a prediction 109 regarding the workload. The system 101 may also have a workload deployment setting unit 1900 that determines a data center 102 to which a workload associated with a received workload execution request 211 is to be deployed, based on conditions under which the workload associated with a received workload execution request 211 is processed and the operation mode 103 of the data center 102 determined by the data center operation mode plan 110.

Description

[Technical Field]

[0001] The present disclosure relates to a technique for efficiently operating information processing resources that process workloads. [Background technology]

[0002] One or more workloads (tasks to be processed, etc.) are deployed to some information processing resource, and then the information processing resource processes the workload. For example, information processing resources (e.g., resources such as a central processing unit (CPU), a graphics processing unit (GPU), and memory) owned by a group of servers on the cloud are assigned to a training processing workload such as training a model (model parameters) by machine learning, or an inference processing workload such as performing inference using a trained model, and once the workload is deployed, the information processing resource processes the workload.

[0003] Conventionally, various methods for improving efficiency have been studied in relation to the processing of workloads by information processing resources. For example, Patent Document 1 discloses a technology that, in response to a determined workload allocation, simulates the temperature and other factors in a data center that has information processing resources for processing the workload, and attempts to minimize the energy consumption of cooling equipment while keeping the temperature within an acceptable range. For example, Patent Document 2 discloses a technology for allocating each workload to each host, taking into consideration the specifications of the information processing resources required to execute the workload (e.g., the specifications of the required hardware accelerator) and the maintenance schedules of each host that is a candidate for allocating the workload. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] U.S. Patent No. 10,034,417 [Patent Document 2] US Patent Application Publication No. 2023 / 0035310 Summary of the Invention [Problem to be solved by the invention]

[0005] There is much room for improvement in the efficiency with which information processing resources handle workloads. The technology disclosed in the aforementioned Patent Document 1 attempts to collectively control all information processing resources that process workloads, which makes it difficult to set up control of the equipment that provides the information processing resources and makes it difficult to perform detailed control. Furthermore, the technology disclosed in Patent Document 1 places emphasis on controlling the equipment that provides the information processing resources that process the workload, and not so much on controlling the workload itself. Therefore, if an attempt is made to control the workload using the technology disclosed in Patent Document 1, it is expected that the workload control will be cumbersome and that it will be difficult to perform detailed control. The technology disclosed in the aforementioned Patent Document 2 attempts to allocate workloads to hosts in a situation where control of information processing resources is largely fixed, although there is some room for adjustment of host maintenance schedules. Therefore, it can be said that the technology disclosed in Patent Document 2 does not place much emphasis on control of the equipment that provides the information processing resources. In other words, if you try to configure the equipment that provides the information processing resources using the technology disclosed in Patent Document 2, the configuration is likely to be cumbersome and it is expected that it will be difficult to perform detailed control. Furthermore, in the technology disclosed in Patent Document 2, when allocating a workload to a host, the specifications of the information processing resources for executing the workload are taken into consideration, but the specifications of the computer resources provided by each host to which the workload is allocated when the host is in operation are specific to that host, and it is not considered that the specifications of the computer resources provided by the host will be dynamically adjusted. As such, the control of allocating workloads to hosts in the technology disclosed in Patent Document 2 lacks flexibility.

[0006] From the above, one of the objectives of the present disclosure can be to provide simple and detailed control when information processing resources process workloads, both in terms of controlling the equipment that provides the information processing resources and in terms of controlling when allocating (deploying) workloads to the information processing resources.

[0007] When the object of the present disclosure is achieved, it becomes possible to simply and precisely control the facilities that provide the information processing resources that process the workloads according to the status of the workloads to be processed, which means that the facilities that provide the information processing resources can be operated with the necessary and sufficient quality or quantity. Furthermore, when the object of the present disclosure is achieved, when allocating information processing resources to workloads, it is possible to simply and precisely control which of the information processing resources that can be provided is precisely controlled in terms of the quality or quantity to which the workload is allocated, thereby enabling highly flexible operation in allocating information processing resources to workloads. [Means for solving the problem]

[0008] In order to achieve at least one of the above objects, the present disclosure may have the following features, for example. One aspect of the present disclosure is a system including a data center operation mode plan creation unit that creates a data center operation mode plan, which is a plan for each operation mode of a data center, based on a prediction regarding a workload requested to be processed in any of the data centers. [Effects of the Invention]

[0009] As described above, the present disclosure determines a plan for the operation mode of each data center that is a candidate for processing a workload based on a workload prediction. In this way, the present disclosure determines an operation mode for each data center, enabling simple and detailed control of the equipment that provides the information processing resources that process the workload. Furthermore, the present disclosure determines an operation mode for each data center based on a workload prediction, enabling the equipment that provides the information processing resources to be operated with the necessary and sufficient quality or quantity. Furthermore, since the present disclosure defines an operation mode for each data center, when allocating information processing resources to a workload, the operation mode allows for simple and detailed control over which information processing resources of each data center to allocate the workload to, where the quality or quantity of resources that can be provided is precisely controlled.

[0010] As a result of the above, the present disclosure enables simple and detailed control when information processing resources process workloads, both in terms of controlling the equipment that provides the information processing resources and in terms of control when allocating (deploying) workloads to the information processing resources.

[0011] Methods and programs that achieve the same processing as the above system can also achieve the same effects as the above system. In the form of a program, costs can often be reduced. Programs also make it easier to make design changes to the processing. Other features that the present disclosure may have and the effects corresponding to those features will be disclosed in this specification, claims, or drawings. [Brief explanation of the drawings]

[0012] [Figure 1] 1 illustrates a basic functional configuration of an embodiment of the present disclosure. [Figure 2] 1 shows an overall configuration including a system 101 according to an embodiment of the present disclosure. [Figure 3] The configuration of the data center area is shown. [Figure 4] 3 shows the functional configuration of a data center control system 301. [Figure 5] 1 shows the functional configuration of a system 101. [Figure 6] 1 shows a data center list table. [Figure 7] 10 shows a data center operation mode list table. [Figure 8] 1 shows a workload performance table. [Figure 9] 1 shows a workload forecast table. [Figure 10] 1 shows a data center operation mode plan table. [Figure 11] 1 shows a workload execution request buffer table. [Figure 12] 1 shows a workload deployment setting table. [Figure 13] 1 shows a workload execution history table. [Figure 14] 1 illustrates a workload redistribution setting table. [Figure 15] 1 shows a computer architecture for implementing the system 101. [Figure 16] 10 shows the processing of a workload result table creation unit. [Figure 17] 10 shows the processing of the workload prediction unit. [Figure 18] 10 shows the processing of a data center operation mode plan creation unit. [Figure 19] 10 shows the processing of a workload deployment setting unit. [Figure 20] 10 shows the processing of a workload reallocation setting unit. [Figure 21]10 shows a modified data center list table. DETAILED DESCRIPTION OF THE INVENTION

[0013] Embodiments of the present disclosure will be described in detail below with reference to the drawings. Note that the embodiments described below do not limit the disclosure according to the claims, and not all of the elements and combinations thereof described in the embodiments are necessarily essential to the solutions of the present disclosure. The following description and drawings are examples for explaining the present disclosure, and appropriate omissions and simplifications have been made for clarity of explanation. The present disclosure can be implemented in various other forms. Unless otherwise specified, each component may be singular or plural. The position, size, shape, range, etc. of each component shown in the drawings may not represent the actual position, size, shape, range, etc., in order to facilitate understanding of the invention. Therefore, the present disclosure is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings. Each of the systems, devices, or functional units disclosed herein may be integrated into a single piece of hardware, or may be divided into multiple parts that work together to perform their functions. Several systems, devices, or functional units may be integrated into one hardware configuration. Each of the systems, devices, or functional units may be realized by causing a computer to execute software (programs) (as in FIG. 15). Some of the functions of the system, device, or functional unit may be realized by hardware (e.g., hardwired logic or a field programmable gate array (FPGA)), and the remaining functions may be realized by executing software (programs). All of the functions of each of the systems, devices, or functional units may be realized by hardware. Some or all of the steps shown in the flowcharts, etc. described in this disclosure may be realized by hardware. One or more systems, devices, or functional units of the present disclosure may be realized using one or more hardware resources. For this purpose, each of the systems, devices, or functional units of the present disclosure may be virtually realized. For example, a virtual computer or virtual container technique may be used. The term "program" may be included in the general concept of software, in which software and hardware resources cooperate to construct a specific system or method of operation according to the intended purpose. In other words, the term "program" is not limited to a specific type or form of program. The program may also be initially recorded in a compressed format. The same reference numbers are used in multiple drawings. In the drawings showing flowcharts, rectangular boxes represent processing steps, and hexagonal boxes represent conditional branching steps. In the drawings showing flowcharts, "step" is abbreviated as "S."

[0014] 1. Basic functional configuration (Figure 1) FIG. 1 shows a basic functional configuration 100 (and the information handled) of a system 101 according to an embodiment of the present disclosure. Note that not all of the functional configurations shown in FIG. 1 are essential. Furthermore, the existence of functional configurations other than those shown in FIG. 1 is not precluded. In FIG. 1 (and FIGS. 4 and 5), solid rectangles with the word "unit" in their names indicate functional units.

[0015] In addition to a system 101 according to an embodiment of the present disclosure, FIG. 1 also illustrates a plurality of data centers 102, a plurality of result consumption devices 104, and a plurality of execution request devices 105. (There may be only one execution request device 105.) The execution request device 105 is a device that issues a workload execution request 211 to request workload processing. The result consumption device 104 is a device that consumes the results of workload processing. Each of the execution request device 105 and the result consumption device 104 may be a user device, another type of device, or some kind of functional unit implemented by executing a program or the like in some kind of system device or the like. Also, although the result consumption device 104 and the execution request device 105 are shown as separate devices in FIGS. 1 and 2, the result consumption device 104 and the execution request device 105 may be the same device. The data center 102 refers to a set of equipment having information processing resources (e.g., a central processing unit (CPU), a graphics processing unit (GPU), and memory) for processing workloads. The data center 102 here may be what is called a container-type data center. The container-type data center may include multiple servers having information processing resources inside a container-like storage space. Furthermore, the data center 102 may be capable of performing equipment control on a container-type data center basis. Here, the equipment control may include one or more of air conditioning control, storage battery control, and emergency generator control. Details of the container-type data center will be described later with reference to FIGS. 3 and 4. Note that in the present disclosure, it is sufficient if the equipment having the information processing resources can be controlled for each certain amount of information processing resources. Therefore, the equipment having the information processing resources is not necessarily limited to a container-type data center, but may be another data center or some other facility other than a data center. 1, one of the execution request devices 105 issues a workload execution request 211 to request the processing of a workload. The workload associated with the issued workload execution request 211 is assigned to one of the data centers 102, and the workload is deployed to the assigned data center 102. The data center 102 to which the workload is deployed executes the processing of the workload.

[0016] In the series of workload handling operations described above, the system 101, which is an embodiment of the present disclosure, assumes at least the role played by a functional unit called a data center operation mode plan creation unit 1800 in Fig. 1. Specifically, the data center operation mode plan creation unit 1800 in the system 101 creates a data center operation mode plan 110, which is a plan for each operation mode 103 of the data center 102, based on a prediction 109 regarding a workload requested to be processed in one of the data centers 102. The operation mode 103 set for each data center 102 may be used to determine the control details of the equipment of the data center 102. The control details of the equipment of the data center 102 may affect the details (e.g., type, performance, and quantity) and availability of information processing resources that the data center 102 can provide. The control details of the equipment of the data center 102 may also affect various indicators related to the data center 102 (e.g., running costs (operating costs), the proportion of electricity generated by power generation with relatively low carbon emissions in the total amount of electricity consumed (green rate), and an index related to adjusted compensation).

[0017] In other words, if the operation mode 103 is appropriately set for each of the data centers 102, it is possible to achieve both the ability to process the workload group indicated by the workload prediction 109 in the data centers without excess or deficiency, and the ability to improve various indicators related to the data centers 102. Also, since the operation mode 103 is set for each data center 102, simple control can be achieved in terms of controlling the equipment that provides information processing resources. Furthermore, since it is possible to set different operation modes 103 for each data center 102, detailed control can be achieved in terms of controlling the equipment that provides information processing resources.

[0018] 1, the system 101 may also centrally control the allocation (deployment) of workloads to information processing resources. To this end, the system 101 may include a workload deployment setting unit 1900. When a workload execution request 211 is received from one of the execution request devices 105, the workload deployment setting unit 1900 may determine a data center 102 to which a workload associated with the received workload execution request 211 is to be allocated (deployed). As a result of this determination, the workload deployment setting unit 1900 may create a workload to data center deployment setting 112. Based on this workload to data center deployment setting 112, the workload may be deployed to the data center 102.

[0019] Here, the workload deployment setting unit 1900 may determine the data center 102 to which the workload will be assigned (deployed) based on the conditions under which the workload associated with the workload execution request 211 is processed and the operation mode 103 of each data center 102 defined by the data center operation mode plan 110. The conditions imposed when a workload is processed may include one or more of the following: constraints when processing the workload (e.g., a constraint indicating a data center 102 to which the workload can be deployed if other conditions are met), requirements when processing the workload (e.g., a requirement regarding the degree to which processing is not interrupted while processing the workload (degree of availability) and requirements regarding the content of information processing resources (e.g., type, performance, and amount)), the amount of time required to process the workload, and an allowable delay time for starting or ending processing of the workload. Note that other conditions may also be imposed. Each operational mode 103 of the data center 102 may be associated with, for example, requirements that can be provided when processing a workload (for example, requirements that can be provided regarding the degree to which execution is not interrupted midway when processing a workload (degree of availability) and requirements that can be provided regarding the content of information processing resources (for example, type, performance, quantity)). Therefore, the workload deployment setting unit 1900 can appropriately allocate (deploy) the workload associated with the workload execution request 211 to a data center 102 that can process the workload.

[0020] Unlike Fig. 1, the system 101 does not have to centrally control the allocation (deployment) of workloads to information processing resources. It is also possible to achieve distributed control when allocating (deploying) workloads to information processing resources. Details will be explained later as Modification A. In this case, the system 101 does not need to include the workload deployment setting unit 1900.

[0021] For simplicity of explanation, the following mainly describes the manner in which the system 101 centralizes control over the allocation (deployment) of workloads to information processing resources.

[0022] The system 101 according to the embodiment of the present disclosure has the above-described functional configuration, and therefore can have the effects described in the above-described [Effects of the Invention].

[0023] 2. Overall configuration including system 101 (Fig. 2) Fig. 2 shows an overall configuration 200 including a system 101 according to an embodiment of the present disclosure. Note that not all of the functional configurations shown in Fig. 2 are essential. Furthermore, the presence of functional configurations other than those shown in Fig. 2 is not prohibited.

[0024] The multiple (container-type) data centers 102, the multiple result consumption devices 104, the one or more execution request devices 105, and the system 101 of the embodiment of the present disclosure in FIG. 1 may be capable of communicating with each other via a network 299 as shown in FIG. 2. Note that although five networks 299 are shown in FIG. 2, this is for convenience of illustration. The actual topology of the network 299 may be arbitrary. Furthermore, it is sufficient that communication entities that actually communicate with each other (e.g., the data centers 102, the result consumption devices 104, the execution request devices 105, and the systems 101) are capable of communicating with each other, and it is not necessary that all combinations of the data centers 102, the result consumption devices 104, the execution request devices 105, and the systems 101 shown in FIG. 2 be capable of communicating with each other.

[0025] Each of the (container-type) data centers 102 may be located at some base 202. The example of FIG. 2 shows a configuration in which multiple (container-type) data centers 102 are geographically distributed and hierarchically arranged according to their roles. Specifically, as shown in FIG. 2, there may be a core base 202-C (core base) that is a central base for services provided by the multiple (container-type) data centers 102, a regional base 202-R (regional base) that is a base for each region, and an edge base 202-E (regional base) that is located in an area smaller than a region. One or multiple (container-type) data centers 102 may be located in each of these bases 202. Here, the geographically closest site 202 from the perspective of the result consumption device 104-U may be the edge site 202-R (regional site) in the region where the result consumption device 104-U is located. In other words, if the result consumption device 104-U wants to obtain the results of processing the workload associated with the workload execution request 211 as soon as possible, it is often appropriate to allocate (deploy) the workload to the (container-type) data center 102-E located in the edge site 202-E (regional site) in the region where the result consumption device 104-U is located (assuming all (container-type) data centers 102 require similar processing time). On the other hand, the requirements that can be provided by the (container-type) data center 102 located in the site 202 (for example, requirements that can be provided regarding the degree to which processing is not interrupted midway when processing a workload (degree of availability) and requirements that can be provided regarding the content of information processing resources (for example, type, performance, and quantity)) are often better at the regional site 202-R (regional site) than at the edge site 202-E (area site), and are often better at the core site 202-C (core site) than at the regional site 202-R (regional site). Therefore, depending on the content of the conditions when the workload associated with the workload execution request 211 is processed, the (container-type) data center 102-E located in the edge site 202-E (area site) (and the operation mode 103 set for the data center 102-E) may not satisfy the conditions. In the case of even stricter conditions, even the (container-type) data center 102-R (and the operation mode 103 set for the data center 102-R) located in the regional base 202-R (regional base) may not satisfy the conditions.

