Device and method for creating it base construction plan

Abstract

To acquire a pareto optimum IT base construction plan in consideration of a plurality of indexes.SOLUTION: In an IT base construction plan creation method, a plurality of search candidates indicating application combinations selected from application candidate information is created, values of the plurality of indexes of the IT base construction plan indicating the application combination and a data center installation destination for deploying a work road of the application combination is calculated for each of the plurality of search candidates by referring to the application candidate information and the data center installation candidate information, and, out of IT base construction plans of the plurality of search candidates, a plurality of IT base construction plans that are not inferior to other IT base construction plans in the IT base construction plans are output.SELECTED DRAWING: Figure 7

Description

[Technical field]

[0001] The present invention relates to creating an IT infrastructure construction plan. [Background technology]

[0002] Background technology of this disclosure is Non-Patent Document 1. The optimization method in Non-Patent Document 1 finds a combination of data center installation locations from multiple DC installation location candidates that maximizes a single score calculated by weighting the initial investment, power cost, and power cost discount due to the use of renewable energy. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] C. Guo et al., "Optimal Placement of Cloudlets Considering Electric Power Communication Network and Renewable Energy Resource," 2019 IEEE International Conference on Smart Cloud (SmartCloud), Tokyo, Japan, 2019, pp. 199-203, doi: 10.1109 / SmartCloud.2019.00041. Summary of the Invention [Problem to be solved by the invention]

[0004] As a measure against climate change, the introduction of renewable energy (RE) is progressing rapidly along with the efficient use of energy. As a result, even when the proportion of renewable energy in the total power is low in the power grid, renewable energy is generated that cannot be fully utilized depending on the time of day or region. As one of the large consumers that consumes a large amount of power, data centers are required to provide the adjustment power to adjust the power consumption to match the aforementioned unused renewable energy, in addition to decarbonization. In this context, in the design of data centers, it is required not only to increase revenue to increase profits and reduce operating costs, but also to simultaneously improve multiple indicators such as efficient use of renewable energy and minimizing the use of power sources other than renewable energy.

[0005] In Non-Patent Document 1, the indices to be considered are limited, and since a single score is assigned, the weights between the indices must be predefined. As a result of giving priority to the indices with larger weights for optimization, there are cases where any of the indices cannot meet the requirements of the data center designer. [Means for solving the problem]

[0006] One aspect of the present invention is an apparatus for creating an IT infrastructure construction plan indicating data center installation locations and the workloads of applications to be deployed at the data center installation locations, the apparatus including a processor and a storage device, the storage device stores application candidate information and data center installation location candidate information, the application candidate information indicates information about multiple applications for calculating multiple preset indicators for the construction and operation of an IT infrastructure, and the data center installation location candidate information indicates information about multiple data center installation location candidate information for calculating the multiple indicators, the processor creates multiple search candidates indicating application combinations selected from the application candidate information, and for each of the multiple search candidates, refers to the application candidate information and the data center installation location candidate information to calculate values ​​of the multiple indicators of an IT infrastructure construction plan indicating the application combination and the data center installation location to deploy the workload of the application combination, and outputs multiple IT infrastructure construction plans for the IT infrastructure construction plans of the multiple search candidates that are not superior to other IT infrastructure construction plans in terms of the multiple indicators. Effect of the Invention

[0007] According to one aspect of the present invention, it is possible to determine an IT infrastructure construction plan taking into consideration multiple indicators at the time of construction and operation.

[0008] Configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating an example of a schematic configuration of a system according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram illustrating an example of hardware and functions of a planning server. [Diagram 3] FIG. 13 is a diagram illustrating an example of a DC installation location candidate table. [Figure 4] FIG. 13 illustrates an example of an application candidate table. [Diagram 5] FIG. 13 illustrates an example of a search history management table. [Figure 6]FIG. 13 illustrates an example of a scenario management table. [Figure 7] 13 is a flowchart illustrating an example of a DC and application configuration optimization process. [Figure 8] 13 is a flowchart illustrating an example of a search candidate determination process. [Figure 9] 13 is a flowchart illustrating an example of a scenario generation process. [Figure 10] 13 is a flowchart showing an example of a DC configuration and WL placement optimization process. [Figure 11] 13 is a flowchart illustrating an example of an index calculation and search history registration process. [Figure 12] FIG. 13 is a diagram showing an example of a planning result display screen. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] In the following, when necessary for convenience, the description will be divided into multiple sections or examples, but unless otherwise specified, they are not unrelated to each other, and one is related to the other as a partial or complete modification, detail, supplementary explanation, etc. Furthermore, in the following, when the number of elements, etc. (including the number, numerical value, amount, range, etc.) is mentioned, it is not limited to the specific number, and may be more or less than the specific number, unless otherwise specified or clearly limited in principle to a specific number, etc.

[0011] The system or device in this specification may be a physical computer system (one or more physical computers) or a system built on a group of computing resources (multiple computing resources) such as a cloud infrastructure. The computer system or the group of computing resources may include one or more interface devices (including, for example, a communication device and an input / output device), one or more storage devices (including, for example, a memory (main memory) and an auxiliary storage device), and one or more arithmetic devices.

