Distributed processing system, distributed processing method and program

Abstract

A distributed processing system, a distributed processing method, and a program are provided that create a processing schedule plan for a target task based on energy cost efficiency and enable efficient task processing. [Solution] A distributed processing system (1) comprises two or more distributed servers (2) installed in multiple geographically dispersed locations, and a computer (3) that controls the distributed servers (2). The distributed servers (2) operate using power supplied from at least one of green power (4), storage batteries (5), and grid power (6) at the locations where the distributed servers (2) are installed. The computer (3) comprises an acquisition unit that acquires processing task requests (100) from users (7), and a processing unit that executes a processing task request (100) in response to the requests (100). , the amount of tasks that a distributed server can process per unit time divided by the current and future electricity prices Based on the energy cost efficiency of each distributed server, The processing time and a planning unit that plans how long it will take to process the job.

Description

[Technical Field]

[0001] The present invention relates to a distributed processing system, a distributed processing method, and a program for planning task processing on two or more distributed servers installed in multiple geographically dispersed locations. [Background technology]

[0002] Conventionally, task processing by distributed servers has been achieved by distributing the load mainly based on performance factors such as CPU usage, memory usage, and communication latency for each distributed server. However, with the spread of renewable energy, it is becoming necessary to balance loads taking into account energy supply forecasts, fluctuations in electricity market prices, and the energy costs of each server site.For example, when operating distributed servers using solar power, a renewable energy source, it is important to consider energy management factors such as the forecast of solar power generation at each server site, the remaining capacity of storage batteries, and the market trading price of grid electricity at the Japan Electric Power Exchange (JEPX).

[0003] For example, Patent Document 1 discloses a technology in which a target power value for a base where a server is installed is set based on requests for increasing or decreasing power consumption, power charges, and power generation amount at the base, and load is transferred between multiple servers based on the target power value. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Application No. 2024-129969 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the technology disclosed in Patent Document 1 did not allow a user to plan which distributed server should process one or more tasks requested by the user and for how long based on current or future electricity prices, and then have the distributed server process the tasks based on the plan.

[0006] Therefore, the inventors focused on the fact that by creating and efficiently managing a processing schedule plan for the target task based on the energy cost efficiency calculated from current or future electricity prices and the amount of tasks that can be processed per unit of electricity by the distributed server, the load on multiple distributed servers can be optimized, making it possible to achieve efficient energy use and cost reduction.

[0007] In view of these problems, the present invention aims to provide a distributed processing system, a distributed processing method, and a program that create a processing schedule plan for a target task based on energy cost efficiency, enabling efficient task processing. [Means for solving the problem]

[0008] The present invention provides the following solutions.

[0009] The present invention provides a distributed processing system comprising two or more distributed servers installed in multiple geographically dispersed locations and a computer that controls the distributed servers, The distribution server operates on power supplied from at least one of green power, a storage battery, and grid power at the location where the distribution server is installed, The computer an acquisition unit that acquires a processing task request from a user; In response to said request , the amount of tasks that the distributed server can process per unit time, divided by the current and future electricity prices. Based on the energy cost efficiency of the distributed servers, The processing time and a planning unit that plans how long it will take to process each task.

[0010] Although the present invention is categorized as a system, the same effects and advantages can be obtained even when it is a method or a program. [Effects of the Invention]

[0011] According to the present invention, it is possible to provide a distributed processing system, a distributed processing method, and a program that create a processing schedule plan for a target task based on energy cost efficiency and enable efficient task processing. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram illustrating a basic overview of a distributed processing system 1 according to an embodiment of the present invention. [Figure 2] 1 is a flowchart of distributed processing executed by the distributed processing system 1 of this embodiment. [Figure 3] 1 is a block diagram showing a basic configuration of a distributed processing system 1 according to an embodiment of the present invention. [Figure 4] 10 is a flowchart of a real-time task / non-real-time task separation process executed by the distributed processing system 1 of the present embodiment. [Figure 5] 10 is a flowchart of an energy cost efficiency improvement process executed by the distributed processing system 1 of this embodiment to optimize the energy cost efficiency of each distributed server 2. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, the best mode for carrying out the present invention will be described with reference to the drawings. In the following drawings, the same elements are designated by the same numbers or symbols throughout the description of the embodiments. Note that this is merely an example, and the technical scope of the present invention is not limited to this.