[0026] The system 101 according to an embodiment of the present disclosure may be located at a control center 201 (Energy Management System (EMS)), or may be located at one of the other centers shown in FIG. When the system 101 creates the data center operation mode plan 110 (FIG. 1), the system 101 transmits data center operation mode control information 210 (DC-mode) to each data center 102 in order to communicate the setting of the operation mode 103 for each data center 102 included in the data center operation mode plan 110. The data center operation mode control information 210 (DC-mode) includes the setting content of the operation mode 103 for the data center 102 that is the destination of the data center operation mode control information 210 (DC-mode). When the system 101 centrally controls the allocation (deployment) of workloads to information processing resources, the system 101 transmits workload deployment control information 212 (WL-deploy). The workload deployment control information 212 (WL-deploy) includes information about the workload to be allocated (deployed) to the data center 102 that is the destination of the workload deployment control information 212 (WL-deploy). Each of the data centers 102 may transmit workload execution history information 213 (WL-log), which is information on the execution history or execution state related to the processing of the workload, as the data center 102 processes the workload. The system 101 may grasp the information on the execution history or execution state related to the processing of the workload in each of the data centers 102 by receiving the workload execution history information 213. Note that the workload execution history information 213 (WL-log) may be transmitted and received whether the control for allocating (deploying) the workload to the information processing resources is concentrated in the system 101 or distributed to each of the data centers 102.

[0027] Control of the operation mode 103 in the data centers 102 based on the data center operation mode plan 110 (FIG. 1) may result in a change in the operation mode 103 in one of the data centers 102. Then, with the change in the operation mode 103 of the data center 102, the conditions for processing the workload that was previously deployed in the data center 102 may no longer match the requirements that can be provided by the data center 102 in the changed operation mode 103. In the above case, the system 101 may transmit workload redeployment control information 214 (WL-migration) to redeploy (rebalance, migrate) the incompatible workload to another data center 102. The workload redeployment control information 214 (WL-migration) includes information for changing the data center 102 to which the workload to be redeployed (rebalance, migrate) is to be deployed. The workload redeployment control information 214 (WL-migration) may be received by both the data center 102 that was the deployment destination before the redeployment and the data center 102 that is the redeployment destination.

[0028] 3. Data center periphery configuration (Fig. 3, Fig. 4) Fig. 3 shows a configuration 300 of a (container-type) data center 102 and its surroundings that exist in each of the bases 202 shown in Fig. 2. Note that not all of the configurations shown in Fig. 3 are required. Also, it is not prohibited that configurations other than those shown in Fig. 3 exist. 3 does not show an uninterruptible power supply (UPS) that acts as a buffer for the amount of power between various power sources and the (container-type) data center 102. Also, FIG. 3 does not show a detailed configuration for making the power source for the (container-type) data center 102 redundant or triplexed.

[0029] The (container-type) data center 102 serves as a server room and may include multiple servers 302 therein. Each of the servers 302 may include intra-server IT resources 320 as information processing resources. The intra-server IT resources 320 may include, for example, a central processing unit (CPU) 321, a graphics processing unit (GPU) 325, and memory 322. The intra-server IT resources 320 may also include other types of information processing resources. Each of the servers 302 included in one (container-type) data center 102 may be individually powered on / off controlled. Alternatively, the percentage of the servers 302 to be powered on (the percentage of the servers 302 to be powered off) among the servers 302 included in one (container-type) data center 102 may be controlled. Alternatively, the operating voltage or operating frequency of each of the servers 302 may be controlled. The (container-type) data center 102 may include a data center control system 301. Functional units included in the data center control system 301 are shown in FIG. 4 (details will be described later). The data center control system 301 may perform on / off control of the servers 302, control related to air conditioning for the (container-type) data center 102, control related to storage batteries for the (container-type) data center 102, and control related to emergency generators for the (container-type) data center 102. The data center control system 301 can receive data center operation mode control information 210 (DC-mode), workload deployment control information 212 (WL-deploy), and workload redeployment control information 214 (WL-migration), which are described in FIG. 2, and can also transmit workload execution history information 213 (WL-log). In addition, in Figure 3, the server 302 and the data center control system 301 are shown as separate entities, but the data center control system 301 may be constructed by executing a data center control program on one of the servers 302.

[0030] Peripheral equipment of the (container-type) data center 102 within the base 202 may include one or more of a data center unit air conditioning unit 330, a data center unit storage battery 340, and a data center unit fuel 351 (e.g., hydrogen fuel) for an emergency generator. Furthermore, one or more of an on-site emergency power generator 350, a fuel generating device 352 (e.g., a hydrogen electrolysis device), a carbon-based fuel 353, and an on-site solar power generation device 360 ​​may be present within the site 202. Note that one or more of the emergency power generator, the fuel generating device, the carbon-based fuel, and the solar power generation device may also be present on a data center basis. In Figure 3, white arrows without letters indicate the flow of electricity, while thick black arrows indicate the flow of things other than electricity. Numbers with circles indicate that items with the same number are connected.

[0031] The data center unit air conditioner 330 adjusts the temperature and humidity of the (container-type) data center 102 using the amount of power from any of the power transmission and distribution system 370, the data center unit storage battery 340, the on-site emergency generator 350, and the data center unit storage battery 340. (Note that, although air conditioners are mentioned in the descriptions of the embodiments of the present disclosure, including FIGS. 3, 4, and 7, control of the temperature and the like of the data center 102 is not limited to air conditioners. For example, other types of cooling devices, such as water-cooled or liquid-cooled, may be used instead of or together with air conditioners. In that case, the other types of cooling devices may be included as targets of equipment control shown in FIG. 7.) The data center unit storage battery 340 stores power using the amount of power from either the power transmission and distribution system 370, the in-site emergency power generator 350, or the in-site solar power generation device 360. The data center unit storage battery 340 can also supply the stored amount of power to the (container-type) data center 102. The on-site emergency generator 350 generates power using a data center unit fuel 351 for the emergency generator (e.g., hydrogen fuel) or generates power using a carbon-based fuel 353. Alternatively, the on-site emergency generator 350 may generate power using both the fuel 351 (e.g., hydrogen fuel) and the carbon-based fuel 353. The on-site emergency generator 350 can also supply the generated power to the (container-type) data center 102. The fuel generator 352 generates fuel for the on-site emergency generator 350 using electric power from either the power transmission and distribution system 370 or the on-site solar power generator 360. The fuel generator 352 may be, for example, a hydrogen electrolysis device. The generated fuel is stored as data center unit fuel 351 (e.g., hydrogen fuel) for the emergency generator. The on-site solar power generation device 360 ​​can supply the generated power to the (container-type) data center 102.

[0032] The amount of electricity transmitted from the power transmission and distribution system 370 to the base 202 is made up of one or both of the amount of electricity from a power generation source 380 with relatively high carbon emissions and the amount of electricity from a power generation source 390 with relatively low carbon emissions.

[0033] FIG. 4 shows a functional configuration 400 of a data center control system 301. In FIG. 4, solid rectangles with the word "unit" attached to their names indicate functional units. Each of the functional units may be constructed by executing a program for that functional unit. Alternatively, any of the functional units may be realized more by hardware. Alternatively, all of the functional units may be realized more by hardware. Furthermore, for any of the functional units shown in FIG. 4, some of the functions may be realized more by hardware, and the remaining functions may be constructed by executing a program.

[0034] The data center control system 301 may have, as functional units, an operation mode setting unit 418 and a workload allocation control unit 419. The operation mode setting unit 418 may have, as internal functional units, a server on / off unit 420, an air conditioning control unit 430, a power storage control unit 440, and an emergency power generation control unit 450. The air conditioning control unit 430 may have, as internal functional units, an air conditioning on / off unit 431, a temperature control unit 432, and a humidity control unit 433. The power storage control unit 440 may have, as internal functional units, a remaining charge management unit 441. The emergency power generation control unit 450 may have, as internal control units, a fuel storage management unit 451.

[0035] The operation mode setting unit 418 sets the (container type) data center 102 and its peripheral facilities according to the operation mode 103 instructed by the data center operation mode control information 210 (DC-mode). The server on / off unit 420 controls the power on / off of each server 302 according to the operation mode 103. Alternatively, the server on / off unit 420 with expanded functionality may control the operating voltage and operating frequency of each server 302 according to the operation mode 103. The air conditioning control unit 430 controls the data center unit air conditioner 330 in accordance with the operation mode 103. The air conditioning on / off unit 431 controls the on / off of the air conditioning function of the data center unit air conditioner 330 in accordance with the operation mode 103. The temperature control unit 432 controls the temperature of the (container type) data center 102 by air conditioning by the data center unit air conditioner 330 in accordance with the operation mode 103. The humidity control unit 433 controls the humidity of the (container type) data center 102 by air conditioning by the data center unit air conditioner 330 in accordance with the operation mode 103. The power storage control unit 440 controls the data center unit storage battery 340 in accordance with the operation mode 103. The remaining charge management unit 441 controls the remaining charge of the data center unit storage battery 340 in accordance with the operation mode 103. The emergency power generation control unit 450 performs control related to the in-site emergency generator 350 according to the operation mode 103. The fuel stockpile management unit 451 controls the stockpile amount of data center unit fuel 351 for the emergency generator according to the operation mode 103.

[0036] The workload deployment control unit 419 controls the deployment of workloads to the (container type) data center 102 in accordance with instructions from the workload deployment control information 212 (WL-deploy) and the workload redeployment control information 214 (WL-migration). In addition, the workload deployment control unit 419 may transmit information on the execution history or execution status when the (container type) data center 102 processes the workload to the system 101 as workload execution history information 213 (WL-log) as appropriate.

[0037] 4. Functional configuration of system 101 (Figs. 5 to 14) Fig. 5 shows a functional configuration 400 (and information handled) of the system 101 according to an embodiment of the present disclosure. Note that not all of the functional configurations shown in Fig. 5 are essential. Furthermore, the presence of functional configurations other than those shown in Fig. 5 is not prohibited. The content of the processing performed by the system 101 shown in Fig. 5 will be described in detail in the section "6. Processing Performed by an Embodiment of the Present Disclosure" below. In this section "4. Functional Configuration of the System 101", an outline of the content of the processing performed by the system 101 and an outline of the information (tables) handled by the system 101 will be described. Note that parts that have already been described with reference to Fig. 1 may be omitted below. In Fig. 5, solid rectangles with the word "unit" in their names indicate functional units. Also, in Fig. 5, dotted rectangles indicate handled information (tables). Also, numbers with circles indicate that items with the same number are connected.

[0038] As shown in FIG. 5 , the system 101 may have, as functional units, a workload performance table creation unit 1600, a workload prediction unit 1700, a data center operation mode plan creation unit 1800, a data center operation mode control information transmission unit 510, a workload execution request reception unit 511, a workload deployment setting unit 1900, a workload deployment control information transmission unit 512, a workload execution history information reception unit 513, a workload reallocation setting unit 2000, and a workload reallocation control information transmission unit 514. In a computer architecture 1500 such as that shown in Fig. 15 described below, when a program corresponding to each functional unit is executed to realize each functional unit in software, such software-realized functional unit does not need to be constantly realized in the system 101. For example, when a function provided by a functional unit is needed, the functional unit may be realized in software in the system 101. Also, in the system 101, any of the functional units (or a part of the function of any of the functional units) may be implemented more in hardware.

[0039] As shown in FIG. 5, the system 101 may have the following tables for managing information handled by any of the functional units: a data center list table 600, a data center operation mode list table 700, a workload actual performance table 800, a workload forecast table 900, a data center operation mode plan table 1000, a workload execution request buffer table 1100, a workload deployment setting table 1200, a workload execution history table 1300, and a workload redeployment setting table 1400. Each of the tables shown above may be recorded as part of data group 1532 in non-volatile recording medium (storage device) 1503 in computer architecture 1500 in FIG. 15 (described below). Alternatively, each of the tables shown above may be stored as part of various buffers 1523 in storage device (memory) 1502 in computer architecture 1500. Alternatively, each of the tables shown above may be held in a recording medium, storage medium, device, system, server, etc. that is accessible to or communicable with computer architecture 1500. Furthermore, in FIG. 5, the information handled by any of the functional units is shown as being managed by a table, but a method other than a table may be used as a form of managing the information.

[0040] 4.1. Overview of Functional Parts Below is an overview of the processing of each of the functional units shown in Figure 5. Note that in the explanation of the overview of the processing of each of the functional units, each of the tables is also mentioned, but the overview of each table will be explained later in "4.2. Overview of tables that manage handled information."

[0041] 4.1.1. Overview of the functional departments involved in the creation of workload performance information The workload result table creating unit 1600 creates a record including information regarding the actual receipt of the workload execution request 211. More specifically, the workload result table creating unit 1600 may create a record including information regarding the actual receipt of the workload execution request 211, based on information regarding the workload execution request 211 read from the workload execution request buffer table 1100 and information regarding the execution history of the workload associated with the workload execution request 211 read from the workload execution history table 1300. The workload result table creating unit 1600 stores the created record in the workload result table 800. Details of the processing by the workload result table creating unit 1600 will be described later with reference to FIG. 16 .

[0042] 4.1.2. Overview of Functional Departments Related to the Development of Data Center Operational Mode Plans The workload prediction unit 1700 creates a workload-related prediction 109. More specifically, the workload prediction unit 1700 may create the workload-related prediction 109 based on information about the actual receipt of a workload execution request 211, which is read from a workload actual result table 800. The workload prediction unit 1700 stores information indicating the created workload-related prediction 109 in a workload prediction table 900. Details of the processing by the workload prediction unit 1700 will be described later with reference to FIG. 17 . The data center operation mode plan creation unit 1800 creates the data center operation mode plan 110. More specifically, the data center operation mode plan creation unit 1800 may create the data center operation mode plan 110 based on information on the contents (e.g., type, performance, and amount) of information processing resources possessed by each of the data centers 102, read from the data center list table 600, information on the degree (level) of availability that can be achieved for each combination of the data center 102 and the operation mode 103, read from the data center operation mode list table 700, information on the contents (e.g., type, performance, and amount) of the information processing resources possessed by the data center that can be provided to the workload, information on the values ​​of one or more types of indicators, and information indicating the workload prediction 109, read from the workload prediction table 900. The data center operation mode plan creation unit 1800 stores information indicating the created data center operation mode plan 110 in the data center operation mode plan table 1000. The details of the processing of the data center operation mode plan creating unit 1800 will be explained later with reference to FIG. The data center operation mode control information transmitter 510 may generate the data center operation mode control information 210 (DC-mode). More specifically, the data center operation mode control information transmitter 510 may generate the data center operation mode control information 210 (DC-mode) indicating each operation mode 103 of the data center 102 for a predetermined time period, based on information indicating the data center operation mode plan 110 read from the data center operation mode plan table 1000. The data center operation mode control information transmitter 510 may transmit the generated data center operation mode control information 210 (DC-mode) to the destination data center 102. Details of the processing of the data center operation mode control information transmitter 510 will be described later.