[0012] When a function is realized by executing a program by a computing device, the defined process is performed using a storage device and / or an interface device, etc., as appropriate, and therefore the function may be at least a part of one or more computing devices. A process described using a function as the subject may be a process performed by a system including one or more computing devices.

[0013] The program may be installed from a program source. The program source may be, for example, a program distribution computer or a computer-readable storage medium (for example, a computer-readable non-transitory storage medium). The description of each function is an example, and multiple functions may be combined into one function, or one function may be divided into multiple functions.

[0014] In one embodiment of the present specification, an IT infrastructure plan for providing an application service is created and presented to a user. The IT infrastructure plan indicates a combination of one or more data center candidates and one or more application program type candidates. The plan includes information on the installation locations of the data center candidates.

[0015] In general, the efficiency of renewable energy utilization in urban data centers is low. In addition, as the amount of renewable energy generated increases, there is a demand for efficient use of renewable energy without waste and for reducing the load on the power grid. For this reason, for example, it is conceivable to install a small data center near a renewable power generation source in a rural area and decarbonize the data center by locally producing and consuming renewable energy. In one embodiment of the present specification, an IT infrastructure plan is created and presented based on such demands, and the load on the power grid can be reduced while decarbonizing the data center. EXAMPLES

[0016] FIG. 1 is a diagram showing an example of a schematic configuration of a system according to this embodiment. The system includes multiple data centers (DC) 160. In FIG. 1, as an example, only one data center (DC1) is indicated by reference numeral 160. The data center 160 includes multiple IT devices including a server device 165, a storage device 166, and a network (NW) device. The server device 165 executes one or multiple application programs (apps) 161 that are deployed. The data center 160 includes an execution environment (not shown) of the app 161.

[0017] Each application 161 provides a specific service, such as a virtual desktop service or a machine learning service, to a user. Each application 161 executes a workload (application workload (WL)) 162, which is an assigned process. Each application 161 executes the assigned application workloads in sequence.

[0018] The data center 160 includes, in addition to IT equipment, facilities necessary to function as a data center, such as air conditioning equipment. In the data center 160, in addition to the IT equipment, the facilities for the IT equipment consume power. In the embodiment of the present specification, the data center 160 may take various forms. For example, the data center 160 may be a container-type data center, a data center of a specific scale (e.g., small or medium scale), or may take a form in which services from the data center are intermediated to users (DC in DC). The container-type data center may be a portable type or a non-portable type.

[0019] The data center 160 is supplied with power from a power distribution network 173. A plurality of renewable energy generating sources 171 are connected to the power distribution network 173 and supply power. Renewable energy is also simply called "renewable energy." A power distribution substation 174 is connected to the power distribution network 173 and a power transmission network 175. FIG. 1 shows, as an example, one power distribution network indicated by reference numeral 173, one renewable energy generating source indicated by reference numeral 171, and one power distribution substation indicated by reference numeral 174. A large-scale output renewable energy generating source 172 such as a mega solar power plant is directly connected to the power transmission network 175.

[0020] Any type of energy may be used by the renewable energy power generation sources 171 and 172. Any type of renewable energy power generation source may be used, such as solar power generation, wind power generation, biomass power generation, hydroelectric power generation, and geothermal power generation.

[0021] The green computing service 100 is a cloud computing service provided to a green computing service user 12 by an application executed in a data center 160. The green computing service user 12 uses the green computing service 100 on a green computing service user terminal 102.

[0022] The planning server 50 and the management server 150 enable the provision of virtual zero-emission computing services.

[0023] The management server 150 receives a service request from a green computing service user terminal 102 via a network 107. The management server 150 assigns an application workload 162 to a specific application 161 in a specific data center 160 via a network 108.

[0024] The green computing service management server 150 transmits the processing result of the application 161 to the green computing service user terminal 102 via the networks 108 and 107. The type of the networks 107 and 108 is arbitrary, and may be, for example, a public network such as the Internet, a LAN (Local Area Network), or a WAN (Wide Area Network).

[0025] The planning server 50 creates an IT infrastructure construction plan for a green computing service and presents it to a green computing service designer 11. The IT infrastructure construction plan indicates one or more data centers 160 and the types of one or more application programs (apps) 161 to be deployed in the one or more data centers 160. The details of the processing by the planning server 110 will be described later.

[0026] The green computing service designer 11 uses a green computing service designer terminal 101 to access the planning server 50 and the management server 150 via a network 107 .

[0027] 2 is a diagram for explaining an example of hardware and software included in the planning server 50. The planning server 50 includes a processing device 51, a memory 52, a storage device 53, and a communication device 55. The processing device 51 includes processors such as a CPU (Central Processing Unit), a DSP (Digital Signal Processor), a GPU (Graphics Processing Unit), and an FPGA (Field-Programmable Gate Array). The processing device 51 executes various programs stored in the memory 52 or the storage device 53 to implement the IT infrastructure construction plan creation method according to this embodiment.

[0028] The memory 52 is called a main storage device and is composed of a read only memory (ROM), a random access memory (RAM), etc. The storage device 53 is an auxiliary storage device such as a hard disk drive (HDD) or a solid state drive (SSD).