[0014] [Basic Overview of Distributed Processing System 1] The basic overview and configuration of a distributed processing system 1 according to one embodiment of the present invention will be described with reference to Fig. 1 and Fig. 2. Fig. 1 is a diagram for explaining the basic overview of the distributed processing system 1 according to one embodiment of the present invention. Fig. 2 is a flowchart of distributed processing executed by the distributed processing system 1 according to this embodiment.

[0015] As shown in FIG. 1, a distributed processing system 1 of this embodiment is made up of a plurality of distributed servers 2 installed in a plurality of geographically dispersed locations, and a computer 3 that controls the distributed servers 2. The distributed server 2 operates on power supplied from at least one of green power 4 generated at the location where the distributed server 2 is installed, a storage battery 5, and grid power 6. The distributed servers 2 (2-1, 2-2, . . . , 2-n) are server computers for processing tasks described below.

[0016] Here, green electricity 4 (4-1, 4-2, . . . , 4-n) refers to electricity generated by renewable energy sources installed at the location where the distribution server 2 is installed. For example, this electricity may be generated using solar panels 4-2 or windmills 4-3. The following description will be focused on green electricity 4 generated using solar panels. Note that there may be one or more types of equipment for generating the green electricity 4 supplied to the distribution server 2, and the number and type are not limited. Because these green electricity sources 4 are installed geographically distributed, just like the distribution servers 2, their output depends on the weather in the areas where they are installed. For example, neither solar nor wind power can generate electricity on rainy, windless days.

[0017] Furthermore, the storage batteries 5 (5-1, 5-2, . . . , 5-n) may be charged with generated green electricity 4 and / or grid electricity 6 (6-1, 6-2, . . . , 6-n), where grid electricity 6 refers to electricity supplied from a power plant via a power transmission and distribution network owned by an electric power company. Note that there may be one or more types of storage batteries 5 and grid electricity 6 for each distribution server 2 that supply power to the distribution server, and the number and types are not limited. Grid power6 is electricity generated at large-scale power plants such as thermal and nuclear power plants and supplied to individual households via the power grid. Generally, nuclear power plants are unable to precisely adjust their output to meet demand, so they are used as a baseload power source, and stable power is maintained by adjusting output using thermal power plants, which are easy to adjust. Nuclear power plants cannot reduce their output even when power demand decreases, so during times of extremely low power demand, even if thermal power plant output is minimized, excessive power may still be generated.

[0018] The computer 3 of the distributed processing system 1 has a server function and may be realized, for example, by a single computer, or may be realized by multiple computers, such as a cloud computer. The computer 3 in the embodiment of this specification is a cloud computer. The cloud computer in this specification may be either one that uses any computer in a scalable manner to perform a specific function, or one that includes multiple functional modules to realize a system and uses the functions in any combination.

[0019] Here, the computer 3 may be equipped with a user terminal 8 to receive a request 100 from a user 7 for a task to be processed by the distributed server 2. The user terminal 8 may be, for example, an information terminal such as a mobile phone, a smartphone, a tablet terminal, a personal computer, a laptop computer, a virtual reality headset, a voice input terminal, a self-driving car, or various types of mobility, and the number of such user terminals may be one or more for each user. The user terminal 8 in the embodiment of this specification is a laptop computer. Furthermore, the object that accesses the computer 3 is not limited to the user terminal 8 operated by the user 7, but may also be a device that autonomously generates processing tasks. For example, this includes cases where an autonomous driving mobility or a proactive training data generation AI tool directly accesses the computer 3 and requests task processing.