[0043] 4.1.3. Overview of the functional parts related to workload deployment configuration in data centers The workload execution request receiver 511 may receive the workload execution request 211 from the execution request device 105 that is the issuer of the workload execution request 211. Furthermore, the workload execution request receiver 511 may generate a record including a condition for when the workload associated with the received workload execution request 211 is processed. In addition, the workload execution request receiver 511 may store the generated record in the workload execution request buffer table 1100. Details of the processing of the workload execution request receiver 511 will be described later. The workload deployment setting unit 1900 determines the data center 102 in which the workload associated with the workload execution request 211 is to be deployed. More specifically, the workload deployment setting unit 1900 may determine the data center 102 to which the workload associated with the workload execution request 211 will be deployed, based on information about the contents (e.g., type, performance, quantity) of the information processing resources held by each of the data centers 102, read from the data center list table 600, information about the degree (level) of availability that can be achieved for each combination of the data center 102 and the operation mode 103, read from the data center operation mode list table 700, information about the contents (e.g., type, performance, quantity) of the information processing resources held by the data center 102 that can be provided to the workload, information about the settings of each operation mode 103 of the data center 102, read from the data center operation mode plan table 1000, information about the execution history or execution status of other workloads deployed to each of the data centers, read from the workload execution history table 1300, and records associated with the workload execution request 211, read from the workload execution request buffer table 1100. The workload deployment setting unit 1900 may store a record including information for identifying the determined data center 102 in the workload deployment setting table 1200. Details of the processing of the workload deployment setting unit 1900 will be described later with reference to FIG. The workload deployment control information transmitter 512 may generate the workload deployment control information 212 (WL-deploy). More specifically, based on a record read from the workload deployment setting table 1200 and including information identifying the data center 102 to which the workload is to be deployed, the workload deployment control information transmitter 512 may generate workload deployment control information 212 (WL-deploy) indicating that the workload indicated by the record is to be deployed to the data center 102 indicated by the record. The workload deployment control information transmitter 512 may transmit the generated workload deployment control information 212 (WL-deploy) to the destination data center 102. Furthermore, the workload deployment control information transmitter 512 may store a record indicating that the workload has been deployed to the data center 102 in the workload execution history table 1300. Details of the processing by the workload deployment control information transmitter 512 will be described later.

[0044] 4.1.4. Overview of the Functionality for Receiving Workload Execution History Information The workload execution history information receiver 513 may receive, from the data center 102, workload execution history information 213 (WL-log), which is information on the execution history or execution state of workloads deployed to the data center 102. Furthermore, the workload execution history information receiver 513 may update the contents of records in the workload execution history table 1300 based on the received workload execution history information 213 (WL-log). Details of the processing of the workload execution history information receiver 513 will be explained later.

[0045] 4.1.5. Overview of the functional parts related to workload redeployment settings to data centers When the operation mode 103 of any of the data centers 102 is changed, the workload redeployment setting unit 2000 determines, based on the data center operation mode plan 110, to redeploy (rebalance, migrate) the workload, whose changed operation mode 103 no longer complies with the conditions under which the workload is processed, to another data center 102 that complies with the conditions. More specifically, the workload reallocation setting unit 2000 reads information on the details (e.g., type, performance, and amount) of information processing resources held by each of the data centers 102 from the data center list table 600, information on the degree (level) of availability that can be achieved for each combination of the data center 102 and the operation mode 103 from the data center operation mode list table 700, information on the details (e.g., type, performance, and amount) of the information processing resources held by the data center 102 that can be provided to the workload, and information on the setting of each operation mode 103 of the data center 102 from the data center operation mode plan table 1000. The workload reallocation setting unit 2000 may determine, based on the information about the workload execution history 103 of one of the data centers 102, information about the execution history or execution status of other workloads deployed to each of the data centers 102 read from the workload execution history table 1300, and records associated with each workload execution request 211 read from the workload execution request buffer table 1100, that a workload whose changed operation mode 103 no longer conforms to a condition for processing the workload be redeployed (rebalanced or migrated) to another data center 102 that conforms to the condition. The workload reallocation setting unit 2000 may store, in the workload reallocation setting table 1400, a record including information identifying the determined redeployment destination (migration destination) data center 102. Furthermore, the workload reallocation setting unit 2000 may update the content of the record in the workload deployment setting table 1200. Details of the processing by the workload reallocation setting unit 2000 will be described later with reference to FIG. 20 . The workload redeployment control information transmitter 514 may generate the workload redeployment control information 214 (WL-migration). More specifically, based on a record read from the workload redeployment setting table 1400 and including information identifying a redeployment destination (migration destination) data center 102 to which the workload is to be redeployed (rebalanced, migrated), the workload redeployment control information transmitter 514 may generate the workload redeployment control information 214 (WL-migration) indicating that the workload indicated by the record is to be deployed to the redeployment destination (migration destination) data center 102 indicated by the record. The workload redeployment control information transmitter 514 may transmit the generated workload redeployment control information 214 (WL-migration) to the destination data center 102 (migration source) before the redeployment and the destination data center 102 (migration destination), which are destinations. Furthermore, the workload reallocation control information transmitter 514 may update the contents of the record in the workload execution history table 1300 to indicate that the data center 102 to which the workload is deployed has been changed. Details of the processing by the workload reallocation control information transmitter 514 will be described later.

[0046] 4.2. Overview of tables that manage information handled Below, an overview of each of the tables shown in FIG. 5 that manages the information handled by each of the functional units will be shown.

[0047] 4.2.1. Overview of Data Center List Table 600 (Figure 6) FIG. 6 shows a data center list table 600 . The data center list table 600 shows a list of (container-type) data centers 102 that exist in each of the bases 202. The data center list table 600 may also have information for each existing (container-type) data center 102 that shows the information processing resources (for example, the type, performance, and amount of information processing resources (IT resources)) held by the data center 102 and the details of the equipment related to the data center (for example, the type and performance of the equipment). (Hereinafter, information processing resources may be referred to as "IT resources.") In the example of Figure 6, the data center list table 600 contains, for each record, information such as identification information for the base 202, identification information for the data center 102 (data center identifier (DC-ID)), total amount of owned IT resources (number of central processing units (CPUs), number of graphics processing units (GPUs), memory capacity), maximum power source multiplicity, maximum amount of electricity from power sources with relatively low carbon emissions (green-derived electricity), and whether or not there is a data center unit storage battery 340.

[0048] For example, the first record in the data center list table 600 in the example of Figure 6 indicates that there is a data center 102 with a data center identifier (DC-ID) of "C-1" at "core site 202-C (core site C)", that the data center 102 has 15,000 central processing units (CPUs), 2,000 image processing units (CPUs), and a memory capacity of 15,000 gigabytes (GB), that the maximum power supply multiplicity of the data center 102 is 3, and that a data center unit storage battery 340 exists for the data center.

[0049] 4.2.2. Overview of Data Center Operation Mode List Table 700 (FIG. 7) FIG. 7 shows a data center operation mode list table 700. The data center operation mode list table 700 shows a list of operation modes 103 that can be set for each data center 102 whose existence is indicated in the data center list table 600. Furthermore, the data center operation mode list table 700 may include, for each combination of a data center 102 and an operation mode 103, information on the degree (level) of availability that can be achieved when processing a workload, and information on the contents (e.g., type, capacity, and amount) of information processing resources (IT resources) that can be provided for processing a workload (out of the contents (e.g., type, performance, and amount) of information processing resources (IT resources) possessed by the data center 102). Furthermore, the data center operation mode list table 700 may include, for each combination of a data center 102 and an operation mode 103, information on the contents of control over equipment related to the data center 102 in order to achieve the above-mentioned degree (level) of availability that can be achieved. In addition, the data center operation mode list table 700 may have one or more indicators that serve as a basis for determining whether or not to select the operation mode 103 for each combination of data center 102 and operation mode 103. In the example of Figure 7, the data center operation mode list table 700 has, for each record, information such as identification information of the data center 102 (data center identifier (DC-ID)), identification information of the operation mode 103 (operation mode number), control details for the IT resources held by the data center 102 (for example, control details for the ratio of servers 302 held by the data center 102 that are powered on), IT resources that can be provided for processing workloads (for example, the ratio of the performance and amount of information processing resources that can be provided for processing workloads in the operation mode 103 to the maximum performance and amount of information processing resources that can be achieved by all of the IT resources held by the data center 102), control details for equipment related to the data center 102, the degree (level) of availability that can be achieved when processing workloads in the operation mode 103, and a group of indicators associated with the operation mode 103. Here, the control of the equipment related to the data center 102 may specifically include the on / off control of the data center unit air conditioner 330, the temperature adjustment of the data center unit air conditioner 330, the humidity adjustment of the data center unit air conditioner 330, the control of the remaining charge of the data center unit storage battery 340, and the control of the stockpile of data center unit fuel 351 for the emergency generator. In addition, the group of indicators associated with the operation mode 103 may specifically include an index related to the unit cost in the operation mode 103 (which is given the highest priority (priority 1) when setting the operation mode 103) (a certain portion of the cost here may be the electricity bill), the proportion of electricity from power generation sources with relatively low carbon emissions (green electricity) in the amount of electricity consumed in the operation mode 103 (which is given the second highest priority (priority 2) when setting the operation mode 103), and an index related to the adjustment compensation in the operation mode 103 (the compensation that the operator of the base 202 / data center 102 can receive from the supply and demand adjustment market by reducing the amount of electricity that the base 202 / data center 102 receives from the transmission and distribution system 370) (which is given the third highest priority (priority 3) when setting the operation mode 103).

[0050] By associating the operation mode 103 with each of the control information listed above in advance, simple handling of setting the operation mode 103 for the data center 102 can be performed, while detailed handling can be achieved in controlling the equipment corresponding to the operation mode 103. For example, the degree (level) of availability realized in the data center 102 and the details (e.g., type, performance, and amount) of information processing resources (IT resources) that can be provided can be controlled in conjunction with setting the operation mode 103 in the data center 102. Here, the degree (level) of availability realized in the data center 102 and the details (e.g., type, performance, and amount) of information processing resources (IT resources) that can be provided can be easily compared with the "conditions" when processing a workload, so it is expected that it will be easier to determine whether or not a workload having a predetermined "condition" can be assigned (deployed) to the data center 102 to which the operation mode 103 is set. Furthermore, for example, in conjunction with the setting of operation mode 103 in data center 102, control settings can be made for facilities such as air conditioners, storage batteries, and emergency generators in order to achieve a predetermined level of availability in data center 102. In other words, specific control of facilities can be quickly implemented to achieve a predetermined level of availability. Since the operation modes 103 are pre-associated with the respective indicators listed above and the priorities between the indicators are pre-set, when creating the data center operation mode plan 110, it becomes easier to identify a combination that should be given priority among the set of combinations of values ​​(operation mode numbers) of the operation modes 103 for each data center 102 that can process the workload group included in the workload prediction 109.

[0051] 7, the data centers 102 (a group of data centers located at the core base 202-C) whose identification information (data center identifier (DC-ID)) of the data center 102 is "C-1," "C-2," or "C-3" share common records in the data center operation mode list table 700. Similarly, the data centers 102 (a group of data centers located at the regional base 202-R-1 or the regional base 202-R-2) whose identification information (data center identifier (DC-ID)) of the data center 102 is "R-1-1," "R-1-2," "R-2-1," or "R-2-2" share common records in the data center operation mode list table 700. Similarly, for data centers 102 (a group of data centers located at edge (area) site 202-E-1-1 or edge (area) site 202-E-1-2) whose identification information (data center identifier (DC-ID)) is "E-1-1-1" or "E-1-2-1", the records in the data center operation mode list table 700 are common. However, a record of the data center operation mode list table 700 may exist for each data center 102 that exists in the same base 202. Also, for example, a record of the data center operation mode list table 700 may be provided in common to a group of data centers that exist in the core base 202-C (core base C) and the regional base 202-R-1.

[0052] 7, the second-to-first record in the data center operation mode list table 700 indicates that operation mode "1" exists in a data center 102 whose data center identifier (DC-ID) is one of "C-1," "C-2," or "C-3." This record also indicates that when operation mode "1" is set in a data center 102 whose data center identifier (DC-ID) is one of "C-1," "C-2," or "C-3," the ratio of servers 302 held by the data center 102 that are powered on is controlled to 30% (and accordingly, the ratio of the performance and capacity of information processing resources that can be provided for workload processing in this operation mode 103 to the maximum value of the performance and capacity of information processing resources that can be realized by all of the IT resources held by the data center 102 is also set to 30%). Furthermore, the record indicates that when operation mode "1" is set for a data center 102 whose data center identifier (DC-ID) is any of "C-1," "C-2," or "C-3," the power to the data center unit air conditioner 330 is turned on, the temperature is controlled to be within a range of 18 to 27 degrees Celsius, the humidity is controlled to be within a range of 40 to 60 percent, the remaining charge of the data center unit storage battery 340 is controlled to be 60 percent or more, and the stockpile of data center unit fuel 351 for the emergency generator is controlled to be 60 percent or more, and based on the control of these facilities, the degree (level) of availability that can be achieved when processing workloads is set to "2." In addition, the record indicates that when operation mode "1" is set for a data center 102 whose data center identifier (DC-ID) is either "C-1," "C-2," or "C-3," in the group of indicators associated with operation mode 103, the unit cost for operation mode 103 will be "0.05 million yen / month," the proportion of electricity from power generation sources with relatively low carbon emissions (green electricity) in the amount of electricity consumed in operation mode 103 will be "0.7 (70%)," and the index related to the adjusted compensation for operation mode 103 will be "0.1." The example in Figure 7 shows that as the amount of information processing resources (IT resources) that can be provided increases, the proportion of electricity (green electricity) from power generation sources with relatively low carbon emissions in the total power consumption decreases. This example shows a case where there is an upper limit on the amount of green electricity that can be used by one data center 102. However, there are various methods for achieving carbon neutrality, such as purchasing non-fossil fuel certificates, purchasing green electricity from power companies, and procuring renewable energy through power purchase agreements (PPAs). Therefore, in other examples, the trend in the proportion of green electricity for each operating mode may differ from that shown in Figure 7.

[0053] 4.2.3. Overview of Workload Performance Table 800 (Figure 8) FIG. 8 shows a workload performance table 800 . The workload performance table 800 shows information about the performance of the workload execution request 211 (performance of receiving the workload execution request 211 and performance of processing the workload) received by the system 101 from the execution request device 105 that is the issuer of the workload execution request 211. The workload performance table 800 may have a record for each workload execution request 211 received by the system 101. 8, the workload actual table 800 includes, for each record, information such as the date and time when the system 101 received the workload execution request 211, the constraints or requirements imposed on the data center 102 that will be the processing entity when processing the workload associated with the workload execution request 211, the details of the IT resources used when processing the workload (mainly the amount of IT resources in the example of FIG. 8), identification information (data center identifier (DC-ID)) of the data center 102 where the workload was actually deployed (actual deployment destination data center), and the processing time (actual required time) required when the workload was actually processed. (Note that, among the information shown above, if the identification information of the actual deployment destination data center is not used in the processing by the workload prediction unit 1700, the identification information of the actual deployment destination data center does not need to be included in the record of the workload actual table 800 shown in FIG. 8.) Here, the constraints or requirements imposed on the data center 102 that will be the processing subject when processing a workload may specifically include information (data center identifier (DC-ID)) that identifies the data center 102 (hereinafter simply referred to as the "deployable data center") to which the workload can be deployed if other conditions (constraints, requirements, amount of allocatable free IT resources, etc.) are met, and the level of availability required when processing the workload. Also, the content of the IT resources used when processing the workload may specifically include the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity used when processing the workload. The constraints or requirements imposed on the data center 102 that is the processing entity when processing the workload associated with the workload execution request 211, combined with the content of the IT resources used when processing the workload, may represent the ``conditions'' under which the workload is processed.

[0054] For example, the first record in the workload performance table 800 in the example of FIG. 8 indicates that the system 101 received a workload execution request 211 at 00:01:34 on April 21, 2024. The record also indicates that, with regard to the conditions for processing the workload associated with the workload execution request 211, the data center identifier (DC-ID) of the "deployable data center" is one of "C-1," "C-2," "C-3," "R-1-1," "R-1-2," or "E-1-1-1," the level of availability required for processing the workload is "1 or higher (1, 2, 3, ...)," the number of central processing units (CPUs) used for processing the workload is "3," the number of graphics processing units (GPUs) is "1," and the memory capacity is "2 gigabytes (GB)." Furthermore, the record indicates that the data center identifier (DC-ID) of the data center 102 (actual deployment data center) to which the workload associated with the workload execution request 211 was actually deployed is "E-1-1-1," and that the time required to actually process the workload (actual required time) was "60 seconds."