[0029] The communication device 55 is configured using a known network interface such as a NIC (Network Interface Card), a wireless communication module, a USB (Universal Serial Interface) module, or a serial communication module. These components of the planning server 50 can communicate with each other via an internal bus 57. The planning server 50 may further include an input device such as a keyboard, mouse, or touch panel that accepts input operations from a user, and an output device such as a display that displays processing results to the user.

[0030] A DC and application type optimization program 54 is stored in the storage device 53 of the planning server 50. In addition to the programs shown here, other programs are also stored in the storage device 53. A DC installation location candidate table 200, an application candidate table 300, a search history management table 400, and a scenario management table 500 are stored in the storage device 53, and are referenced and updated by each program 54 executed by the processing device 51. In the example described below, the DC and application type optimization program 54 references these tables.

[0031] The DC and application type optimization program 54 creates multiple Pareto-optimal IT infrastructure construction plans from combinations of DC installation candidate locations and application candidate locations. The details of the process will be described later.

[0032] Next, a detailed description will be given of each piece of information stored in the planning server 50. In each table described below, some items may be omitted, and other items may be added.

[0033] 3 is a diagram showing an example of a DC installation candidate table 200. In the DC installation candidate table 200, one record is configured by a DC installation candidate ID column 201, a renewable energy supply characteristic column 202 of the DC installation candidate, a cost column 203 for installing a DC in the DC installation candidate, and an installed column 204 indicating the amount of computational resources if a DC has already been installed.

[0034] The DC installation candidate ID column 201 shows IDs assigned to data center installation candidate columns. The renewable energy supply characteristics column 202 shows information on the average and variance of the renewable energy supply amount for each time period in the DC installation candidate. In this example, information for each hour is shown. If these averages and variances vary significantly depending on the day of the week or season, information on the average and variance of the renewable energy supply amount for each time period may be stored for each day of the week or season.

[0035] The DC installation cost column 203 indicates the cost required for installing a DC. The cost includes the cost of the land where the DC is installed, the cost of the building in which the computational resources are installed, discounts from the power grid, and the like.

[0036] The installed column 204 indicates the amount of computational resources when a DC has already been installed. If the DC has not yet been installed in the DC installation candidate, "None" is stored, and if the DC has already been installed, the amount of computational resources is stored. Here, the amount of computational resources is expressed as maximum power consumption (W), and the larger the maximum power consumption, the larger the amount of computational resources. Note that the method of expressing the amount of computational resources is not limited to this.

[0037] This information is set by the system designer or the green computing service designer 11.

[0038] 4 is a diagram showing an example of an application candidate table 300. In the application candidate table 300, each record is composed of an application candidate ID column 301, a characteristic column 302 of a WL executed to operate the corresponding application candidate, a WL execution price column 303 indicating the profit obtained from the WL executed to operate the corresponding application candidate, and an in operation column 304 indicating the scale of an application already in operation. The scale of an application indicates its processing volume, and in this example, is expressed by the amount of power consumption. A large amount of power consumption means a large amount of processing.

[0039] The application candidate ID column 301 shows IDs assigned to application candidates.

[0040] Some WLs executed to operate candidate applications have a grace period (delay tolerance) before deployment depending on the purpose, such as batch processing. Therefore, the WL characteristics column 302 shows information on the average and variance of WL power consumption for each WL delay tolerance in each time slot for the candidate application. Each WL is deployed between the start time of the assigned time slot and the time after the delay allowable time. In this example, information for each hour is shown. If these averages and variances vary significantly depending on the day of the week or season, information on the average and variance of WL power consumption in each time slot may be stored for each day of the week or season.

[0041] In addition, in cloud services, WLs with delay tolerance can often be executed at a lower cost than those without delay tolerance. Also, the price of WL execution often varies depending on the services accompanying the application. Therefore, the WL execution price column 303 stores the cost to be charged to the green computing service user for WL execution for each application candidate and each delay tolerance. "node" indicates the number of hosts (computing nodes) that execute the application.

[0042] If the application is already in operation, its scale is shown in the "in operation" column 304. If the candidate application is in operation, its scale is stored, and if it is not in operation, "None" is stored.

[0043] This information is set by the system designer or the green computing service designer 11.

[0044] 5 is a diagram showing an example of a search history management table 400. In the search history management table 400, one record is configured with a search history ID information column 401, a configuration column 402 indicating an IT infrastructure construction plan in the corresponding search history, a target value column 403 indicating values ​​in the corresponding search history and configuration for multiple indicators taken into account during operation and construction, and a searched column 404 indicating whether the indicators in the corresponding search history have been calculated.

[0045] The search ID indicates the ID assigned to the search candidate generated by the search candidate determination process S4.

[0046] The configuration column 402 indicates the amount of computational resources to be installed in each DC installation candidate, the scale to be provided to the user for each application candidate of the application candidate combination, and the policy indicating the priority in determining the DC configuration. In the example shown in FIG. 5, the amount of computational resources is indicated by the maximum power consumption (W), and the scale of the application candidate is indicated by the annual power consumption. If a DC is not installed in any of the DC installation candidates, or if any of the application candidates is not provided, the corresponding value becomes 0. Also, if the corresponding record has not been processed by the scenario generation process S5 or later, the DC configuration is undecided, so None is stored. Also, the policy indicates, for example, a weight indicating which of the initial cost and the operating cost is to be prioritized as shown in the figure. Alternatively, other weights may be substituted or added as necessary by the system designer or the green computing service designer 11. The policy may be omitted.