[0020] The distributed server 2, the computer 3, and the user terminal 8 are connected to each other via a network 10 such as a public line network or a dedicated line so as to enable data communication.

[0021] As shown in FIGS. 1 and 2, the computer 3 of the distributed processing system 1 in this embodiment receives a processing task request 100 from a user 7 via a user terminal 8 (step S1).

[0022] Next, in response to the processing task request 100, the computer 3 of the distributed processing system 1 in this embodiment creates a plan 110 that determines which distributed server will process which task and for how long, based on the energy cost efficiency of each distributed server 2 (step S2).The processing tasks 120 are then assigned to the distributed servers based on the plan 110 (step S3).

[0023] Here, the energy cost efficiency for each distributed server is a formula derived by dividing the amount of tasks that can be processed per unit of power by the distributed server by the current or future power price.

[0024] The above is the basic outline of the distributed processing system 1 in this embodiment. With such a distributed processing system 1, efficient management is possible by creating a processing schedule plan for a target task based on energy cost efficiency.

[0025] [Basic configuration of distributed processing system 1] 3 is a block diagram showing the configuration of the distributed processing system 1. The basic configuration of the distributed processing system 1 will be described with reference to FIG.

[0026] As shown in FIG. 3, the distributed processing system 1 in this embodiment is composed of two or more distributed servers 2 (2-1, 2-2, . . . , 2-n), a computer 3, and a user terminal 8. The distributed processing system 1 in this embodiment is a system in which the computer 3 is connected to the distributed servers 2 and the user terminals 8 so as to be able to communicate data with them via a network 10 such as a public line network or a dedicated line. As described above, each distributed server 2 includes at least one of a device that generates green power 4, a storage battery 5, and a device for receiving grid power 6. Furthermore, the distributed processing system 1 may include other terminals and devices in addition to the distributed servers 2, the computers 3, and the user terminals 8, and the number, types, and functions of the other terminals and devices can be designed as appropriate.

[0027] The distributed server 2 and the computer 3 each include a control unit such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), etc., a memory unit such as a data storage unit using a hard disk, semiconductor memory, recording medium, memory card, etc., and a communication unit such as a device for enabling communication with other terminals, devices, etc., such as a Wi-Fi compatible device (not shown). The storage unit of the computer 3 may be located outside the computer 3 and may be capable of network communication.

[0028] The computer 3 includes an acquisition module 200 that acquires processing task requests 100 from users 7 as an acquisition unit, a planning module 210 that plans which processing tasks 120 corresponding to the acquired requests 100 will be processed on which distributed server 2 and for how long based on the energy cost efficiency of the distributed server 2, and a classification module 220 that classifies the acquired processing tasks 120 into real-time tasks 130 that are processed in real time and non-real-time tasks 140 as a classification unit.

[0029] As shown in FIG. 3, in the computer 3, a control unit (not shown) reads a predetermined program, thereby realizing an acquisition module 200, a planning module 210, and a classification module 220 in cooperation with a memory unit and a communication unit (not shown).

[0030] The user terminal 8 is equipped with a CPU, GPU, RAM, ROM, etc. as a terminal control unit, a device, etc. that enables communication with other terminals and devices, etc. as a communication unit, and various devices, etc. that accept predetermined inputs, input and output of various data, etc. as an input / output unit (not shown).

[0031] The above is the basic configuration of the distributed processing system 1. Below, each process executed by the distributed processing system 1 will be explained in more detail together with the process executed by each of the above-mentioned units.

[0032] [Real-time and non-real-time separation processing] 4 is a flowchart of the real-time / non-real-time separation process executed by the distributed processing system 1 of this embodiment. The real-time / non-real-time separation process executed by the distributed processing system 1 will be described with reference to FIG.