[0055] 4.2.4. Overview of Workload Forecast Table 900 (Figure 9) FIG. 9 shows a workload forecast table 900 . The workload prediction table 900 indicates the content of the workload prediction 109. The workload prediction table 900 may have information indicating the content of the workload to be processed over a period of a predetermined length (e.g., one day (24 hours) or seven days (168 hours)). The information indicating the content of the workload to be processed here may include, for example, information such as the conditions under which the workload is processed (constraints or requirements imposed on the data center 102 that processes the workload), the content of the IT resources used to process the workload (e.g., type, performance, and amount), and the temporal amount of the workload. Furthermore, the workload prediction 109 indicated by the workload prediction table 900 may consist of a prediction for each time period of a predetermined length (e.g., one hour or four hours). 9, a record exists in the workload prediction table 900 for each combination of a time period to be predicted (a time period of one hour in the example of FIG. 9) and constraints or requirements imposed on the data center 102 that will be the processing entity when processing the workload. The record may include information on the content of the IT resources (e.g., type, performance, and amount) and the temporal amount of the workload for the time period and the combination of constraints or requirements. In other words, the workload prediction table 900 may include multiple records for one time period. 9, the workload prediction table 900 has, for each record, information such as the type and number of IT resources (number of central processing units (CPUs), number of graphics processing units (GPUs), and memory capacity) to be used when processing the workload, and the time (required time) predicted to be required to process the workload using these types and numbers of IT resources, for a combination of the predicted target time period (a one-hour time period in the example of FIG. 9), the data center identifier (DC-ID) of the "available data center," and the level of availability required when processing the workload. (Note that if the required time in each record of the workload prediction table 900 is always set to a constant value (for example, 3600 seconds), the records of the workload prediction table 900 do not need to explicitly include information about the required time.)

[0056] For example, the first record in the workload forecast table 900 in the example of Figure 9 indicates that a group of workloads for the time period from midnight to 1:00 on April 23, 2024, in which the data center identifier of the "deployable data center" is one of "C-1," "C-2," "C-3," "R-1-1," "R-1-2," or "E-1-1-1," and in which the level of availability required for processing the workload is "1 or higher (1, 2, 3, etc.)," ​​will collectively require 3,600 seconds of processing by information processing resources including 10,000 central processing units (CPUs), 1,000 graphics processing units (GPUs), and 7,500 gigabytes (GB) of memory.

[0057] 11 described later, for all workload execution requests 211, the processing of the workload associated with the workload execution request 211 does not necessarily need to be immediately executed in one of the data centers 102. In other words, the workload execution request 211 may be set with a length of time (permissible execution delay time) for which the execution of the processing of the workload associated with the workload execution request 211 is allowed to be delayed. In this regard, the workload forecast table 900 of FIG. 9 may have a record for each combination of the predicted target time period (a time period of one hour in the example of FIG. 9), the length of the allowable execution delay time, and the constraints or requirements imposed on the data center 102 that will be the processing entity when processing the workload.

[0058] 4.2.5. Overview of Data Center Operation Mode Plan Table 1000 (Fig. 10) FIG. 10 shows a data center operation mode plan table 1000. The data center operation mode plan table 1000 shows a data center operation mode plan 110. The data center operation mode plan table 1000 may show, for example, the operation modes 103 to be set for each of the data centers 102 for a period of a predetermined length (for example, one day (24 hours) or seven days (168 hours)). Furthermore, the data center operation mode plan 110 shown by the data center operation mode plan table 1000 may consist of the setting of the operation modes 103 for each time slot of a predetermined length (for example, one hour or four hours). In the example of Figure 10, the data center operation mode plan table 1000 has, for each record, information such as identification information for the location 202, identification information for the data center 102 (data center identifier (DC-ID)), time period (in the example of Figure 10, the length of the time period is one hour), and identification information for the operation mode 103 (operation mode number).

[0059] For example, the first record of the data center operation mode plan table 1000 in the example of Figure 10 indicates that the operation mode for the data center 102 located at "core site 202-C (core site C)" and having a data center identifier (DC-ID) of "C-1" is planned to be set to "2" from 0:00 to 1:00 on April 23, 2024.

[0060] 4.2.6. Overview of Workload Execution Request Buffer Table 1100 (Figure 11) FIG. 11 shows a workload execution request buffer table 1100. The workload execution request buffer table 1100 buffers information about workload execution requests 211 received by the system 101. The information about the workload execution requests 211 held in the workload execution request buffer table 1100 may include information determined by the workload execution request receiver 511 about the workload associated with the workload execution request 211, in addition to information explicitly held in the workload execution request 211 itself. The workload execution request buffer table 1100 may have a record for each workload execution request 211 received by the system 101. (However, after the processing of the workload associated with the workload execution request 211 corresponding to the record is completed in one of the data centers 102 and the information about the workload execution request 211 is reflected in the workload performance table 800, the record may be deleted from the workload execution request buffer table 1100.) 11, the workload execution request buffer table 1100 includes, for each record, information such as identification information assigned to the workload (workload identification number (WL-ID)), the date and time when the system 101 received the workload execution request 211, constraints or requirements imposed on the data center 102 that will be the processing subject when processing the workload associated with the workload execution request 211, the details of the IT resources used when processing the workload (mainly the amount of IT resources in the example of FIG. 11), the length of time that the execution of the workload processing can be delayed (allowable execution delay time), and the length of time that is expected to be required to process the workload in one of the data centers (expected required time). Here, the constraints or requirements imposed on the data center 102 that will be the processing subject when processing the workload may specifically include information (data center identifier (DC-ID)) that identifies the data center 102 that can deploy the workload (the "deployable data center") if other conditions (constraints, requirements, the amount of allocatable free IT resources, etc.) are satisfied, and information on the degree (level) of availability required when processing the workload. In addition, the IT resources used when processing a workload may specifically include the number of central processing units (CPUs), the number of graphics processing units (GPUs), and the memory capacity used when processing the workload. In addition, the constraints or requirements imposed on the data center 102 that is the processing entity when processing the workload associated with the above-mentioned workload execution request 211, combined with the content of the IT resources used when processing the workload, may be considered to indicate the ``conditions'' under which the workload is processed.

[0061] By configuring the "conditions" when a workload is processed as described above, the geographical conditions of the data center 102 that will be the processing entity (or the requirement for a fast response time for the processing results), the availability conditions, and the processing speed and processing volume conditions can all be taken into consideration when allocating (deploying) the workload to the data center 102. Furthermore, by making it possible to set the allowable execution delay time as described above, it is possible to schedule workload processing according to the characteristics of the workload. For example, if the workload has a strong batch processing characteristic, it is often acceptable to delay the start of processing of that workload to some extent. By utilizing the characteristics of the workload as described above, it is expected that processing schedules in the data centers 102 can be made more flexible.

[0062] For example, the first record in the workload execution request buffer table 1100 in the example of FIG. 11 indicates that the system 101 received a workload execution request 211 at 00:01:01 on April 23, 2024. The record also indicates that a WL identification number (WL-ID) of "20240423-1" was assigned to the received workload execution request 211 (or the workload associated with the request). The record further indicates that, with regard to the conditions for processing the workload associated with the workload execution request 211, the data center identifier (DC-ID) of the "available data center" is one of "C-1," "C-2," "C-3," "R-1-1," "R-1-2," or "E-1-1-1," the level of availability required for processing the workload is "1 or higher (1, 2, 3, ...)," the number of central processing units (CPUs) used to process the workload is "2," the number of graphics processing units (GPUs) is "1," and the memory capacity is "3 gigabytes (GB)." Additionally, the record indicates that the allowable execution delay time is set to "immediate" (i.e., execution delay is not allowed). Finally, the record indicates that the estimated time required to process the workload associated with the workload execution request using information processing resources with two central processing units (CPUs), one graphics processing unit (GPU), and three gigabytes (GB) of memory is "60 seconds."

[0063] 4.2.7. Overview of Workload Deployment Setting Table 1200 (Figure 12) FIG. 12 shows a workload deployment setting table 1200 . The workload deployment setting table 1200 indicates the deployment setting 112 of a workload to a data center. The workload deployment setting table 1200 may have, for example, a record for each workload execution request 211 accepted by the system 101. (However, after the processing of the workload associated with the workload execution request 211 corresponding to the record is completed in one of the data centers 102, the record may be deleted from the workload deployment setting table 1200.) 12, the workload deployment setting table 1200 has, for each record, information such as identification information assigned to the workload (workload identification number (WL-ID)), identification information (datacenter identifier (DC-ID)) of the data center 102 to which the workload is actually deployed (actual deployment destination data center), the time when processing of the workload is started (started) or the time when processing is scheduled to start (WL execution start time or scheduled WL execution start time), and details of IT resources allocated by the deployment destination data center 102 for processing the workload (mainly the amount of IT resources in the example of FIG. 12). Here, the information on the details of IT resources allocated by the deployment destination data center 102 for processing the workload (mainly the amount of IT resources in the example of FIG. 12) may include information on the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity.

[0064] For example, the first record in the workload deployment setting table 1200 in the example of Figure 12 indicates that for a workload assigned the WL identification number (WL-ID) "20240423-1", the identification information (data center identifier (DC-ID)) of the data center 102 to which the workload is actually deployed is "E-1-1-1", the execution start time of the workload is set to "immediate" (execution delay is not allowed), and the IT resources allocated for processing the workload include "2" central processing units (CPUs), "1" graphics processing unit (GPU), and "3 gigabytes (GB)" of memory.

[0065] 4.2.8. Overview of the Workload Execution History Table 1300 (Figure 13) FIG. 13 shows a workload execution history table 1300 . The workload execution history table 1300 indicates information about the execution history and execution status of each workload associated with a workload execution request 211 received by the system 101. The workload execution history table 1300 may have, for example, a record for each workload execution request 211 received by the system 101. (However, after the processing of the workload associated with the workload execution request 211 corresponding to the record is completed in one of the data centers 102 and information about the workload execution request 211 is reflected in the workload result table 800, the record may be deleted from the workload execution history table 1300.) 13 , the workload execution history table 1300 has, for each record, information such as identification information assigned to the workload (workload identification number (WL-ID)), identification information (datacenter identifier (DC-ID)) of the data center 102 where the workload was actually deployed (actual deployment destination data center), information about the time interval during which the workload was executed (executed) (actual WL execution time interval), details of IT resources allocated by the deployment destination data center 102 for processing the workload (mainly the amount of IT resources in the example of FIG. 13 ), and the execution state of the workload (WL execution state). Here, the information about the time interval during which the workload was executed (executed) (actual WL execution time interval) may include the time when processing of the workload started or the time when processing is scheduled to start (WL execution start time or scheduled WL execution start time), and (if processing of the workload has finished (completed)) the time when processing of the workload finished (completed). In addition, information on the content of IT resources allocated by the data center 102 where the workload is deployed to process it (mainly the amount of IT resources in the example of Figure 13) may include information on the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity.

[0066] Possible states for the workload execution state (WL execution state) include, for example, "waiting for execution," which is the state before the execution start time arrives, "running," which is the state after the workload processing has started but before the processing has ended (completed), and "completed," which is the state after the workload processing has ended (completed).

[0067] For example, the first record in the workload execution history table 1300 in the example of Figure 13 indicates that for a workload assigned the WL identification number (WL-ID) "20240423-1", the identification information (data center identifier (DC-ID)) of the data center 102 where the workload is actually deployed is "E-1-1-1", the start time is 00:01:01 on April 23, 2024, the end time (completion time) is 00:02:01 on April 23, 2024, the IT resources allocated to process the workload include "2" central processing units (CPUs), "1" graphics processing unit (GPU), and "3 gigabytes (GB)" of memory, and the execution status of the workload is "completed".

[0068] Furthermore, if the data center to which a workload is deployed is changed (the workload is redeployed (rebalanced, migrated)), the identification information (data center identifier (DC-ID)) of the data center to which the workload was actually deployed in the record of the workload execution history table 1300 for that workload may also be changed. The lower part of Figure 13 (showing the status of the workload execution history table 1300 at a later time than the upper part of Figure 13 (the part above the "downward arrow" in Figure 13)) shows that for a workload assigned the WL identification number (WL-ID) "20240423-87", the identification information of the data center where the workload was actually deployed (data center identifier (DC-ID)) in the record of the workload execution history table 1300 for that workload is changed from "C-1" to "C-2" (redeployment (rebalancing, migration) is performed).

[0069] 4.2.9. Overview of the Workload Redistribution Setting Table 1400 (Fig. 14) FIG. 14 shows a workload redistribution setting table 1400. As shown in FIG. The workload redeployment setting table 1400 indicates the details of the change (redeployment (rebalancing, migration)) when the destination data center of a workload is changed (when workload redeployment (rebalancing, migration) is performed). The workload redeployment setting table 1400 may have a record corresponding to each setting of the change of the destination data center of a workload (setting of workload redeployment (rebalancing, migration)). (However, after the redeployment (rebalancing, migration) of the workload corresponding to the record is performed and the details of the redeployment (rebalancing, migration) are reflected in the workload deployment setting table 1200 and the workload execution history table 1300, the record may be deleted from the workload redeployment setting table 1400.) 14, the workload redeployment setting table 1400 has, for each record, information such as identification information assigned to the workload (workload identification number (WL-ID)), identification information of the deployment destination data center before redeployment (deployment-in-progress data center) (datacenter identifier (DC-ID)), identification information of the redeployment destination data center (datacenter identifier (DC-ID)), the scheduled time of redeployment (scheduled redeployment time), and details of IT resources to be allocated by the redeployment destination data center 102 for processing the workload (mainly the amount of IT resources in the example of FIG. 14). Here, the details of IT resources to be allocated by the redeployment destination data center 102 for processing the workload (mainly the amount of IT resources in the example of FIG. 14) may include information such as the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity.

[0070] For example, the first record in the workload redeployment setting table 1400 in the example of Figure 14 indicates that for a workload assigned the WL identification number (WL-ID) of "20240423-87", the data center 102 to which the workload is deployed will be changed from a data center 102 having a data center identifier (DC-ID) of "C-1" to a data center 102 having a data center identifier (DC-ID) of "C-2" (redeployment (rebalancing, migration) will be performed), the scheduled redeployment time is "1:00 a.m. on April 23, 2024", and the IT resources allocated for processing the workload include "20" central processing units (CPUs), "10" graphics processing units (GPUs), and "50 gigabytes (GB)" of memory.

[0071] 5. Computer Architecture for Implementing Embodiments of the Present Disclosure (FIG. 15) 15 illustrates a computer architecture 1500 for implementing the system 101 of the embodiment of the present disclosure. The computer architecture 1500 illustrated in FIG. 15 may be referred to as an information processing device or an information processing system. To realize the system 101, some or all of the processing unit 1501, storage device 1502, non-volatile storage medium (storage device) 1503, external storage medium drive 1504, input device 1506, display or output device 1507, communication device 1508, external input / output port 1509, and reading device 1510 may be interconnected via an interconnection unit 1511. (Note that some or all of the interconnection unit 1511 may be a network. In that case, the system 101 is realized by a plurality of devices connected via the network.) The arithmetic processing device 1501 may be, for example, a processor. Examples of this processor include a CPU, an MPU, or a GPU. Alternatively, the processor referred to here may be any other semiconductor device that executes predetermined processing. Furthermore, the arithmetic processing device 1501 may be one or more (micro)processors. For example, the arithmetic processing device 1501 may be a multi-core processor having multiple processing cores (CPU cores). The storage device 1502 may be, for example, a memory. The non-volatile recording medium (recording device) 1503 may be, for example, a non-volatile memory (e.g., a flash memory) or a non-volatile disk device. The external recording medium drive 1504 may be, for example, a disk drive. The input device 1506 may be, for example, a mouse, a keyboard, an imaging device, a sensor, a touch panel, or a pointing device. The display or output device 1507 may be, for example, a display, a printer, or a speaker. The communication device 1508 may be, for example, a communication device for wired communication or a communication device for wireless communication. The communication device 1508 may be a network interface device (NIC) that controls communication with other systems, data centers, devices, terminals, or servers according to a predetermined protocol. The interconnection unit 1511 may be, for example, a bus or a crossbar switch. (As described above, part or all of the interconnection unit 1511 may be a network.)

[0072] The non-volatile recording medium (recording device) 1503 may record various programs included in the program group 1531 (for example, programs for realizing the functional configuration related to the present disclosure; for example, various programs for implementing each of the functional units realized in the system 101), various data groups included in the data group 1532, or information included in the various information 1533. The program group 1531 may include various programs for realizing each of the functional units indicated as "units" in the functional configuration diagrams of Figures 1 and 5. Some of the above programs may be integrated into one program. Also, any of the above programs may be divided into multiple programs. The data group 1532 may include information (data, etc.) handled by the above-mentioned functional units. For example, the data group 1532 may include information constituting each of the various "tables" in the functional configuration diagram of FIG. 5. (Note that some or all of the information contained in the various "tables" may be stored in the storage device 1502 (memory).) Alternatively, some or all of the various programs included in the program group 1531, the various data groups included in the data group 1532, or the information included in the various information 1533 may be obtained from outside the configuration shown in FIG. 15.