[0047] The target value column 403 indicates the values ​​of each index during operation and construction in the corresponding configuration, such as DC construction cost, renewable energy utilization amount, non-renewable energy utilization amount, operation cost, and WL execution profit, calculated by the index calculation and search history registration process S7. These multiple indexes include indexes that have a trade-off relationship. The types of indexes to be stored may be set by the system designer or green computing service designer 11, with deletion and addition as necessary. Also, if the corresponding record has not been processed by the scenario generation process S5 or later, the corresponding index has not been calculated, so "None" is stored.

[0048] These pieces of information are updated by the search candidate determination process S4 and the index calculation and search history registration process S7.

[0049] 6 is a diagram showing an example of a scenario management table 500. The scenario management table 500 includes a scenario ID column 501, a WL scenario column 502 indicating the WL amount for each time period and for each delay tolerance, and a renewable energy supply scenario column 503 indicating the renewable energy supply amount for each time period and for each DC installation candidate. Furthermore, the table includes a WL allocation column 504 indicating the corresponding WL scenario, the execution DC of each WL in the renewable energy supply scenario, and the execution time, and a power consumption column 505 indicating the power consumption for each DC and for each time period in the corresponding WL allocation. One record is composed of information of these items.

[0050] The scenario ID column 501 indicates an ID assigned to a scenario generated by the scenario generation process S5. The ID is stored in the format of XY, where X is the search ID 401 in the search history management table 400 and Y is the ID assigned to the scenario.

[0051] The WL scenario column 502 stores the WL amount for each time period and each delay tolerance, which is generated by the scenario generation process S5. As shown in the figure, the total of all applications is stored as this information, or it can be stored individually for each application.

[0052] The renewable energy supply scenario column 503 stores the amount of renewable energy supply for each time period and each DC installation candidate, which is generated by the scenario generation process S5. As shown in the figure, the total value regardless of the supply source is stored for this information, or it can be stored individually for each supply source in the form of self-generated power and supplied renewable energy.

[0053] The WL allocation column 504 stores information on the WL allocation in the relevant WL scenario and renewable energy supply scenario, which is obtained by the DC configuration and WL allocation optimization process S6. Specifically, as shown in the figure, the WL execution amount for each time period and each DC allocation destination, for each time period in the WL scenario, and for each delay tolerance is stored.

[0054] The power consumption column 505 stores information on the power consumption for each DC in the corresponding WL arrangement, which is calculated by the DC configuration and WL arrangement optimization process S6. Specifically, as shown in the figure, the power consumption for each time period and each DC arrangement destination is stored. The corresponding power consumption includes not only the increase in power consumption due to WL execution, but also the power consumption for air conditioning of the DC, the standby power consumption of the computing resources, and the power consumption of other equipment lighting, etc.

[0055] This information is updated by the scenario generation process S5 and the DC configuration and WL placement optimization process S6.

[0056] Next, the process executed by the planning server 50 will be described.

[0057] 7 is a flowchart for explaining an example of the DC and application configuration optimization process S1. A program corresponding to this process is included in the DC and application type optimization program 54 stored in the storage device 53 of the planning server 50. This process is started by a request from the green computing service designer 11.

[0058] In the DC and application configuration optimization process S1, first, a use case is selected in response to a request from the green computing service designer 11 (S2). If the use case is an initial plan, i.e., if there is no existing DC or application in operation, the process proceeds to the search candidate determination process (S4). If the use case is a plan update, i.e., if there is an existing DC or application in operation, records indicating information on the installed DC and the application in operation are added to the DC installation candidate table 200 and the application candidate table 300 (S3). When performing the process of (S3), as described later, the IT infrastructure construction plan indicates the optimal DC and application to be added to the existing DC and application. This process is updated by the green computing service designer 11 through an appropriate UI or the like, or is automatically updated using the log of the management server 150 containing information on the existing DC and application. After that, the process proceeds to the search candidate determination process (S4).

[0059] Next, a search candidate determination process (S4), a scenario generation process (S5), a DC candidate and WL placement optimization process (S6), and an index calculation and search history registration process (S7) are performed. The contents of these processes will be described later.

[0060] Next, it is confirmed whether searches have been performed a prescribed number of times or for a prescribed period of time (S8). If Yes, the process ends, and if No, the process returns to the search candidate determination process (S4) and executes the processes from S4 onwards again.

[0061] 8 is a flowchart showing an example of the search candidate determination process S4. The functional unit that executes this process corresponds to the search candidate generation unit. This process generates information that uniquely determines one proposal for an IT infrastructure construction plan, and the process from S5 onward complements the IT infrastructure construction plan, and calculates indicators for construction and operation.

[0062] In this embodiment, a genetic algorithm (NSGA-II) is used as an example, but search candidates can also be generated using other metaheuristic methods related to multi-objective optimization, such as particle swarm optimization. In this embodiment, application configuration and policies are generated as information for uniquely determining one proposal for an IT infrastructure construction plan, and DC allocation is determined later in the DC configuration and WL allocation optimization process S6, but application configuration and DC configuration may be generated directly (for reasons such as distributing and executing WL allocation optimization for each scenario).