[0033] The computer 3 of the distributed processing system 1 of this embodiment acquires a processing task request 101 from the user 7 (step S10). The processing task request 101 may include deadline date and time information for the task.

[0034] Next, the computer 3 of this embodiment classifies the processing tasks from the acquired request 101 into real-time tasks 130 that process the tasks in real time and non-real-time tasks 140 (step S11).

[0035] Here, the classification module 220 of the computer 3 of this embodiment may assign task completion deadline date and time information 150 to the classified non-real-time tasks 140 (step S12). The classification of the processing tasks may be performed automatically based on a request 101 set by the user 7, or the processing tasks may be manually classified as real-time tasks 130 or non-real-time tasks 140 as specified by the user 7. Here, task completion deadline date and time information 151 may be assigned to the real-time tasks 130. Furthermore, as in the basic overview described above, a plan 111 is created for the processing tasks classified as real-time tasks 130, which plan determines which distributed server will process which task and for what period, based on the energy cost efficiency of each distributed server 2, in response to the request 101, and the processing tasks are then allocated to the distributed servers based on the plan 111.

[0036] For the classified non-real-time tasks 140 assigned with task completion deadline date and time information 150, a plan 112 is created that determines which task will be processed by which distributed server and for how long, based on the energy cost efficiency of each distributed server 2, in accordance with the request 101 and the task completion deadline date and time information 150 (step S13).

[0037] Next, the computer 3 of this embodiment allocates the non-real-time tasks 140 to the distributed servers 2 based on the created plan 112 (step S14).

[0038] The above is the real-time / non-real-time separated processing executed by the distributed processing system 1 of this embodiment. This type of processing makes it possible to plan a schedule with better energy cost efficiency within a predetermined time set up for task processing.

[0039] [Energy cost-effective processing] FIG. 5 is a flowchart of an energy cost efficiency improvement process executed by the computer 3 of this embodiment to optimize the energy cost efficiency of each distributed server 2.

[0040] In this embodiment, the computer 3 of the distributed processing system 1 optimizes the energy cost efficiency of each distributed server 2. The acquisition module 200 acquires from each distributed server 2 performance information 160 including at least the current and future CPU utilization rate, memory utilization rate, communication latency for each distributed server 2, storage battery status information including the amount of charge (SOC) of the storage battery 5 that supplies power to the distributed server 2 and the state (SOH: State Of Health) of the storage battery 5, and green power status information including the amount of power generated by the green power 4 that supplies power to the distributed server 2 and the state of the green power 4; The distribution server 2 acquires input information 180 including at least power price information 170 including at least information on the wholesale power price of the grid power 6 that supplies power to the distribution server 2 and the predicted future wholesale power price (step S20).

[0041] The input information 180, which includes at least the performance information 160 and the power price information 170, may be acquired from the distribution server 2. The power price information 170 may be acquired not only from the distribution server 2 but also from the electric power company supplying the grid power 6, the Japan Electric Power Exchange, or other sources. As described above, the grid power 6 adjusts the output of thermal power plants according to power demand. However, during periods of low power demand, solar power plants in sunny areas or wind power plants in windy areas may increase their output regardless of power demand, resulting in power generation exceeding power demand even when the thermal power plant's output is minimized. During such periods, power prices fall. In such situations, the selling price of the grid power 6 falls sharply, and the electricity generated by green power 4 is essentially purchased at zero price, resulting in unused power being discharged. Taking such situations into account, the Japan Electric Power Exchange discloses constantly fluctuating power price data, making it ideal for use as the power price information 170.

[0042] In order to improve energy cost efficiency, the acquisition module 200 of the distributed processing system 1 in this embodiment may acquire input information 180 including at least a newly acquired processing task request 103, performance information 160, and power price information 170, by associating them with the items in Table 1 below.