[0073] The external recording medium drive 1504 can be connected to an external recording medium 1505. The external recording medium 1505 may be, for example, a portable recording disk (such as a DVD), an IC card, an SD card, a nonvolatile memory (such as a flash memory), or a portable hard disk. Various programs included in the program group 1531, various data included in the data group 1532, or information similar to the information included in the various information 1533 may be transferred and stored from the external recording medium 1505 to the nonvolatile recording medium (recording device) 1503 or the storage device 1502. The external recording medium 1505 may be used to record programs and data handled in the system 101. The external recording medium drive 1504 and the external recording medium 1505 may be connected to the system 101 shown in FIG. 15 via a network. The various programs included in the program group 1531, the various data included in the data group 1532, or the information included in the various information 1533 may be brought via the communication device 1508, the external input / output port 1509, the input device 1506, or the reading device 1510, and recorded or stored in the non-volatile recording medium (recording device) 1503 or the memory device 1502.

[0074] In order for the architecture of FIG. 15 to function as the system 101, each functional unit within the system 101, or a part of each functional unit (to execute one or a series of processes (steps)), various programs included in the program group 1531 may be loaded into the storage device 1502 (for example, from a non-volatile recording medium (recording device) 1503). The loaded program is indicated by 1521 in FIG. 15. The arithmetic processing device 1501 may then execute the program 1521 (using, as necessary, various data and the like included in the data group 1532 stored in the non-volatile recording medium (recording device) 1503, or information included in the various information 1533). Execution of the program 1521 realizes the function of the system 101, each functional unit within the system 101, or a part of each functional unit (to execute one or a series of processes (steps)). At this time, various buffers 1523 temporarily formed in the storage device 1502 may also be used as appropriate.

[0075] 6. Processing performed by the embodiment of the present disclosure The following describes the processing performed by an embodiment of the present disclosure (system 101). Note that it is not necessary to realize all of the functional configurations and perform all of the processing described below. Furthermore, it is not prohibited to realize functional configurations and perform processing other than the functional configurations and processing described below. Furthermore, the steps of the processes described below may be combined to form a method executed by a system (information processing device or information processing system). In the flowcharts shown in FIGS. 16, 17, 18, 19 and 20, numbers enclosed in circles are connected to each other if they have the same number.

[0076] 6-1. Workload performance table creation process (Figure 16) Fig. 16 shows a flowchart of the processing executed by the workload result table creating unit 1600. The processing will be explained below in the order shown in Fig. 16. Note that each of the processing steps performed by the workload result table creating unit 1600 in the flowchart of Fig. 16 may be considered to form a "workload result table creating step." Since the functions described below are realized, it is possible to create a workload performance table 800 having records that appropriately reflect the performance of receiving workload execution requests 211 and the performance of processing the workload associated with the workload execution requests 211. In this way, since the workload performance table 800 having records that appropriately reflect the performance related to the workload is created, it is possible to prepare information for making predictions 109 related to the workload.

[0077] In step 1601 of Fig. 16, the workload result table creation unit 1600 determines whether it is time to record a new record in the workload result table 800. More specifically, the workload result table creation unit 1600 determines whether a new record in which the workload execution status is "completed" has appeared in the workload execution history table 1300 shown in Fig. 13. If the determination result in step 1601 is positive, control transitions to step 1602. If the determination result in step 1601 is negative, step 1601 is repeated. 16, the workload result table creation unit 1600 acquires information about the record for which the workload execution state has been determined to have newly become "completed" in step 1601 from the workload execution history table 1300. Then, the workload result table creation unit 1600 grasps each piece of information included in the acquired record information, such as workload identification information (workload identification number (WL-ID)), data center (DC-ID) where the workload is deployed, start time (of processing the workload), and end time (of processing the workload). In step 1603 of FIG. 16, the workload actual table creation unit 1600 calculates the time actually required to process the workload (actual required time) based on the start time (of processing the workload) and end time (of processing the workload) determined in step 1602. In step 1604 of FIG. 16, the workload result table creating unit 1600 identifies one of the records in the workload execution request buffer table 1100 shown in FIG. 11 using the workload identification information (workload identification number (WL-ID)) identified in step 1602. The workload result table creating unit 1600 acquires information about the identified record from the workload execution request buffer table 1100. The workload result table creating unit 1600 acquires information about the acquired record, including the date and time when the workload execution request 211 was accepted by the system 101 (workload execution request acceptance date and time), the constraints and requirements imposed on the data center 102 that is the processing entity when processing the workload, and the details of the IT resources used when processing the workload (mainly the amount of IT resources in the example of FIG. 11). At this time, the workload result table creating unit 1600 may also acquire information about the time a delay is allowed when executing the processing of the workload (allowable execution delay time). In step 1605 of Figure 16, the workload performance table creation unit 1600 adds a new record to the workload performance table 800 shown in Figure 8, which reflects the performance of receiving the workload execution request 211 and the performance of processing the workload associated with the workload execution request 211. Specifically, the workload actual table creation unit 1600 may add to the newly added record the date and time when the workload execution request 211 grasped in step 1604 was received by the system 101 (workload execution request reception date and time), the constraints and requirements imposed on the data center 102 that will be the processing entity when processing the workload grasped in step 1604, the details of the IT resources (mainly the amount of IT resources in the example of Figure 11 or Figure 13) used when processing the workload grasped in step 1604 (or step 1602), the data center identifier (DC-ID) of the data center where the workload is actually deployed grasped in step 1602, and the time actually required to process the workload calculated in step 1603 (actual required time), and then store the record in the workload actual table 800. After step 1605 in FIG. Furthermore, if the data center identifier (DC-ID) of the data center where the actual results are deployed among the information shown above is not used in the processing of the workload prediction unit 1700, then in the processing shown in Figure 16 above, the workload actual table creation unit 1600 does not need to handle the data center identifier (DC-ID) of the data center where the actual results are deployed.

[0078] 6-2. Workload Prediction Processing (Figure 17) Fig. 17 shows a flowchart of the processing executed by the workload prediction unit 1700. The processing will be described below in the order shown in Fig. 17. Note that each of the processing steps executed by the workload prediction unit 1700 in the flowchart of Fig. 17 may be considered to form a "workload prediction step." Since the functions described below are realized, it is possible to create a workload prediction 109 that appropriately reflects the results of receiving workload execution requests 211 and the results of processing the workload associated with the workload execution requests 211, which are included in the workload result table 800. Furthermore, since the created workload prediction 109 is made up of predictions corresponding to each time period in the data center operation mode plan 110, it becomes easier to create the data center operation mode plan 110 using the workload prediction table 900 that stores information related to the workload prediction 109.

[0079] 17, the workload prediction unit 1700 determines whether it is time to create a new workload prediction 109. The workload prediction unit 1700 may determine, for example, any timing before the time to create a new data center operation mode plan 110 as the time to create a new workload prediction 109. Specifically, for example, if the data center operation mode plan 110 includes 24 hours from midnight to midnight (one day's worth) and the data center operation mode plan 110 is created at 23:30 the previous day (created at 23:30 every day), the creation of the workload prediction 109 (creation of a new workload prediction table 900) may be performed at 23:00 the previous day (or at 23:00 every day). Alternatively, for example, if the data center operation mode plan 110 includes 168 hours (seven days) from midnight every Sunday to midnight every Saturday, and the data center operation mode plan 110 is created at 11:30 PM every Saturday, the creation of the workload prediction 109 (creation of a new workload prediction table 900) may be performed at 11:00 PM every Saturday. The length of the period for which the operation mode 103 is set by the entire data center operation mode plan 110 may be a length other than the above-mentioned 24 hours (one day) or 168 hours (seven days). If the determination result of step 1701 is positive, control transitions to step 1702. If the determination result of step 1701 is negative, step 1701 is repeated.

[0080] 17, the workload prediction unit 1700 selects one of the time slots that are the time unit for controlling the operation mode 103 in the data center operation mode plan 110. For example, the length of a time slot in the data center operation mode plan 110 may be one hour or four hours (or may be any other length). For example, if the length of a time slot in the data center operation mode plan 110 is one hour and the length of the period for setting the operation mode 103 by the entire data center operation mode plan 110 is 24 hours (one day, from midnight to midnight), the workload prediction unit 1700 may select one of the time slots from midnight to 1:00, the time slot from 1:00 to 2:00, and the time slot from 11:00 to midnight in this step 1702. Alternatively, if the length of a time period in the data center operation mode plan 110 is one hour and the length of the period for which the operation mode 103 is set by the entire data center operation mode plan 110 is 168 hours (seven days, for example, from midnight on Sunday to midnight on Saturday), in this step 1702, the workload prediction unit 1700 may select one of the time period from midnight to 1:00 on Sunday, the time period from 1:00 to 2:00 on Sunday, and the time period from 11:00 to midnight on Saturday.

[0081] 17, the workload prediction unit 1700 obtains information used to create the prediction 109 for the workload for the time slot selected in step 1702 from the workload result table 800 shown in FIG. 8. The workload prediction unit 1700 may obtain, from the workload result table 800, records related to the results of receiving workload execution requests 211 for a certain period in the past (e.g., the most recent month) and records related to the results of processing the workload associated with the workload execution requests 211. Specifically, the workload prediction unit 1700 may obtain, from the workload result table 800, a group of records corresponding to the time slot selected in step 1702. For example, if the time slot selected in step 1702 is the time slot from midnight to 1:00, in step 1703, the workload prediction unit 1700 may obtain, from the workload result table 800, a group of records whose workload execution request date and time falls within the range from midnight to 1:00 (e.g., of any day included in the most recent month). Alternatively, for example, if the time period selected in step 1702 is the time period from midnight to 1:00 on Sunday, in step 1703, the workload prediction unit 1700 may obtain from the workload actual table 800 a group of records whose workload execution request date and time falls within the range from midnight to 1:00 on any Sunday (for example, within the past month).

[0082] 17, the workload prediction unit 1700 selects one of the possible settings as the constraints and requirements to be imposed on the data center 102 that will be the processing subject when processing the workload included in the record group acquired in step 1703. Here, the constraints and requirements to be imposed on the data center 102 that will be the processing subject when executing the workload may be a combination of a "data center that can be deployed" and the degree (level) of availability required when processing the workload. In other words, in step 1703, one of the possible settings may be selected as the combination of a "data center that can be deployed" and the degree (level) of availability. In accordance with the example of FIG. 8, for example, a combination of "C-1," "C-2," "C-3," "R-1-1," "R-1-2," or "E-1-1-1" as the data center identifier (DC-ID) of the "deployable data center" and "1 or more (1, 2, 3,...)" as the degree (level) of availability may be selected. Alternatively, a combination of "C-1," "C-2," "C-3," "R-1-1," or "R-1-2" as the data center identifier (DC-ID) of the "deployable data center" and "2 or more (2, 3,...)" as the degree (level) of availability may be selected. Alternatively, a combination of "C-1," "C-2," or "C-3" as the data center identifier (DC-ID) of the "deployable data center" and "3 or more (3, 4,...)" as the degree (level) of availability may be selected.

[0083] 17, the workload prediction unit 1700 extracts, from the records acquired in step 1703, a record group that corresponds to one of the possible settings of constraints and requirements imposed on the data center 102 that will be the processing subject when processing the workload selected in step 1704. Then, for the extracted records, the workload prediction unit 1700 totals the combination of the content of the IT resources (here, mainly the amount of IT resources) used to process the workload and the actual required time. 8 , if a combination of a “distributable data center” and a degree (level) of availability is selected in step 1704 as a combination of “available data center” and a degree (level) of availability in which the data center identifier (DC-ID) of the “distributable data center” is one of “C-1,” “C-2,” “C-3,” “R-1-1,” “R-1-2,” or “E-1-1-1,” and the degree (level) of availability is “1 or more (1, 2, 3, . . .),” the workload prediction unit 1700 extracts a set of records having the values ​​of the above combination in step 1705. Then, the workload prediction unit 1700 sums up, for each of the extracted records, the products of the values ​​indicated by the amount of IT resources (the number of central processing units (CPUs), the number of graphics processing units (GPUs), and the memory capacity) and the values ​​of the actual required time. The sum may be a set of parameters, such as the number of central processing units (CPUs), the number of graphics processing units (GPUs), the memory capacity, and the actual required time. The sum may be calculated separately for each day on which the workload is processed.

[0084] In step 1706 of FIG. 17, the workload prediction unit 1700 calculates predicted values ​​of the IT resource content (here, mainly the amount of IT resources) and required time to be used to process the workload group, which corresponds to one of the possible settings of constraints and requirements imposed on the data center 102 that is the processing subject when processing the workload selected in step 1704 during the time slot selected in step 1702. The workload prediction unit 1700 may calculate the predicted value by, for example, performing predetermined processing on the total value (for example, for each day on which the workload is processed) calculated in step 1705 (represented by, for example, a set of parameters expressed in the form of the number of central processing units (CPUs), the number of graphics processing units (GPUs), the memory capacity, and the actual required time). The predetermined processing here may be any processing that calculates a predicted value, and may be, for example, processing using an autoregressive technique or least squares method (LSM). (Note that the estimated required time may be set to the same value as the length of the time slot (e.g., one hour).)

[0085] In step 1707 of FIG. 17, the workload prediction unit 1700 stores in the workload prediction table 900 shown in FIG. 9 a record including the predicted value calculated in step 1706, which corresponds to one of the possible settings of constraints and requirements imposed on the data center 102 that will be the processing subject when processing the workload selected in step 1704 during the time slot selected in step 1702. (Note that if the required time is set to the same value as the length of the time slot (for example, one hour), information indicating the required time does not need to be included in the record of the workload prediction table 900.) As shown in FIG. 9, the workload forecast table 900 has a record for each combination of a time period and possible settings of constraints and requirements imposed on the data center 102 .

[0086] In step 1708 of Fig. 17, the workload prediction unit 1700 determines whether all of the settings that can be constraints and requirements to be imposed on the data center 102 that will be the processing subject when processing the workload, which are included in the record group acquired in step 1703, have been selected in step 1704. If the determination result in step 1708 is positive, control transitions to step 1709. If the determination result in step 1708 is negative, control returns to step 1704, and another setting that can be a constraint and requirement to be imposed on the data center 102 that will be the processing subject when processing the workload is selected. 17, the workload prediction unit 1700 determines whether all of the time slots, which are time units for setting the operation modes 103 in the data center operation mode plan 110, have been selected in step 1702. If the determination result in step 1709 is positive, control is returned to step 1701, and the workload prediction unit 1700 essentially waits until it is time to create a prediction 109 for the next workload. If the determination result in step 1709 is negative, control is returned to step 1702, and a time slot that has not yet been selected is newly selected.

[0087] 6-3. Processing of Data Center Operation Mode Planning Unit (Figure 18) Fig. 18 shows a flowchart of processing executed by the data center operation mode plan creation unit 1800. The processing will be described below in the order shown in Fig. 18. Note that each of the processing steps executed by the data center operation mode plan creation unit 1800 in the flowchart of Fig. 18 may be understood to form a "data center operation mode plan creation step." The functions described below are realized, so that a data center operation mode plan 110 can be created in consideration of priorities while appropriately responding to the workload prediction 109 indicated by the workload prediction table 900.

[0088] 18, the data center operation mode plan creation unit 1800 determines whether it is time to create a new data center operation mode plan 110. The data center operation mode plan creation unit 1800 may determine, for example, any timing before the entire period (for example, a period of 24 hours (one day) or a period of 168 hours (seven days)) during which the setting of the operation mode 103 by the data center operation mode plan 110 is used, as the time to create a new data center operation mode plan 110. Specifically, for example, if the data center operation mode plan 110 includes 24 hours (one day) from midnight to midnight, the data center operation mode plan 110 may be created at 23:30 the previous day (or at 23:30 every day). Alternatively, for example, if the data center operation mode plan 110 includes 168 hours (seven days) from midnight every Sunday to midnight every Saturday, the data center operation mode plan 110 may be created at 11:30 PM every Saturday. The length of the period for which the operation mode 103 is set by the entire data center operation mode plan 110 may be a length other than the above-mentioned 24 hours (one day) or 168 hours (seven days). If the determination result of step 1801 is positive, control transitions to step 1802. If the determination result of step 1801 is negative, step 1801 is repeated.