[0063] The search candidate determination process S4 first checks whether there are any unsearched records, i.e., records for which the searched column 404 is False, in the search history management table 400. If Yes, the process proceeds to S402, and if No, the process proceeds to S404.

[0064] In S402, an unsearched record is selected in the search history management table 400. Next, the configuration column 402 of the corresponding record is referenced, and the application configuration and policy are output (S403).

[0065] In S404, it is confirmed whether or not a record exists in the search history management table 400. If Yes, the process proceeds to S405, and if No, the process proceeds to S407.

[0066] In S405, new search candidates are generated from the searched records. From all records, the target value column 403 is referenced, and the top M individuals that are determined to be promising for multiple indicators by a predetermined algorithm, for example, non-dominated sorting, are selected (S405). Here, the individuals refer to application configurations and policies.

[0067] Then, for the M individuals, genetic operations such as crossover and mutation are performed to generate P individuals, and for each individual that satisfies the constraints on the applications in operation, that is, for each individual that does not reduce the scale of the applications in operation, a search ID is assigned and a record is added to the search history management table 400 (S406). Then, proceed to S402. Parameters such as M and P are set by the system designer or the green computing service designer 11.

[0068] In this way, steps S405 and S406 can generate search candidates that are thought to be closer to the Pareto optimum from the already searched records. By repeatedly executing the generation of search candidates in steps S405 and S406 and the calculation of indices for the search candidates in S5, S6, and S7, it is possible to bring the Pareto front of all the search candidates closer to the Pareto optimum and generate search candidates for IT infrastructure construction plans that are not dominated by others in multiple indices or are dominated to a small extent by others.

[0069] In S407, initial search candidates are generated randomly so as to satisfy the constraints on the applications in operation, that is, to the extent that the scale of each application in operation is not reduced. Then, the process proceeds to S402.

[0070] 9 is a flowchart showing an example of the scenario generation process S5. The functional unit that executes this process corresponds to the scenario generation unit. This process generates a scenario for calculating an operational index in an IT infrastructure construction plan.

[0071] First, the scenario generation process S5 receives the output of the search candidate determination process S4 as input (S501), and then initializes the scenario list to an empty list (S502).

[0072] Then, one WL scenario is generated. First, the WL scenario is initialized with a zero vector (S503). Then, an unprocessed application type in the current WL scenario is selected (S504). Then, the application configuration and the WL characteristics column 302 of the application candidate table 300 are referenced and multiplied to calculate the average variance of the WL amount for each time period and each possible delay time for the corresponding application type, and the WL amount for each time period and each possible delay time is generated by sampling according to the normal distribution N(μ,σ) (S505). Here, a normal distribution is used to generate the scenario, but other distributions or other methods for generating random noise may be used.

[0073] Then, the WL amount for each time period and delay allowable time generated for the corresponding application type is added to the WL scenario (S506). Then, it is determined whether there is an unprocessed application type in the current WL scenario, and if Yes, the processes of S504, S505, and S506 are executed again, and if No, the process proceeds to S508 (S507).

[0074] Next, one renewable energy supply scenario is generated. First, the renewable energy supply scenario is initialized with an empty list (S508). After that, unprocessed DC installation candidate in the current renewable energy supply scenario is selected (S509). After that, the DC installation candidate table is referenced, and the RE supply amount for each time period is generated by sampling according to the normal distribution N(μ,σ) using the respective average variance (S510). Here, a normal distribution is used to generate the scenario, but other distributions or other methods for generating random noise may be used.

[0075] Then, the renewable energy supply amount for each time period generated for the DC installation candidate is added to the renewable energy supply scenario (S511). Then, it is determined whether there is an unprocessed DC installation candidate in the current renewable energy supply scenario, and if Yes, the processes of S509, S510, and S511 are executed again, and if No, the process proceeds to S513 (S512).

[0076] Next, a WL scenario and a renewable energy supply scenario are added to the scenario list (S513). It is determined whether the number of elements in the scenario list is less than N (S514). If Yes, the process returns to S503 and executes the subsequent processing, and if No, the application configuration, policy, and scenario list are output (S515). The parameter N indicating the number of scenarios is set by the system designer or green computing service designer 11.

[0077] FIG. 10 is a flowchart showing an example of the DC configuration and WL placement optimization process S6. This process optimizes the DC configuration and the WL placement of each scenario. The DC configuration is treated as a set of variables common to all scenarios, and a robust configuration that is not overly specialized for a specific scenario and can be used for multiple scenarios can be obtained. The functional unit that executes this process corresponds to the DC placement and WL placement optimization unit. In addition, in the search candidate determination process S4, when the DC placement has already been determined, the functional unit that executes the process omitting the processes of S602, S603, S615, and S616 corresponds to the WL placement optimization unit. The optimization problem shown here is an example, and includes adding constraints and variables and changing the purpose, such as when the WL of a specific application is executed only at a specific base from the viewpoint of information security, or when the delay is reduced as much as possible among the allowable delays for a WL that has delay tolerance.

[0078] The DC configuration and WL placement optimization process S6 receives the output of the scenario generation process S5 as input (S601).