[0043] [Table 1]

[0044] Table 1 above shows an example of the main input information 180 for this optimization problem. For the purposes of the calculation formulas described below, the distributed server 2 corresponds to server k, the processing task 120 corresponds to task i, green power 4 corresponds to solar power (PV: Photovoltaic), and grid power 6 corresponds to the grid. In the items and explanations, t indicates time. In this embodiment, t is calculated as 24 hours with 48 frames (1 frame = 0.5 hours), but the time per frame is not limited. JEPX (Japan Electric Power Exchange) refers to the Japan Electric Power Exchange, and the JEPX price refers to the current or future electricity price per frame of the grid power 6 in the area where server k is located, or its predicted value. Task load and server processing capacity refer collectively to the CPU, memory, etc., and response time refers to communication latency. The inputs described here are merely representative components and can be flexibly expanded depending on the implementation objectives. For example, task start times and completion deadlines can be specified not only as integers but also in actual time format and then internally converted to units. The time granularity (units) can also be expanded beyond the current 30-minute units (48 units per day) to more frequent scheduling units such as 15-minute or 1-minute units. For processing tasks, it is desirable to design the system so that constraints and objective functions can be flexibly configured by specifying task-specific attributes (priority, type, interruptibility, parallel processing, dependencies, etc.) in addition to the load and required time on the server. For storage batteries, inputs such as unit-by-unit charge / discharge limits, charge / discharge efficiency, and even charge / discharge constraints based on degradation coefficients and SOH can be accepted, enabling expansion to achieve scheduling closer to realistic charging and discharging behavior. Furthermore, usable energy resources include not only storage batteries but also batteries installed in electric vehicles (EVs). With regard to renewable energy, by adding power generation forecasts for solar power, wind power, biomass, etc. to the inputs, it becomes possible to achieve optimization that is adapted to a variety of regular uses. In this way, the input items defined in Table 1 are merely a representative configuration, and can be flexibly redesigned as an expandable framework according to future needs and designs.

[0045] Here, the computer 3 of the distributed processing system 1 in this embodiment executes an optimization algorithm based on various input information 180 (shown in Table 1), including performance information 160 and power price information 170, as well as the objective function and constraints described below, and calculates the optimization target variables (shown in Table 2) obtained as the solution (step S21). The purpose of this optimization is to improve the energy cost efficiency of each distributed server 2 and to derive an optimal operation schedule for the server's operating status and task allocation, as well as renewable energy, storage batteries, and market procurement. Note that the computer 3 of the distributed processing system 1 in this embodiment may refer to previously requested plans for current or future processing and create a new plan, including modifications to the plan, based on the optimized energy cost efficiency of each distributed server 2.

[0046] [Table 2]

[0047] Here, based on the predicted price value per unit time of the location of the distributed server 2 in Table 1, the total amount of grid power 6 used by the distributed server 2 in Table 2, the amount of tasks processed by the distributed server 2 at time t, which is the load score, and the weighting coefficient (alpha) that converts the load score into cost, it is possible to improve the energy cost efficiency of each distributed server 2 by using an objective function (Equation 1) that minimizes the sum of the grid power 6 cost and server operation cost incurred in all time frames t of each distributed server k during the optimization period. The objective function can be expanded to meet actual operational needs. For example, by simultaneously setting the objective of maximizing the ratio of self-consumption of solar power generation on each server, it is possible to expand the model to multi-objective optimization aimed at improving the utilization rate of renewable energy. Furthermore, by adding a penalty term that takes into account the number of charge / discharge cycles and the impact on the lifespan of the storage battery, the design can also accommodate optimization that takes into account the sustainability of resources.

[0048]

number

[0049] The above objective function is optimized when the objective function value satisfies the condition and the values ​​of the following constraint functions satisfy the constraint conditions.