[0089] In step 1802 of FIG. 18, the data center operation mode plan creation unit 1800 selects one of the time periods that is the time unit for controlling the operation mode 103 in the data center operation mode plan 110. For example, the length of a time period in the data center operation mode plan 110 may be one hour or four hours (or may be any other length). For example, if the length of a time period in the data center operation mode plan 110 is one hour and the length of the period for setting the operation mode 103 by the entire data center operation mode plan 110 is 24 hours (one day, from midnight to midnight), in step 1802 the data center operation mode plan creation unit 1800 may select one of the time period from midnight to 1:00, the time period from 1:00 to 2:00, and the time period from 11:00 to midnight. Alternatively, if the length of a time period in the data center operation mode plan 110 is one hour and the length of the period for which the operation mode 103 is set by the entire data center operation mode plan 110 is 168 hours (seven days, from midnight on Sunday to midnight on Saturday), in this step 1802, the data center operation mode plan creation unit 1800 may select one of the time period from midnight to 1:00 on Sunday, the time period from 1:00 to 2:00 on Sunday, and the time period from 11:00 to midnight on Saturday.

[0090] 18, the data center operation mode plan creation unit 1800 acquires a group of records indicating the prediction 109 related to the workload for the time period selected in step 1802 from the workload prediction table 900 shown in FIG. 9. For example, if the time period selected in step 1802 is the time period from midnight to 1 o'clock, then in step 1803 the data center operation mode plan creation unit 1800 may acquire a group of records in which the time period is set from midnight to 1 o'clock from the workload prediction table 900. Alternatively, if the time period selected in step 1802 is the time period from midnight to 1 o'clock on Sunday, then in step 1803 the data center operation mode plan creation unit 1800 may acquire a group of records in which the time period is set from midnight to 1 o'clock on Sunday from the workload prediction table 900.

[0091] 18, the data center operation mode plan creation unit 1800 selects one of the combinations of operation mode 103 settings that can be made for each of the data centers 102 that are the targets of the data center operation mode plan 110. In accordance with the example of FIG. 7, for example, the data center operation mode plan creation unit 1800 sets the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "C-1" to "1", the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "C-2" to "2", the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "C-3" to "3", the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "R-1-1" to "0", the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "R-1-1" to "1", the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "R-1 ...1" to "2", the operation mode 103 setting of the data center 102 whose data center identifier (DC-ID) is "R-1-1" to "3", the operation mode 103 setting of the data center 102 whose data center identifier (DC The following combinations may be selected: setting the operation mode 103 of a data center 102 whose data center identifier (DC-ID) is "R-2-2" to "1", setting the operation mode 103 of a data center 102 whose data center identifier (DC-ID) is "E-1-1-1", and setting the operation mode 103 of a data center 102 whose data center identifier (DC-ID) is "E-1-2-1" to "2", setting the operation mode 103 of a data center 102 whose data center identifier (DC-ID) is "E-1-2-1" to "1", etc. When executing step 1804, the data center operation mode plan creation unit 1800 may select a combination in descending order of priority. The data center operation mode list table 700 shown in FIG. 7 indicates index values ​​corresponding to combinations of a data center identifier (DC-ID) and an operation mode 103 (operation mode number). Here, the index corresponding to a combination of a data center identifier (DC-ID) and an operation mode 103 (operation mode number) may be an index related to the unit cost of the operation mode 103, the proportion of the amount of power consumed in the operation mode 103 that is derived from a power generation source with relatively low carbon emissions (green-derived power), or the adjustment compensation for the operation mode 103 (the compensation that the operator of the site 202 / data center 102 can receive from the balancing market by reducing the amount of power received by the site 202 / data center 102 from the power transmission / distribution system 370). Priorities may also be set among these three indexes. For example, the unit cost may be associated with the highest priority (priority 1). The second highest priority (priority 2) may be associated with the proportion of green-derived electricity. The third highest priority (priority 3) may be associated with an index related to the adjustment compensation. The data center operation mode planning unit 1800 sequentially selects combinations of possible settings for the operation modes 103 that each of the data centers 102 can be in, in order of the best of these indexes.

[0092] 18, the data center operation mode planning unit 1800 selects one record (a record in the workload prediction table 900) from the group of records acquired in step 1803. The data center operation mode planning unit 1800 grasps information on constraints and requirements imposed on the data center 102 that will be the processing subject when processing the workload, information on the content (quantity) of IT resources used to process the workload, and information on the required time, all of which are included in the selected record. 9, when the first record in the workload prediction table 900 is selected in step 1805, it is understood that the constraints and requirements imposed on the data center 102 that will be the processing subject when processing the workload are that the data center identifier (DC-ID) of the "data center available for deployment" is one of "C-1," "C-2," "C-3," "R-1-1," "R-1-2," or "E-1-1-1," and that the level of availability required when processing the workload is "1 or higher (1, 2, 3, ...)." It is also understood that the amount of IT resources used to process the workload and the required time are predicted to be a total workload amount equivalent to information processing resources having 10,000 central processing units (CPUs), 1,000 graphics processing units (GPUs), and 7,500 gigabytes (GB) of memory performing processing for 3,600 seconds.

[0093] 18, the data center operation mode planning unit 1800 attempts to allocate (virtually allocate) the predicted workload groups indicated by the records (in the workload prediction table 900) selected in step 1805 to each of the data centers 102, assuming the combination of operation modes 103 for each data center 102 selected in step 1804. Specifically, the data center operation mode planning unit 1800 attempts to virtually secure a combination of IT resource amounts and time (providable in accordance with the operation mode 103) equivalent to the combination of IT resource amounts and required time used to process the workload groups indicated by the records (in the workload prediction table 900) selected in step 1805, in one or more data centers 102 that satisfy the constraints and requirements imposed on the data center 102 that will be the processing subject when processing the workload groups indicated by the records (in the workload prediction table 900) selected in step 1805. For example, in the example of Figure 9, if the first record in the workload prediction table 900 is selected in step 1805, then in step 1806 the data center operation mode plan creation unit 1800 identifies a data center 102 whose data center identifier (DC-ID) is one of "C-1", "C-2", "C-3", "R-1-1", "R-1-2", or "E-1-1-1" and whose degree (level) of availability that can be achieved for processing the workload is "1 or higher (1, 2, 3, etc.)" (the data center operation mode list table 700 may be referenced at this time). Then, the data center operation mode plan creation unit 1800 attempts to virtually secure available resources in one or more of the identified data centers 102, corresponding to the total workload of an information processing resource having 10,000 central processing units (CPUs), 1,000 graphics processing units (GPUs), and 7,500 gigabytes (GB) of memory performing processing for 3,600 seconds (at this time, the data center list table 600 and the data center operation mode list table 700 may be referenced).

[0094] 18, the data center operation mode plan creation unit 1800 determines whether the virtual allocation (virtual reservation of available resources) attempted in step 1806 was successful. If the determination result in step 1807 is positive, control transitions to step 1810. If the determination result in step 1807 is negative, control transitions to step 1808.

[0095] (In response to the negative determination result in step 1807, which indicates that the combination of operation modes 103 for each data center 102 selected when step 1804 was most recently executed cannot accommodate the workload prediction 109 for the time slot selected when step 1802 was most recently executed), in step 1808 of FIG. 18 , the data center operation mode plan creation unit 1800 determines whether all of the combinations of operation modes 103 that can be taken by each data center 102 for the time slot selected when step 1802 was most recently executed have been selected in step 1804. If the determination result in step 1808 is positive, control transitions to step 1809. If the determination result in step 1808 is negative, control returns to step 1804, and one of the combinations of operation modes 103 for each data center 102 that has not yet been selected is newly selected. (When step 1809 in FIG. 18 is reached, it means that a combination of operation modes 103 for each data center 102 that can correspond to the workload prediction 109 for the time period selected when step 1802 was most recently executed has not been found.) In step 1809 in FIG. 18, the data center operation mode plan creation unit 1800 issues an alert to the user of the system 101, notifying that creation of the data center operation mode plan 110 has failed. After this, the user of the system 101 will deal with the alert.

[0096] (In response to the affirmative determination result in step 1807), in step 1810 of FIG. 18, the data center operation mode plan creation unit 1800 determines whether all of the records included in the record group (of the workload prediction table 900) acquired the most recent time step 1803 was executed have been selected in step 1805. If the determination result in step 1810 is affirmative, control transitions to step 1811. If the determination result in step 1810 is negative, control returns to step 1805, and one of the records that has not yet been selected is newly selected.

[0097] (When step 1811 of FIG. 18 is reached, a combination of operation modes 103 for each data center 102 that can correspond to the workload prediction 109 for the time period selected when step 1802 was most recently executed has been found, so in step 1811 of FIG. 18, the data center operation mode plan creation unit 1800 stores information about the combination of operation modes 103 (operation mode numbers) for each data center 102 for the time period selected when step 1802 was most recently executed in the data center operation mode plan table 1000 shown in FIG. 10. In this case, the information about the combination of operation modes 103 (operation mode numbers) for each data center 102 is information that indicates the combination of operation modes 103 for each data center 102 that was selected when step 1804 was most recently executed.

[0098] 18, the data center operation mode plan creation unit 1800 determines whether all of the time periods that are time units for controlling the operation modes 103 in the data center operation mode plan 110 have been selected in step 1802. If the determination result in step 1812 is positive, creation of the data center operation mode plan 110 is complete, so control is returned to step 1801, and the processing of the data center operation mode plan creation unit 1800 is essentially on standby until the time comes to create the next data center operation mode plan 110. If the determination result in step 1812 is negative, control is returned to step 1802, and one of the time periods that has not yet been selected is newly selected.

[0099] 6-4. Processing of Data Center Operation Mode Control Information Transmitter (Fig. 5, Fig. 10) Information on the setting of the operation mode 103 for each time period and each data center 102, which is stored in the data center operation mode plan table 1000, is transmitted to each data center 102. The data center operation mode control information transmission unit 510 is responsible for generating and transmitting the information used for this transmission (data center operation mode control information 210 (DC-mode)). 10, the data center operation mode control information transmitter 510 may create data center operation mode control information 210 (DC-mode) addressed to each of the data centers 102 that are targets of the data center operation mode plan 110. Each piece of data center operation mode control information 210 (DC-mode) may include identification information of the base 202 that is the destination of the data center operation mode control information 210 (DC-mode), identification information (data center identifier (DC-ID)) of the destination data center 102, identification information of the time period, and the value (operation mode number) of the operation mode 103 (set for the time period). The data center operation mode control information transmission unit 510 may create data center operation mode control information 210 (DC-mode) for each combination of data center 102 and time period that is the destination of the data center operation mode control information 210 (DC-mode), and then transmit the data center operation mode control information 210 (DC-mode). Alternatively, the data center operation mode control information transmitter 510 may create one piece of data center operation mode control information 210 (DC-mode) for the destination data center 102 by summarizing some time periods or all time periods (targeted by the data center operation mode plan 110), and then transmit the data center operation mode control information 210 (DC-mode). In this case, one piece of data center operation mode control information 210 (DC-mode) will include multiple sets of time period identification information and values ​​(operation mode numbers) of the operation mode 103 (set for the time period). Furthermore, the data center operation mode control information transmission unit 510 may create a single piece of data center operation mode control information 210 (DC-mode) for multiple data centers 102 included in the same base 202, and then transmit the data center operation mode control information 210 (DC-mode). In this case, one piece of data center operation mode control information 210 (DC-mode) will include multiple pieces of identification information (data center identifiers (DC-IDs)) for the destination data centers 102, and will further include, for each data center 102, one or more sets of time zone identification information and values ​​(operation mode numbers) for the operation modes 103 (set for the time zone). The process performed by the data center operation mode control information transmission unit 510 may be considered to form a "data center operation mode control information transmission step."

[0100] The data center operation mode control information transmission unit 510 realizes the functions described above, so that the information on the setting of the operation mode 103 for each time period and each data center 102 stored in the data center operation mode plan table 1000 can be transmitted to each data center 102. Then, each data center 102 can operate in the operation mode 103 in accordance with the data center operation mode plan 110.

[0101] 6-5. Workload Execution Request Receiving Process (Figure 5, Figure 11) A workload execution request 211 from the execution request device 105 is accepted by the system 101, a record containing information about the workload execution request 211 is generated, and the generated record is stored in the workload execution request buffer table 1100. The role of this acceptance, generation, and storage is played by the workload execution request receiving unit 511. As already explained, the workload execution request buffer table 1100 may store the record shown in FIG. 11 for each workload execution request 211 accepted by the system 101. Of the information included in the record shown in FIG. 11, the workload identification information (WL identification number (WL-ID)) may be assigned by the workload execution request receiving unit 511 itself or by another functional unit of the system 101. (If the workload identification information (WL identification number (WL-ID)) is assigned in advance by the execution request device 105 that is the issuer of the workload execution request 211, this previously assigned information may be reused within the system 101.) Of the information contained in the record shown in Figure 11, the date and time when the workload execution request 211 was received (WL execution request reception date and time) may be assigned by the workload execution request receiving unit 511 itself or by other functional units of the system 101. Of the information contained in the record shown in Figure 11, the constraints and requirements imposed on the data center 102 that is the processing entity when the workload is processed, the content of the IT resources used when processing the workload (mainly the amount of IT resources in the example of Figure 11), the time allowed when processing the workload (allowable execution delay time), and the time expected to be required when processing the workload (expected required time) may each be based on information explicitly assigned to the workload execution request 211 itself, or may be based on the result of a judgment arrived at by the workload execution request receiving unit 511 through some kind of judgment. After generating a record including information about the received workload execution request 211 , the workload execution request receiver 511 stores the record in the workload execution request buffer table 1100 . The processing performed by the workload execution request receiver 511 may be considered to form a "workload execution request receiving step."

[0102] The workload execution request receiving unit 511 realizes the functions described above, and as information regarding the workload execution request 211 received by the system 101, a record containing information that is easy for other functional units included in the system 101 to use can be generated and stored in the workload execution request buffer table 1100.

[0103] 6.6. Workload deployment setting process (Figure 19) Fig. 19 shows a flowchart of the processing executed by the workload deployment setting unit 1900. The processing will be described below in the order shown in Fig. 19. Note that each of the processing steps executed by the workload deployment setting unit 1900 in the flowchart of Fig. 19 may be understood to form a "workload deployment setting step." The functions described below are realized, so that the conditions for processing the workload associated with the workload execution request 211 received by the system 101 can be matched with the operation mode 103 for each data center 102 based on the data center operation mode plan 110, and settings can be made to appropriately deploy the workload to one of the data centers 102.

[0104] 19, the workload deployment setting unit 1900 determines whether a record for a new workload execution request 211 has been stored in the workload execution request buffer table 1100. If the determination result in step 1901 is affirmative, control transitions to step 1902. If the determination result in step 1901 is negative, step 1901 is repeated.

[0105] 19, the workload deployment setting unit 1900 acquires information about the new record that was the subject of determination in step 1901 from the workload execution request buffer table 1100. Based on the information about the acquired record, the workload deployment setting unit 1900 ascertains the constraints and requirements imposed on the data center 102 that will be the processing subject when the workload is processed (in the example of FIG. 11, the data center identifier (DC-ID) of the "data center where the workload can be deployed" and the degree (level) of availability required when the workload is processed), the time allowed for delay when processing the workload (allowable execution delay time), the details of the IT resources used when processing the workload (mainly the amount of IT resources in the example of FIG. 11), and the time expected to be required when the workload is processed (expected required time).

[0106] In step 1903 of FIG. 19 , the workload deployment setting unit 1900 identifies a time period during which the workload can be deployed (processed) based on the permissible execution delay time determined in step 1902. For example, if the permissible execution delay time is "immediate" (a setting that does not allow execution delay), the time period during which the workload can be deployed (processed) may be limited to the time period including the time when step 1903 is executed. (Alternatively, if the time is close to the end of the time period, the next time period may be used.) Furthermore, if the permissible execution delay time is set to, for example, "up to two hours later," the time period during which the workload can be deployed (processed) may be multiple time periods that include the time interval from the time when step 1903 is executed (or the time when the workload execution request 211 is generated or accepted) to two hours later. The permissible execution delay time may be set in either hours or minutes. Alternatively, the permissible execution delay time may be set to "until a specific time."

[0107] In step 1904 of Figure 19, the workload deployment setting unit 1900 identifies (one or more) data centers 102 within the range of one or more time periods identified in step 1903 that satisfy the constraints and requirements (in the example of Figure 11, the data center identifier (DC-ID) of the "deployable data center" and the degree (level) of availability required when processing the workload) imposed on the data center 102 that will be the processing entity when processing the workload, as identified in step 1902. Note that an operation mode 103 is set for each data center 102 for each time period. Therefore, the degree (level) of availability that can be achieved when processing a workload may differ for each data center 102 for each time period. Therefore, the workload deployment setting unit 1900 may refer to the data center operation mode plan table 1000 to determine the setting of the operation mode 103 for each time period and each data center 102, and then refer to the data center operation mode list table 700 to determine the degree (level) of availability that can be achieved when processing a workload according to the determined operation mode 103 (and combination of data center 102).