[0079] Next, we describe the optimization problem of DC configuration and WL placement.

[0080] First, variables and constraints related to the DC configuration that are common to all scenarios are added. First, a variable that indicates the amount of computational resources for each candidate DC installation location is added (S602). If this variable is 0, it means that a DC will not be installed in that candidate DC installation location. Next, a constraint is added for each DC: "amount of computational resources >= amount of computational resources already installed" (S603).

[0081] Next, we add variables and constraints related to WL placement for each scenario, and add expressions related to the objective.First, we select an element of the unprocessed scenario list (S604).

[0082] First, the variables related to WL allocation for each scenario are explained. First, a variable indicating the "amount of renewable energy used for each combination of DC and time period" in the relevant scenario is added (S605). Next, a variable indicating the "WL execution amount for each combination of DC, application candidate, time period, and WL delay possible time" in the relevant scenario is added (S606). Next, a variable indicating the "DC power consumption for each combination of DC and time period" in the relevant scenario is added (S607).

[0083] Next, the constraints on WL allocation for each scenario are explained. First, the following constraint is added for the scenario: "For each combination of application candidate and WL delay time, the total WL execution amount for the corresponding time period and corresponding DC = corresponding WL amount" (S608). This constraint indicates that all WLs are executed on one of the DCs within the range of their delay tolerance. Next, the following constraint is added for the scenario: "WL execution amount for each combination of DC and time period <= constraint on the amount of computational resources of the corresponding DC installation candidate" (S609). This constraint indicates that the WL amount allocated to each DC does not exceed the limit of the DC's computational resources.

[0084] Next, a constraint is added for the scenario: "Amount of renewable energy used for each combination of DC and time period <= min (RE supply amount for the DC and time period, DC power consumption for the DC and time period)" (S610). This constraint indicates that even if renewable energy is supplied to each DC in excess of its power consumption, the renewable energy cannot be consumed. Next, a constraint is added for the scenario: "Power consumption for each DC and time period = WL execution amount for the DC and time period + computational resource amount * (standby power + air conditioning power)" (S611). This constraint is for calculating the power consumption of the DC.

[0085] Next, the formula to be added to the target value for WL allocation for each scenario will be explained. First, "(Total power consumption - Total energy used) * Non-renewable energy (BE: Borne Energy) price + Total energy used * Renewable energy price" is added to the target value (S612). This formula shows the power cost related to DC operation. Next, the sum of "(WL execution amount per delay tolerance) * (WL execution price per delay tolerance)" is subtracted from the target value (S613). This formula shows the revenue from WL execution.

[0086] Next, it is determined whether all elements in the scenario list have been processed, and if No, the process returns to S604 and executes the subsequent processes again, whereas if Yes, the process proceeds to S615 (S614).

[0087] Next, the formula to be added to the target value related to the DC configuration common to all scenarios is explained. First, "When (amount of computational resources>0) for DCs at each base where no existing DC has been installed, the corresponding DC installation price * initial cost priority" is added to the target value (S615). This formula is obtained by introducing a binary variable indicating (amount of computational resources>0) for each DC, and adding a constraint so that if the value is 0, the corresponding DC's amount of computational resources = 0. This formula indicates the cost of installing a DC, and by multiplying it by the initial cost priority included in the policy, it is determined whether to prioritize the operating cost or the initial cost. Next, "(amount of computational resources to be added overall) * computational resource price * initial cost priority" is added to the target value (S616). This formula indicates the procurement cost of computational resources to be installed in a DC, and by multiplying it by the initial cost priority included in the policy, it is determined whether to prioritize the operating cost or the initial cost.

[0088] Next, the optimization problem described in the processes from S602 to S616 is solved so as to minimize the objective value under the constraints (S617). This optimization problem can be written as a mixed integer linear optimization problem and can be solved using a general-purpose solver.

[0089] Using the obtained optimal solution, the optimization results for each scenario are added to the scenario management table 500 (S618). Specifically, the variable of "WL execution amount for each combination of DC, application candidate, each time period, and WL delay possible time" added in S606 corresponds to the WL allocation column 504, and the variable of "DC power consumption for each combination of DC and time period" added in S607 corresponds to the DC power consumption column 505. These optimal values ​​and the corresponding WL scenarios and renewable energy supply scenarios are assigned appropriate scenario IDs and stored. Next, the application configuration and DC configuration are output (S619). The DC configuration is the optimal value of the variable representing the "computational resource amount for each DC installation candidate" added in S602.

[0090] 11 is a flowchart explaining an example of the index calculation and search history registration process S7. This process calculates indexes for the construction and operation of the relevant IT infrastructure construction plan. The functional unit that executes this process corresponds to the index calculation unit. The indexes shown here are examples, and include adding or deleting the target index, such as reducing delay as much as possible from among the delays allowed for a WL that has delay tolerance.

[0091] First, the index calculation and search history registration process S7 receives the output of the DC configuration and WL placement optimization process S6 as input (S701).