[0050] That is, when a requested processing task is assigned to a distributed server 2, the basic form is that the processing task 120 to be assigned is always assigned to one distributed server 2 (Equation 2), the time required to process the processing task 120 on the distributed server 2 is met (Equation 3), and the sum of the processing time of the processing task 120 and the time required for a reply from the distributed server 2 to the computer 3 (central system) meets the set deadline condition (Equation 4), thereby satisfying the constraint conditions. Depending on future operational needs, it is also possible to envision extensions such as partially processing a single task across multiple distributed servers, or allowing parallel processing for specific high-load tasks. In such cases, a consistent task division definition and additional constraints on communication costs are required, which necessitate design changes or reconfiguration of the constraint equations, but the framework can be flexibly extended.

[0051]

number

[0052]

number

[0053]

number

[0054] In order to ensure consistency between the allocation of the processing task 120 and the operating state of the distributed server 2, the following constraints are set. First, if no processing task i is assigned to server k in time frame t, then server k will not be operating during that time period (Equation 5). On the other hand, if server k is not operating in a certain time frame t, task i must not be assigned to server k either (Equation 6). Another constraint is that the load on the distributed server 2 when processing the processing task 120 must be within the maximum processing capacity of the distributed server 2 (Equation 7). In future expansion, if consideration is given to server startup delays, warm-up times, or operation for non-task purposes (e.g., data communication), it will be necessary to introduce additional conditions to complement or modify these constraints.

[0055]

number

[0056]

number

[0057]

number

[0058] Furthermore, the total amount of power used in the processing of the processing task 120, including the green power 4, the amount of discharge from the storage battery 5, and the grid power 6, is equal to the amount of power used by the processing load of the processing task 120 (Equation 8). The amount of power generated by the green power 4 is greater than the total amount of green power 4 used in processing the task and the amount of charge from the green power 4 to the storage battery 5 (Equation 9), The constraint is that the amount of power (procurement amount) of the grid power 6 to be used by the distributed server 2 is equal to the amount of power used for task processing and charging the storage battery 5 using the grid power 6 (Equation 10). It should be noted that this energy balance constraint can be expanded based on a similar structure depending on the type of green power or storage battery resource used.

[0059]

number

[0060]

number

[0061]

number

[0062] Furthermore, with regard to the state of charge (SOC) of the storage battery 5 attached to the distributed server 2, the SOC at time t is successively updated based on the difference between the SOC at the previous point in time, the total amount of charging power from renewable energy (green power 4) and grid power 6 during the same time period, and the amount of discharge used for processing tasks, etc. (Equation 11). Furthermore, the amount of power stored in the storage battery 5 is equal to or greater than the minimum capacity and equal to or less than the maximum capacity of the storage battery 5 (Equation 12). In the future, these battery constraints can be expanded to include considerations such as battery charge / discharge efficiency, loss, maximum charge / discharge capacity, battery deterioration, and cycle life control.

[0063]

number

[0064]

number

[0065] By using an optimization algorithm to minimize the objective function (Equation 1) while satisfying the above-mentioned constraints (Equations 2 to 12), it is possible to optimize the energy usage of each distributed server 2, and by optimizing the energy usage of each distributed server 2, it is possible to improve the energy cost efficiency. This is the energy cost efficiency improvement process.

[0066] With the distributed processing system 1 described above, it is possible to provide a distributed processing system, a distributed processing method, and a program that enable efficient management by creating a processing schedule plan for a target task based on energy cost efficiency.

[0067] In addition, if the power supply exceeds the power demand in the area of ​​the distributed server 2, charging of the storage battery 5 and the load of processing tasks on the distributed server 2 may be promoted, and if the power demand in the area exceeds the power supply, charging of the storage battery 5 and the load of processing tasks on the server may be suppressed.