[0108] 19, the workload deployment setting unit 1900 selects one of the combinations of the time slot identified in step 1903 and the data center 102 identified in step 1904. (Note that the combination is limited to those that are approved in step 1904.) In step 1906 of Figure 19, the workload deployment setting unit 1900 obtains information regarding the combination of the time period and data center 102 selected in step 1905 from each of the data center list table 600, the data center operation mode list table 700, the data center operation mode plan table 1000, and the workload execution history table 1300. Specifically, the workload deployment setting unit 1900 may obtain information on the IT resources held by the data center (in the example of Figure 6, the number of central processing units (CPUs), the number of graphics processing units (GPUs), and the memory capacity) from the record related to the data center 102 related to the combination in the data center list table 600. The workload deployment setting unit 1900 may acquire the value (operation mode number) of the operation mode 103 from the record relating to the data center 102 and time period related to the combination in the data center operation mode plan table 1000. The workload deployment setting unit 1900 may obtain information on the ratio of the content (quantity) of IT resources that can be provided for processing the workload to the content (total quantity) of IT resources held by the data center 102 from records in the data center operation mode list table 700 relating to the combination of data center 102 and operation mode 103 (during the time period). The workload deployment setting unit 1900 may obtain information regarding the execution history and execution status (e.g., information regarding the amount and time of allocated IT resources) of other workloads that have already been deployed to the data center 102 from the group of records related to the data center 102 related to the combination in the workload execution history table 1300.

[0109] In step 1907 of Figure 19, the workload deployment setting unit 1900 calculates, based on the information obtained in step 1906, for the combination of the time period and data center 102 selected in step 1905, the content (e.g., quantity; specifically, the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity) of IT resources that can be provided to process the workload in the data center 102 and the portion of the time that has not been allocated to other workloads (unassigned portion (free portion)). Specifically, in step 1906, based on the information obtained from the data center list table 600, the information obtained from the data center operation mode plan table 1000, and the information obtained from the data center operation mode list table 700, the workload deployment setting unit 1900 can identify the content (e.g., quantity; specifically, the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity) and time of the IT resources that can be provided to process the workload in the data center 102. In addition, in step 1906, based on the information obtained from the workload execution history table 1300, the workload deployment setting unit 1900 can identify the content and time of IT resources available to process the workload in the data center 102 that have already been allocated to other workloads. Therefore, the workload deployment setting unit 1900 can calculate the unallocated portion (free portion) described above.

[0110] 19, the workload deployment setting unit 1900 determines whether the unallocated portion (free portion) calculated in step 1907 is equal to or greater than the amount indicated by the combination of the content (quantity) of IT resources used to process the workload associated with the workload execution request 211 accepted by the system 101 and the estimated required time. If the determination result in step 1908 is affirmative, control transitions to step 1911. If the determination result in step 1908 is negative, control transitions to step 1909.

[0111] In step 1909 of FIG. 19 (in response to the fact that the workload associated with the workload execution request 211 cannot be deployed to the combination of time slot and data center 102 selected in the most recent step 1905), the workload deployment setting unit 1900 determines whether all of the combinations of time slot and data center 102 identified in steps 1903 and 1904 have been selected in step 1905. If the determination result in step 1909 is positive, control transitions to step 1910. If the determination result in step 1909 is negative, control returns to step 1905, and one of the combinations of time slot and data center 102 that have not yet been selected is newly selected. (Upon finding that the workload associated with the workload execution request 211 cannot be deployed for all of the combinations of the time slot and the data center 102 identified in steps 1903 and 1904), in step 1910 of FIG. 19 , the workload deployment setting unit 1900 issues an alert to the effect that the deployment of the workload associated with the workload execution request 211 has failed. This alert may be notified to a user of the system 101. Also, this alert may be notified to the execution request device 105 that is the issuer of the workload execution request 211. Alternatively, this alert may be notified to the result utilization device 104 that was scheduled to utilize the workload associated with the workload execution request 211. Upon receiving this alert, the execution request device 105 may resend the workload execution request 211 or reissue the workload execution request 211 after changing the content of the workload to be associated with the workload execution request 211. After step 1910, control may be returned to step 1901.

[0112] 19, the workload deployment setting unit 1900 stores a record indicating that part or all of the unallocated portion (free space) calculated in step 1907 has been allocated to the workload in the workload deployment setting table 1200. Specifically, the workload deployment setting unit 1900 creates a record indicating that the content (quantity, for example, quantity; specifically, the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity) of IT resources to be used for processing the workload associated with the workload execution request 211 is allocated to the combination of the time slot and data center 102 (selected in the most recent step 1905) corresponding to the unallocated portion (free space), and stores the record in the workload deployment setting table 1200 shown in FIG. After step 1911, the control returns to step 1901, and the processing of the workload deployment setting unit 1900 essentially goes into a waiting state until the next new workload execution request 211 arrives.

[0113] 6-7. Processing of the Workload Placement Control Information Transmitter (Fig. 5, Fig. 12, Fig. 13) The information on the workload deployment setting 112 to the data center stored in the workload deployment setting table 1200 is transmitted to the data center 102 to which the workload is to be deployed. The workload deployment control information transmitter 512 is responsible for generating and transmitting the information used for this transmission (workload deployment control information 212 (WL-deploy)). The workload deployment control information transmitter 512 may generate workload deployment control information 212 (WL-deploy) addressed to the data center 102 where the workload is to be deployed, based on information stored in a workload deployment setting table 1200 shown in Fig. 12. The workload deployment control information 212 (WL-deploy) may include information such as identification information of the base 202 that is the destination of the workload deployment control information 212 (WL-deploy), identification information (data center identifier (DC-ID)) of the destination data center 102, identification information of the workload (workload identification number (WL-ID)), information related to the execution start time or scheduled execution start time of the workload processing, and details of IT resources used to process the workload (in the example of Fig. 12, the number of central processing units (CPUs), the number of graphic processing units (GPUs), and memory capacity). The workload deployment control information transmitter 512 transmits the generated workload deployment control information 212 (WL-deploy) to the destination data center 102 . The process performed by the workload placement control information transmission unit 512 may be considered to form a "workload placement control information transmission step."

[0114] In addition to sending the workload deployment control information 212 (WL-deploy) to the data center 102, the workload deployment control information transmission unit 512 may store a record in the workload execution history table 1300 indicating that the workload is being deployed to the data center 102. 13 , the workload deployment control information transmitter 512 may store a new record in the workload execution history table 1300, including information on the identification information assigned to the workload (workload identification number (WL-ID)), the identification information (data center identifier (DC-ID)) of the data center 102 where the workload was actually deployed (actual deployment destination data center), the time when processing of the workload will start or the time when processing is scheduled to start (WL execution start time or scheduled WL execution start time), and the number of central processing units (CPUs), the number of graphics processing units (GPUs), and the memory capacity that the deployment destination data center 102 will allocate to process the workload. The execution state of the workload (WL execution state) in the new record may be “running” if processing of the workload will start immediately, or “waiting for execution” if processing of the workload will be performed later.

[0115] The workload deployment control information transmitter 512 realizes the functions described above, so that information contained in a record related to a workload associated with a workload execution request 211, which is stored in the workload deployment setting table 1200, can be transmitted to the data center 102 to which the workload is to be deployed. Then, the destination data center 102 can process the workload associated with the workload execution request 211 after it has been deployed. Furthermore, the fact that the workload is deployed to the data center 102 is reflected in the workload execution history table 1300, so that the system 101 can correctly grasp information related to the execution history and execution status of the workload in each of the data centers 102.

[0116] 6-8. Workload Execution History Information Reception Unit Processing (Figure 5, Figure 13) The data center 102 to which a workload has been deployed may process the workload and transmit information about the workload's execution history and execution status to the system 101 as workload execution history information 213 (WL-log). After the workload execution history information 213 (WL-log) is received by the system 101, the information contained in the workload execution history information 213 (WL-log) is reflected in the workload execution history table 1300. The role of this reception and reflection is played by the workload execution history information receiving unit 513. For example, if the received workload execution history information 213 (WL-log) indicates that processing of one of the workloads has ended (completed) in one of the data centers 102, the workload execution history information receiver 513 extracts each piece of information included in the workload execution history information 213 (WL-log), such as identification information assigned to the workload (workload identification number (WL-ID)) and the end time (completion time) of the workload processing. Then, the workload execution history information receiver 513 identifies a record having the extracted workload identification number (WL-ID) from the workload execution history table 1300. The workload execution history information receiver 513 stores information on the end time (completion time) of the extracted workload processing in the identified record, and changes the workload execution state (WL execution state) in the identified record to "completed." The processing performed by the workload execution history information receiving unit 513 may be considered to form a "workload execution history information receiving step."

[0117] The workload execution history information receiving unit 513 realizes the functions described above, so that, for example, when the workload execution history information 213 (WL-log) indicating that the processing of the workload has ended (completion) is transmitted from the data center 102 to the system 101, the information indicated by the workload execution history information 213 (WL-log) can be reflected in the workload execution history table 1300. In other words, the system 101 can correctly grasp information related to the execution history and execution status of the workload in each of the data centers 102.

[0118] 6.9. Workload redistribution setting process (Figure 20) Fig. 20 shows a flowchart of the processing executed by the workload reallocation setting unit 2000. The processing will be described below in the order shown in Fig. 20. Note that each of the processing steps executed by the workload reallocation setting unit 2000 in the flowchart of Fig. 20 may be considered to form a "workload reallocation setting step." As a result of realizing the functions described below, when the operation mode 103 of any of the data centers 102 is changed in accordance with the data center operation mode plan 110, if an inconsistency occurs between the changed operation mode 103 of the data center 102 and the "conditions" when a workload deployed in the data center 102 is processed, the workload can be redeployed (rebalanced, migrated) to another data center 102 where the inconsistency does not occur. In this way, it is possible to appropriately harmonize the control of each facility of the data center 102 based on the workload prediction 109 (control by setting the operation mode 103) and the control of deployment (deployment) and redeployment (rebalancing, migration) in accordance with the "conditions" when a workload associated with an actually accepted workload execution request 211 is processed.

[0119] 20, the workload reallocation setting unit 2000 determines whether the timing for switching time slots in the data center operation mode plan 110 has arrived. For example, if each time slot is one hour long (the length of the time slot is one hour) starting from 0 minutes and 0 seconds every hour, the workload reallocation setting unit 2000 may determine that the timing for switching time slots is 0 minutes and 0 seconds every hour, and that the timing for indicating a positive determination in step 2001 is a predetermined time before the switching timing (for example, 58 minutes and 0 seconds every hour (2 minutes before the switching)). If the determination result in step 2001 is positive, control transitions to step 2002. If the determination result in step 2001 is negative, step 2001 is repeated.

[0120] In step 2002 of Figure 20, the workload redeployment setting unit 2000 selects one of the records in the workload execution history table 1300 in which the workload execution state (WL execution state) is (expected to be) "running" across a time zone change (for example, 0 minutes and 0 seconds of every hour). The workload redeployment setting unit 2000 acquires information about the selected record. The workload redeployment setting unit 2000 grasps the workload identification information (workload identification number) included in the information about the acquired record.

[0121] 20, the workload reallocation setting unit 2000 identifies a record including identification information (workload identification number) of the workload identified in step 2002 from the workload execution request buffer table 1100. The workload reallocation setting unit 2000 acquires information of the identified record from the workload execution request buffer table 1100. 20, the workload reallocation setting unit 2000 identifies the constraints and conditions imposed on the data center 102 that will be the processing subject when processing the workload, which are included in the information of the record acquired from the workload execution request buffer table 1100 in step 2003. In accordance with the example of Fig. 11, the constraints and conditions imposed on the data center 102 that will be the processing subject when processing the workload include the data center identifier (DC-ID) of the "data center where the workload can be deployed" and the degree (level) of availability required when processing the workload.

[0122] 20, the workload redeployment setting unit 2000 obtains the identification information (data center identifier (DC-ID)) of the data center where the workload was actually deployed, which is included in the information of the record in the workload execution history table 1300 acquired in step 2002. This identification information (data center identifier (DC-ID)) of the data center where the workload was actually deployed indicates the data center 102 where the workload was deployed before the workload was redeployed (rebalanced or migrated). In step 2006 of FIG. 20 , the workload reallocation setting unit 2000 identifies the operation mode 103 to be set in the actual deployment data center identified in step 2005 (the data center 102 where the workload was deployed before the workload was redeployed (rebalanced or migrated)), after the time zone is changed. The workload reallocation setting unit 2000 also identifies the degree (level) of availability that can be achieved for processing the workload, based on the identified operation mode 103. To perform the processing of step 2006, the workload reallocation setting unit 2000 may obtain information for identifying the operation mode 103 from the data center operation mode plan table 1000, and may also obtain information for identifying the degree (level) of availability that can be achieved for processing the workload, from the data center operation mode list table 700.

[0123] In step 2007 of FIG. 20 , the workload reallocation setting unit 2000 determines whether the constraints and conditions, determined in step 2004, imposed on the data center 102 that will be the processing subject when processing the workload after the time zone change are satisfied by the actual deployment destination data center determined in step 2005 (the data center 102 to which the workload was deployed before the workload was redeployed (rebalanced or migrated)). Of the constraints and conditions imposed on the data center 102 that will be the processing subject when processing the workload, the constraint of "data center available for deployment" usually does not change when the time zone change occurs. Therefore, the determination in step 2007 may essentially be a comparison between the level of availability required when processing the workload determined in step 2004 and the level of availability that can be achieved for processing the workload after the time zone change determined in step 2006. If the determination result in step 2007 is positive, there is no need to redeploy (rebalance, migration) the workload associated with the record most recently selected in step 2002, and control transitions to step 2008. If the determination result in step 2007 is negative, control transitions to step 2009 to perform redeployment (rebalance, migration) of the workload associated with the record most recently selected in step 2002.

[0124] 20, the workload reallocation setting unit 2000 determines whether all records in the workload execution history table 1300 whose workload execution states (WL execution states) are (expected to be) "running" across time zone changes (for example, at 0 minutes and 0 seconds of each hour) have been selected in step 2002. If the determination result in step 2008 is positive, control is returned to 2001, and the processing of the workload reallocation setting unit 2000 essentially enters a standby state until the next time zone changes. If the determination result in step 2008 is negative, control is returned to step 2002, and a record that has not yet been selected is newly selected.

[0125] In step 2009 of FIG. 20 , the workload reallocation setting unit 2000 identifies a data center 102 that satisfies the constraints and requirements imposed on the data center 102 that will be the processing subject when processing the workload, as determined in step 2004, after the time zone change, and that can be allocated the content (e.g., quantity, such as the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity) of IT resources to be used when processing the workload and time. In performing the processing of step 2009, the workload reallocation setting unit 2000 may acquire information on the content (e.g., quantity, such as the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity) of IT resources possessed by each of the data centers 102 from the data center list table 600. In performing the processing of step 2009, the workload reallocation setting unit 2000 may acquire information on the operation mode 103 to be set in each of the data centers 102 after the time zone change from the data center operation mode plan table 1000. In performing the processing of step 2009, the workload reallocation setting unit 2000 may acquire information on the degree (level) of availability that can be achieved when processing a workload and information on the ratio of the content (quantity) of IT resources that can be provided for processing a workload out of the content (quantity) of possessed IT resources from the data center operation mode list table 700, which correspond to the operation mode 103 to be set in each of the data centers 102 after the time zone change. When performing the processing of step 2009, the workload redeployment setting unit 2000 may obtain information from the workload execution history table 1300 regarding the status of IT resource allocation for workloads to be deployed in each of the data centers 102 after the time zone change.

[0126] 20, the workload redeployment setting unit 2000 creates a record indicating that, in response to a change in time zone, the destination data center for the workload associated with the workload identification number (WL-ID) of the record acquired most recently when step 2002 was executed will be changed (migrated). The workload redeployment setting unit 2000 stores the created record in the workload redeployment setting table 1400. In accordance with the example of FIG. 14, the created record includes information such as identification information (workload identification number (WL-ID)) assigned to the workload, identification information (data center identifier (DC-ID)) of the destination data center before redeployment (currently deployed data center), identification information (data center identifier (DC-ID)) of the destination data center for redeployment, the scheduled time of redeployment (scheduled redeployment time), and details of IT resources (mainly the amount of IT resources in the example of FIG. 14) to be allocated by the destination data center 102 for processing the workload. Here, the content of the IT resources allocated by the redeployment destination data center 102 for processing the workload (mainly the amount of IT resources in the example of Figure 14) may include information such as the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity. In step 2010, the workload reallocation setting unit 2000 may modify the contents of the workload reallocation setting table 1200 as necessary. In accordance with the example of Fig. 12, in the record for the workload that is the target of reallocation (rebalancing, migration), the data center identifier (DC-ID) of the data center where the workload was actually deployed may be modified to indicate the data center 102 where the workload was reallocated.