[0092] Next, the value of the index to be considered during construction is calculated from the DC configuration (S702). Specifically, it is the cost of constructing the DC configuration of the corresponding IT infrastructure construction plan. For example, the DC installation cost of the candidate DC installation site where the DC will be installed in the corresponding DC configuration and a value obtained by multiplying the total amount of computing resources by a predetermined coefficient are added. Here, if necessary, it is possible to multiply another index, for example, the amount of computing resources or air conditioning equipment used, by a predetermined coefficient to calculate the CO2 emissions during manufacturing.

[0093] Next, the calculation of the values ​​of the indices to be considered during operation will be described. First, the values ​​of the indices during operation are initialized to 0 (S703). Next, an unprocessed scenario for the corresponding search candidate is selected from the scenario management table (S704). Next, the indices during operation in the corresponding scenario are calculated from the DC power consumption and WL placement, and added to the indices during operation (S705). Specifically, the amount of renewable energy used, the amount of non-renewable energy used, the operation cost, and the WL execution profit are calculated. Here, it is possible to calculate another index, for example, the total amount of WL delay, if necessary. Next, it is determined whether all scenarios for the corresponding search candidate have been processed. If Yes, proceed to the processing of S707, and if No, return to S704 and execute the subsequent processing again (S706).

[0094] Next, for the corresponding record in the search history management table 400, the DC configuration, operation indicators, and construction indicators are added to the target value column 403, and the searched column 404 is changed to True. At this time, the target value of the operation indicator may be multiplied by an appropriate coefficient, and the average of all scenarios and the total amount throughout the year may be stored.

[0095] 12 is a diagram showing an example of a plan result display screen 1000 for the green computing service designer 11 displayed on the green computing service designer terminal 101. The plan result display screen 1000 includes a list display field 1100 of a plurality of created IT infrastructure construction plans, a DC configuration display field 1200 for one selected IT infrastructure construction plan, an application configuration display field 1300 for one selected IT infrastructure construction plan, an index display field 1400 for one selected IT infrastructure construction plan, and a scenario details display field 1500 for one selected IT infrastructure construction plan.

[0096] The IT infrastructure construction plan list display field 1100 displays a list of IT infrastructure construction plans searched for by the DC and application configuration optimization process S1. Specifically, the contents of the search history management table 400 are formatted and displayed. The displayed contents may be in list form as shown in the figure, or may be visualized, such as a scatter diagram, using the values ​​of the indices in each IT infrastructure construction plan. The IT infrastructure construction plan list display field 1100 may select and display a set of plans that are not dominated by other plans from among the IT infrastructure construction plans searched for by the DC and application configuration optimization process S1.

[0097] The DC search candidate selection field 1104 selects configuration candidates to be displayed in the DC configuration display field 1200 , the application configuration display field 1300 , the index display field 1400 , and the scenario details display field 1500 .

[0098] The DC configuration display field 1200 displays the DC configuration in the configuration candidate selected in the search candidate selection field 1104. Specifically, for each DC installation candidate, a DC installation destination ID 1201, an amount of computational resources to be installed 1202, and a construction cost 1203 are displayed. The display content may be in a list format as shown in the figure, or may be displayed on a map by providing geographical information for each installation candidate.

[0099] An application configuration display field 1300 displays the application configuration in the configuration candidate selected in the search candidate selection field 1104. Specifically, an application ID 1301, a target operation scale 1302, and a revenue forecast 1303 are displayed for each application candidate.

[0100] The indicator display field 1400 displays the DC configuration in the configuration candidate selected in the search candidate selection field 1104. Specifically, the values ​​of the indicators considered during construction and operation are displayed. The display content may be in list form as shown in the figure, or may be accompanied by visualization such as a bar graph.

[0101] A scenario details display field 1500 displays details of each scenario in the configuration candidate selected in the search candidate selection field 1104. A scenario selection field 1501 selects a scenario ID and a DC installation candidate ID to be displayed in a renewable energy supply and DC power consumption display field 1502. The renewable energy supply and DC power consumption display field 1502 displays information about the DC installation candidate selected in the scenario selection field 1501 for the configuration candidate (corresponding to the search ID 501) selected in the search candidate selection field 1104 and the record corresponding to the scenario ID selected in the scenario selection field 1501 in the scenario management table 500.

[0102] Here, the horizontal axis is time, and the vertical axis is power consumption. DC renewable energy supply 1511 indicates the amount of renewable energy supply in the relevant scenario and DC, and DC power consumption 1521 indicates the DC power consumption in the relevant scenario and DC. In Fig. 12, one bar indicating DC power consumption is indicated by reference numeral 1521 as an example. Here, the display method may be a bar graph as shown in the figure, or a stacked bar graph by delay tolerance or by application type, etc. EXAMPLES

[0103] In this embodiment, the execution DC for WL placement is not dynamically determined during operation as in the first embodiment, but is determined for each application type at design time. Only the tables and steps that have been changed from the first embodiment are shown below with reference to Figures 5, 8, and 10.

[0104] 402: Regarding the application configuration, a DC installation destination candidate ID of the application execution DC is added for each application type.

[0105] S406, S407: The application configuration includes the DC installation destination candidate ID of the application execution DC and the application type.

[0106] S606: Instead of adding a variable indicating "the WL execution amount for each DC, each application candidate, each time period, and each possible WL delay time," a variable is added that indicates "the WL execution amount at the candidate DC installation location for the application candidate, for each application candidate, each time period, and each possible WL delay time."