[0068] There are no particular restrictions on where the distributed server 2 can be installed. In Japan, for example, it would be ideal to utilize vacant houses in depopulated areas or abandoned school buildings. Because the processing tasks are electronic data, there are no transmission or reception costs, even in remote depopulated areas. Furthermore, the use of vacant houses is limited due to circumstances such as the owner's base of operations being in a distant urban area. However, if the house is strong enough, it is possible to install green electricity 4 such as solar or wind power on the roof or in the garden, and if it is protected from rain and wind, it is a sufficient location for installing the distributed server 2. Furthermore, installing the distributed server 2 reduces the need for periodic inspections and increases the economic rationality of installing surveillance cameras, making it ideal from a crime prevention perspective. Furthermore, measuring the task processing volume of the distributed server 2 and storing it in memory can serve as proof of the distributed server 2's contribution to data processing, enabling revenue distribution to the owner of the vacant house. Furthermore, it becomes easy to prove that green power 4 has been used instead of thermal power generation. This allows for the creation of new revenue sources, such as the establishment of carbon emission credits based on the stored task processing volume or the amount of green power 4 used, and selling them to carbon-emitting companies. Of course, the distributed server 2 can also be installed in a residential home, a used school, or a regular office. However, since it is necessary to isolate the distributed server 2 from people working inside to prevent them from coming into contact with the server 2 during processing, arranging multiple servers 2 in a high-density arrangement is necessary. In contrast, in vacant houses or abandoned schools, where there are no people inside, there is no need to arrange the distributed servers 2 in a high-density arrangement. Therefore, natural heat dissipation is sufficient, eliminating the need for large-scale cooling facilities like data centers. Moreover, unlike abandoned schools, which are sometimes used as local cultural facilities, vacant houses are even more suitable than abandoned schools because small facilities can be individually locked.

[0069] The above-described means and functions are realized by a computer (including a CPU, an information processing device, and various terminals) reading and executing a predetermined program. The program is provided, for example, in the form of a cloud service or SaaS (Software as a Service) provided from one or more computers via a network. The program is also provided, for example, in the form of a program recorded on a computer-readable recording medium. In this case, the computer reads the program from the recording medium, transfers it to an internal or external recording device, records it, and executes it. The program may also be pre-recorded on a recording device (recording medium) such as a magnetic disk, optical disk, or magneto-optical disk, and provided to the computer from the recording device via a communication line.

[0070] Although the embodiments of the present invention have been described above, the present invention is not limited to these embodiments. Furthermore, the effects described in the embodiments of the present invention are merely a list of the most preferable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the embodiments of the present invention.

[0071] The first aspect disclosed in this embodiment is: A distributed processing system comprising two or more distributed servers installed in multiple geographically dispersed locations and a computer that controls the distributed servers, The distribution server operates on power supplied from at least one of green power generated at the location where the distribution server is installed, a storage battery, and grid power; The computer an acquisition unit that acquires a processing task request from a user; A distributed processing system is provided which includes a planning unit that, in response to the request, plans which task will be processed by which distributed server and for how long based on the energy cost efficiency of the distributed server.

[0072] The second aspect disclosed in this embodiment is When the power supply exceeds the power demand in the area of ​​the distributed server, charging of the storage battery and the load of the processing task of the distributed server are promoted, and when the power demand in the area exceeds the power supply, charging of the storage battery and the load of the processing task of the server are suppressed. According to a first aspect, there is provided a distributed processing system.

[0073] The third aspect disclosed in this embodiment is a classification unit that classifies the processing tasks into real-time tasks that are processed in real time and non-real-time tasks in response to the request, the planning unit plans which distributed server will process the classified real-time tasks and non-real-time tasks and for how long based on the energy cost efficiency of the distributed server; The present invention provides a distributed processing system according to the first or second aspect.

[0074] The fourth aspect disclosed in this embodiment is the acquisition unit acquires input information consisting of performance information including at least the current and future CPU utilization rate, memory utilization rate, and communication latency of the distributed server, storage battery status information including the amount of electricity stored in the storage battery that supplies power to the distributed server, and green power status information including the amount of electricity generated by the green power that supplies power to the distributed server, and electricity price information including at least information on the wholesale electricity price of the grid electricity and a predicted future wholesale electricity price; the planning unit plans the processing of the processing task acquired based on the energy cost efficiency for each distributed server, which is derived by dividing the task processing amount per unit power calculated from the performance information by the current and future power prices calculated from the power price information; A distributed processing system according to a third aspect.