[0127] In step 2011 of Figure 20, the workload reallocation setting unit 2000 determines whether all records in the workload execution history table 1300 whose workload execution states (WL execution states) are (expected to be) "running" across time zone changes (for example, 0 minutes and 0 seconds of every hour) have been selected in step 2002. If the determination result in step 2011 is positive, control is returned to 2001, and the processing of the workload reallocation setting unit 2000 essentially enters a standby state until the next time when the time zone changes. If the determination result in step 2011 is negative, control is returned to step 2002, and a record that has not yet been selected is newly selected.

[0128] 6-10. Processing of the Workload Reallocation Control Information Transmission Unit (Fig. 5, Fig. 13, Fig. 14) The information to change the data center to which a workload is deployed, which is stored in the workload redeployment setting table 1400, may be transmitted to both the data center 102 that was the deployment destination before the redeployment and the redeployment destination data center 102. The workload redeployment control information transmitter 514 is responsible for generating and transmitting the information used for this transmission (workload redeployment control information 214 (WL-migration)). The workload redeployment control information transmitter 514 may generate workload redeployment control information 214 (WL-migration) addressed to both the data center 102 that was the deployment destination before the redeployment and the redeployment destination data center 102, based on information stored in a workload redeployment setting table 1400 shown in Fig. 14. The workload redeployment control information 214 (WL-migration) may include information such as identification information of (one or more) locations 202 that are the destination of the workload redeployment control information 214 (WL-migration), identification information (data center identifiers (DC-IDs)) of the destination data centers 102, identification information of the workload (workload identification numbers (WL-IDs)), information about the scheduled time of the workload redeployment, and details of IT resources used to process the workload in the redeployment destination data center 102 (in the example of Fig. 14 , the number of central processing units (CPUs), the number of graphics processing units (GPUs), and memory capacity). The workload redeployment control information transmitter 514 transmits the generated workload redeployment control information 214 (WL-migration) to the destination data center(s) 102. The process performed by the workload reallocation control information transmitter 514 may be considered to form a "workload reallocation control information transmitting step."

[0129] In addition to transmitting the workload redeployment control information 214 (WL-migration) to the (multiple) data centers 102, the workload redeployment control information transmitter 514 may also make changes to the records in the workload execution history table 1300 to indicate that the data center 102 to which the workload is deployed will be changed. In accordance with the example of Figure 13, the workload redeployment control information transmission unit 514 may change the data center identifier (DC-ID) of the actual deployment destination data center for the record in the workload execution history table 1300 for the workload that has been subject to redeployment (rebalancing, migration) to indicate the redeployment destination data center 102.

[0130] The workload redeployment control information transmitter 514 realizes the functions described above, so that information about changing the data center 102 to which a workload is to be redeployed, which is stored in the workload redeployment setting table 1400, can be transmitted to both the data center 102 that was the destination before the redeployment and the data center 102 to which the workload is to be redeployed. The destination data center (102) can then operate to realize the redeployment (rebalancing, migration) of the workload. In addition, the fact that the workload is being redeployed (rebalancing, migration) is reflected in the workload execution history table 1300, so that the system 101 can correctly grasp information about the execution history and execution status of the workload in each of the data centers 102.

[0131] 7. Other (variations) The present disclosure is not limited to the above-described embodiments and includes various modifications. Part of the configurations and processes of the embodiments may be replaced with the configurations and processes of other conceivable embodiments. The configurations and processes of other conceivable embodiments may be added to the configurations and processes of the embodiments. For example, the present disclosure may include the following modified embodiments.

[0132] (Variation A) Distributed processing of workload deployment settings The above mainly describes a case in which the system 101 centrally controls the allocation (deployment) of workloads to information processing resources. However, the system 101 does not have to centrally control the allocation (deployment) of workloads to information processing resources. For example, a workload execution request 211 from any of the execution request devices 105 in FIG. 1 may be received by any of the data centers 102 without going through the system 101. The data center 102 that receives the workload execution request 211 may determine whether it can process the workload associated with the workload execution request 211. If it cannot process the workload, the data center 102 may transfer the workload execution request 211 to a data center 102 that can (or has the potential to) process the workload associated with the workload execution request 211. In this way, it is possible to realize distributed control when allocating (deploying) workloads to information processing resources. In this case, the system 101 does not need to include the workload deployment setting unit 1900. In the above-described modified example, the functional configuration of the system 101 can be made simpler.

[0133] (Variation B) Variation of Conditioning When Processing a Workload In the above, the conditions for processing a workload include the constraint of "available data centers," as well as the requirements for the level of availability required for processing the workload, the content (mainly quantity) of information processing resources (IT resources) (e.g., number of central processing units (CPUs), number of graphics processing units (GPUs), memory capacity), the expected time required, and the allowable execution delay time. In variations, the conditioning on processing a workload may include only some of those listed above, or may include more than those listed above. In a variant, the conditions under which the workload is processed can be flexibly set.

[0134] (Variation C) Variation of "Deployable Data Center" In the above, the "deployable data center", that is, the data center 102 to which the workload can be deployed if other conditions (constraints, requirements, amount of allocable free IT resources, etc.) are met, is determined by the geographical relationship between the result utilization device 104, which is a device that utilizes the results of processing the workload associated with the workload execution request 211, and the data center 102 (for example, a relationship such as physical distance that allows personnel to travel to the location 202 where the data center 102 is located in the event of a failure in the data center 102), or the connection relationship on the network 299 (for example, a relationship indicating the degree of proximity on the network topology of the network 299 or the degree of delay time in information transmission on the network 299). In the example of Figure 2, if the result consumption device 104-U-1-1-1-1 is a device that consumes the results of processing a workload associated with the workload execution request 211, the group of data centers 102 at the edge (regional) site 202-E-1-1 in the region where the result consumption device 104-U-1-1-1-1 is located, the group of data centers 102 at the regional site 202-R-1 installed in the region where the result consumption device 104-U-1-1-1-1 is located, and the group of data centers 102 at the core site 202-C can be "deployable data centers," while the group of data centers 102 at the edge (regional) site 202-E-1-2 for another region and the group of data centers 102 at the regional site 202-R-2 for another region are not necessarily "deployable data centers." By setting the "available data centers" as described above, it is possible to narrow down the targets for workload deployment to data centers 102 that are in good geographical conditions or have good connectivity on the network 299, from the perspective of the result utilization device 104, which is a device that utilizes the results of workload processing associated with the workload execution request 211. This can be expected to shorten the time it takes to respond to the results of workload processing, for example. However, the setting of the "deployable data center" may be more flexible. For example, not only the district or region to which the result consumption device 104, which is a device that consumes the results of the processing of the workload associated with the workload execution request 211, belongs, but also the data center 102 located in the base 202 associated with any of multiple districts or multiple regions may be set as the "deployable data center." In the modified example, the "data centers that can be deployed" can be set flexibly, which widens the options for the data centers 102 in which workloads can be deployed.

[0135] (Variation D) Management of performance information of information processing resources (IT resources) In the embodiment described above, the numbers and amounts of information processing resources (IT resources) possessed by each data center 102 are explicitly managed, as shown in the data center list table 600 in Fig. 6, the workload result table 800 in Fig. 8, the workload forecast table 900 in Fig. 9, the workload execution request buffer table 1100 in Fig. 11, the workload deployment setting table 1200 in Fig. 12, the workload execution history table 1300 in Fig. 13, and the workload redeployment setting table 1400 in Fig. 14. For example, in the figures pointed out above, the number of central processing units (CPUs), the number of graphics processing units (GPUs), and the memory capacity possessed by the data center 102 are explicitly managed.

[0136] In a modified example, in addition to explicitly managing the number and amount of information processing resources (IT resources) possessed by each of the data centers 102, the performance of the IT resources possessed by each of the data centers 102 may also be explicitly managed. For example, one or more of the generation indicating the performance of the central processing unit (CPU), the generation indicating the performance of the graphics processing unit (GPU), the access speed indicating the performance of the memory possessed by each of the data centers 102, etc. may be managed in each of the figures indicated above. FIG. 21 shows a modified data center list table 2100, which is a data center list table in a modified example in which generations indicating the performance of central processing units (CPUs) and generations indicating the performance of graphics processing units (GPUs) are managed. In FIG. 21, the generation indicating the performance of central processing units (CPUs) is represented as "Gc." Also, in FIG. 21, the generation indicating the performance of graphics processing units (GPUs) is represented as "Gg." As shown in FIG. 21, for one or both of the central processing units (CPUs) and graphics processing units (GPUs) possessed by each data center 102, information indicating the generation in addition to the number may be managed. (Although not explicitly shown in FIG. 21, information indicating memory access speed, etc. may also be managed.) As in Fig. 21, which is a modification of Fig. 6, in Figs. 8, 9, 11, 12, 13, and 14, information indicating the generation of one or both of the central processing units (CPUs) and the graphics processing units (GPUs) may be managed in addition to the number of units. (Information indicating the memory access speed, etc. may also be managed.) In such a variant, when creating a workload forecast 109, creating a data center operation mode plan 110, configuring a workload deployment to a data center 112, and further redeploying a workload to a data center, not only the number and amount of information processing resources (IT resources) but also one or more of the performance of the information processing resources (IT resources) (e.g., the generation of the central processing unit (CPU), the generation of the graphics processing unit (GPU), the memory access speed, etc.) may be taken into consideration. According to the above modification, it is possible to realize more appropriate workload prediction 109, data center operation mode plan 110, and workload deployment setting 112 to the data center, taking into account the performance of information processing resources (IT resources).

[0137] (Variation E) Variation that does not use the expected time required for processing the workload In the embodiment described above, the workload execution request receiving unit 511 acquires information about the expected time required for processing the workload, or determines the expected time required, and then stores the information about the expected time required in a record of the workload execution request buffer table 1100. Then, the workload deployment setting unit 1900 obtains the information about the expected time required in step 1902 of Fig. 19, and then uses the information about the expected time required in step 1908 of Fig. 19 to determine whether the workload can be deployed to the data center 102.

[0138] In a modified example, the expected required time for processing the workload may not be used. For example, depending on the workload execution request 211 received by the system 101, it may be difficult to determine the expected required time for processing the workload indicated by the workload execution request 211. Therefore, in a modified example, the workload execution request receiving unit 511 may not handle information related to the expected required time and may not store information related to the expected required time in the record of the workload execution request buffer table 1100. Accordingly, the workload deployment setting unit 1900 may not grasp information related to the expected required time. In this case, the workload deployment setting unit 1900 may tentatively determine the expected required time and make a determination in a step similar to step 1908 of FIG. 19 . Alternatively, the workload deployment setting unit 1900 may make a determination in a step similar to step 1908 of FIG. 19 without taking time into consideration. According to this modified example, even in cases where it is difficult to determine the expected time required for processing the workload indicated by the workload execution request 211, the deployment setting 112 of the workload to the data center can be realized.

[0139] (Variation F) Variation in which the notation of deployable data centers is replaced with the notation of bases In the embodiments described above, in the records of the workload actual result table 800 in FIG. 8, the workload forecast table 900 in FIG. 9, and the workload execution request buffer table 1100 in FIG. 11, the "data centers where the data can be deployed" were directly indicated by the data center identification information (DC-ID) of each of the data centers 102.

[0140] In a modified example, the "deployable data center" may be indirectly indicated by the identification information of the base 202 where the data center 102 is located. For example, if the group of data centers 102 located at the regional base 202-R-1 in Fig. 2 are included in the "deployable data centers," the "deployable data center" may be represented by "R-1," which is the identification information of the regional base 202-R-1, instead of being represented by "R-1-1," "R-1-2," etc., which are the data center identification information (DC-ID) of the individual data centers 102 such as data center 102-R-1-1, data center 102-R-1-2, etc. Whether or not each data center 102 is a "deployable data center" is determined by geographical conditions, etc., and in principle, in a case where a data center 102 that is a "deployable data center" and a data center 102 that is not a "deployable data center" do not coexist at the same location 202, it is possible to express the "deployable data center" using the identification information of the location 202. According to such a modification, the information indicating "available data centers" can be simplified. Furthermore, when users of system 101 view the information contained in the above-mentioned tables, the information indicating "available data centers" becomes easy to use (easy to understand when viewed).

[0141] The technical matters shown in the above-described embodiments of the present disclosure and the modified examples of the embodiments can be combined as appropriate as long as no technical contradiction occurs.

Claims

1. 1. A system comprising: a data center operation mode plan creation unit that creates a data center operation mode plan, which is a plan for an operation mode of each of the data centers, based on a prediction regarding a workload that will be requested to be processed at any of the data centers.

2. 2. The system of claim 1, a workload placement setting unit that, upon receiving a workload execution request requesting that a workload be processed in one of the data centers, determines the data center to which the workload associated with the received workload execution request will be placed, based on conditions under which the workload associated with the received workload execution request will be processed and the operation modes of each of the data centers defined by the data center operation mode plan.

3. 3. The system of claim 2, the system further comprising: a workload redeployment setting unit that, when the operation mode of one of the data centers is changed based on the data center operation mode plan and the operation mode of the data center after the change does not conform to the conditions under which the workload deployed in the data center is processed, determines another data center that conforms to the conditions as the data center to which the workload is redeployed.

4. 2. The system of claim 1, A system in which each of the operational modes of each of the data centers is associated with information regarding the degree of availability that can be achieved when the data center is in that operational mode and information regarding the contents of the information processing resources that the data center possesses and that can be provided to the workload when the data center is in that operational mode.

5. 5. The system of claim 4, Each of the operational modes of each of the data centers is further associated with information for controlling equipment associated with that data center to achieve the degree of availability.

6. 6. The system of claim 5, The system, wherein the information for controlling the equipment is one or more of control information regarding air conditioning for the data center, control information regarding a storage battery for the data center, and control information regarding an emergency generator for the data center.

7. 5. The system of claim 4, Each of the operation modes of each of the data centers is associated with information on values ​​of one or more types of indicators when the data center is in that operation mode; The data center operation mode plan creation unit creates the data center operation mode plan by using a combination of operation modes for which the index value is relatively good, from among combinations of operation modes of each of the data centers that can process each of the workloads indicated by the prediction regarding the workload.

8. 8. The system of claim 7, There are three types of indicators: a priority is set for each type of index to be taken into consideration when the data center operation mode plan creation unit determines a combination of the operation modes of each of the data centers; A system in which the indicators, in descending order of priority, are indices relating to costs, the proportion of electricity generated with relatively low carbon emissions, and adjustment compensation.

9. 3. The system of claim 2, The system, wherein the information indicating the conditions under which the workload is processed includes information identifying the data center in which the workload can be deployed if other conditions are met, information indicating the degree of availability required when processing the workload, and information indicating the content of information processing resources required when processing the workload.

10. 10. The system of claim 9, The data center is located at a core base, a regional base located in each region, or an edge base located in an area smaller than a region, a result consuming device that consumes a result of processing a workload associated with the workload execution request, the result consuming device being located within one of the areas; The system, wherein the information identifying the data center to which the workload can be deployed if other conditions are met is determined by a geographical or network connection relationship between the result utilization device for the workload and the data center.

11. 10. The system of claim 9, The system, wherein the information indicating the conditioning upon which the workload is processed further includes information indicating an amount of time a delay is allowed in processing the workload.

12. 2. The system of claim 1, The system includes a workload forecasting unit that generates the forecast for a workload based on a history of received workload execution requests that request the workload to be processed at any of the data centers.

13. 4. The system of claim 3, the system further includes a workload prediction unit, a data center operation mode control information transmission unit, a workload execution request reception unit, a workload deployment control information transmission unit, a workload execution history information reception unit, a workload reallocation control information transmission unit, and a workload performance table creation unit; The system further includes a data center list table, a data center operation mode list table, a workload forecast table, a data center operation mode plan table, a workload execution request buffer table, a workload deployment setting table, a workload execution history table, and a workload redeployment setting table.

14. A method performed by a system, comprising:

1. A method comprising: generating a data center operational mode plan, the data center operational mode plan being a plan for an operational mode of each of the data centers, based on a prediction regarding workloads requested to be processed at any of the data centers.

15. A program, The program includes: A program for causing a computer to execute a data center operation mode plan creation step of creating a data center operation mode plan, which is a plan for each operation mode of one of the data centers, based on a prediction regarding a workload requested to be processed in the data center.

Images