[0107] Making these changes will reduce the freedom of the DC that runs the WL, which will result in a decrease in renewable energy consumption, etc.; however, it will eliminate the need to dynamically move the WL of an application between DCs during operation, which will reduce the power consumption required to move the WL and the total power consumption due to reduced data replication, and will simplify implementation. [Explanation of symbols]

[0108] 50 Planning Server 54 DC and application type optimization program 200 DC installation candidate table 300 App candidate table 400 Search candidate management table 500 Scenario Management Table

Claims

1. An apparatus for creating an IT infrastructure construction plan indicating a data center installation location and a workload of an application to be deployed at the data center installation location, Processor and a storage device, the storage device stores application candidate information and data center installation site candidate information; The application candidate information indicates information about a plurality of applications for calculating a plurality of preset indicators in the construction and operation of an IT infrastructure, the data center location candidate information indicates information about multiple data center location candidates for calculating the multiple indexes; The processor: creating a plurality of search candidates indicating application combinations selected from the application candidate information; For each of the plurality of search candidates, by referring to the application candidate information and the data center installation location candidate information, calculate values ​​of the plurality of indicators of an IT infrastructure construction plan indicating the application combination and the data center installation location to which the workload of the application combination will be deployed; An apparatus that outputs, from among the IT infrastructure construction plans of the plurality of search candidates, a plurality of IT infrastructure construction plans that are not superior to other IT infrastructure construction plans in terms of the plurality of indicators.

2. 10. The apparatus of claim 1, The processor selects, from among the IT infrastructure construction plans that have already been created, an IT infrastructure construction plan that has been determined to be promising in terms of the multiple indicators according to a predetermined algorithm, and generates new application combinations to be searched for from the application combinations of the selected IT infrastructure construction plan.

3. 10. The apparatus of claim 1, the application candidate information indicates information about power consumption of workloads of the plurality of applications for each time period; the data center installation site candidate information indicates information regarding the renewable energy supply amount for each of the time periods of the plurality of data center installation site candidates; The processor: generating, for each of the plurality of search candidates, one or more scenarios indicating power consumption for each time period of the workload of the application combination and renewable energy supply for each time period of the data center installation location where the workload is deployed, by referring to the application candidate information and the data center installation location candidate information; An apparatus for determining a data center configuration that optimizes a predetermined objective value for the one or more scenarios, and an execution time period and deployment destination of the workload for each scenario.

4. 4. The device according to claim 2 or 3, The processor generates data center configurations as part of the search candidates.

5. 10. The apparatus of claim 1, When an installed data center or an application in operation exists, the processor creates and outputs the IT infrastructure construction plan without reducing the scale of the installed data center or the scale of the application in operation.

6. 10. The apparatus of claim 1, The processor determines a data center that will execute a workload for each application in the IT infrastructure construction plan.

7. A method for creating an IT infrastructure construction plan indicating a data center installation location and a workload of an application to be deployed at the data center installation location, comprising: The device stores application candidate information and data center installation site candidate information; The application candidate information indicates information about a plurality of applications for calculating a plurality of preset indicators in the construction and operation of an IT infrastructure, the data center location candidate information indicates information about multiple data center location candidates for calculating the multiple indexes; The method further comprises the steps of: creating a plurality of search candidates indicating application combinations selected from the application candidate information; For each of the plurality of search candidates, by referring to the application candidate information and the data center installation location candidate information, calculate values ​​of the plurality of indicators of an IT infrastructure construction plan indicating the application combination and the data center installation location to which the workload of the application combination will be deployed; A method for outputting, from among the plurality of search candidate IT infrastructure construction plans, a plurality of IT infrastructure construction plans that are not superior to other IT infrastructure construction plans in terms of the plurality of indicators.

8. 8. The method of claim 7, The method comprises the steps of: selecting, from among the IT infrastructure construction plans that have already been created, an IT infrastructure construction plan that has been determined to be promising in terms of the multiple indicators according to a predetermined algorithm; and generating new application combinations to be searched for from the application combinations of the selected IT infrastructure construction plan.

9. 8. The method of claim 7, the application candidate information indicates information about power consumption of workloads of the plurality of applications for each time period; the data center installation site candidate information indicates information regarding the renewable energy supply amount for each of the time periods of the plurality of data center installation site candidates; The method further comprises the steps of: generating, for each of the plurality of search candidates, one or more scenarios indicating power consumption for each time period of the workload of the application combination and renewable energy supply for each time period of the data center installation location where the workload is deployed, by referring to the application candidate information and the data center installation location candidate information; The method further comprises determining a data center configuration that optimizes a predetermined objective value for the one or more scenarios, and determining an execution time period and a deployment destination of the workload for each scenario.

10. 10. The method of claim 8 or 9, The method, wherein the apparatus generates data center configurations as part of the search candidates.

11. 8. The method of claim 7, A method in which, when an installed data center or an application in operation exists, the device creates and outputs the IT infrastructure construction plan without reducing the scale of the installed data center or the scale of the application in operation.

12. 8. The method of claim 7, The method wherein the device determines a data center in which to execute a workload for each application in the IT infrastructure construction plan.

Images