[0075] The fifth aspect disclosed in this embodiment is A distributed processing method executed by a computer that controls two or more distributed servers that are installed in multiple geographically dispersed locations and that operate on power supplied from at least one of green power, storage batteries, and grid power, comprising: receiving a processing task request from a user; In response to the request, a step of planning which distributed server will process which task and for how long based on the energy cost efficiency of the distributed server is provided.

[0076] The sixth aspect disclosed in this embodiment is On the computer, obtaining a processing task request from a user; In response to the request, a step of planning which distributed server will process which task and for how long based on the energy cost efficiency of the distributed server; A computer-readable program for executing the above is provided. [Explanation of symbols]

[0077] 1. Distributed Processing System 2. Distributed Server 3. Computer 4. Green electricity 5. Storage battery 6 Grid power 7 users 8 User terminal 10 Network 100, 101, 102, 103 requests 110,111,112 Plan 120 processing tasks 130 Real-time Tasks 140 Non-real-time tasks 150,151 Task completion deadline information 160 Performance Information 170 Electricity Price Information 180 Input Information

Claims

1. A distributed processing system comprising two or more distributed servers installed in multiple geographically dispersed locations and a computer that controls the distributed servers, The distribution server operates on power supplied from at least one of green power generated at the location where the distribution server is installed, a storage battery, and grid power; The computer an acquisition unit that acquires a processing task request from a user; A distributed processing system comprising: a planning unit that, in response to the request, plans which distributed server will process which processing task and for how long, based on the energy cost efficiency of the distributed server, which is calculated by dividing the amount of tasks that the distributed server can process per unit time by the current and future electricity prices.

2. The distributed processing system described in claim 1, wherein the computer promotes charging of the storage battery and loading of processing tasks of the distributed server when the power supply exceeds the power demand in the area of ​​the distributed server, and suppresses charging of the storage battery and loading of processing tasks of the distributed server when the power demand in the area exceeds the power supply.

3. The computer further comprises a classification unit that classifies the processing tasks into real-time tasks that are processed in real time and non-real-time tasks in response to the request; 3. The distributed processing system according to claim 1, wherein the planning unit plans which distributed server will process the classified real-time tasks and non-real-time tasks and for how long based on the energy cost efficiency of the distributed server.

4. the acquisition unit acquires input information consisting of performance information including at least the current and future CPU utilization rate, memory utilization rate, and communication latency of the distributed server, storage battery status information including the amount of electricity stored in the storage battery that supplies power to the distributed server, and green power status information including the amount of electricity generated by the green power that supplies power to the distributed server, and electricity price information including at least information on the wholesale electricity price of the grid electricity and a predicted future wholesale electricity price; 4. The distributed processing system of claim 3, wherein the planning unit plans the processing of the processing task obtained based on the energy cost efficiency for each distributed server, which is derived by dividing the task processing amount per unit power calculated from the performance information by the current and future power prices calculated from the power price information.

5. A distributed processing method executed by a computer that controls two or more distributed servers installed in multiple geographically dispersed locations, which operate on power supplied from at least one of green power, storage batteries, and grid power, comprising: receiving a processing task request from a user; and in response to the request, planning which distributed server will process which processing task and for how long, based on the energy cost efficiency of the distributed server, which is calculated by dividing the amount of tasks that the distributed server can process per unit time by the current and future electricity prices.

6. On the computer, receiving a processing task request from a user; In response to the request, a step of planning which distributed server will process which of the processing tasks and for what period of time, based on the energy cost efficiency of each distributed server, which is calculated by dividing the amount of tasks that the distributed server can process per unit time by the current and future electricity prices; A computer-readable program for executing the program.

Images