JP3896352B2 - Distributed computing system - Google Patents

Description

  The present invention relates to distributed computing systems. In particular, the present invention relates to a distributed computing system aimed at reducing power consumption.

  In recent years, attention has been paid to a technique for effectively using power for the purpose of effectively using resources and reducing the influence on the natural environment. Conventionally, as a technique for effectively using power, a power saving technique for reducing power consumption and a technique for effectively using surplus power are known.

  For example, as a power saving technique for reducing the power consumption of the information processing apparatus, a technique of adjusting the driving frequency or the driving voltage according to the operating state of the information processing apparatus is used. As an example, SpeedStep (registered trademark) of Intel Corporation, LongRun (registered trademark) of Trans Meta Corporation, or a technology for collectively controlling a plurality of information processing apparatuses on a network (see Patent Document 1) is used. Yes.

  Further, as a technique for effectively using surplus power, a technique for leveling power consumption has been proposed. For example, a technique for preferentially using a precharged battery when power consumption is concentrated in a predetermined area is proposed as a technique for leveling the bias of power consumption according to time zones (see Patent Document 2). Has been. In addition, as a method for leveling out the bias of power consumption according to the region, a method of purchasing electricity from a region having surplus power (see Patent Document 3) has been proposed.

  In recent years, attention has been paid to a grid computing system that allows a plurality of information processing apparatuses having surplus computing power to execute a calculation process that requires an enormous amount of time or a calculation process that requires an enormous amount of memory space.

JP 2000-259292 A JP 2002-271992 A JP 2002-58160 A

  However, according to the conventional grid computing technology, it is possible to speed up the calculation process by effectively using an information processing apparatus having surplus computing power, while using a large amount of computer resources in parallel. Thereby, when the information processing apparatus which performs calculation processing concentrates on a predetermined area, the power consumption of the area increases.

  According to the power saving technique, the information processing apparatus reduces power consumption by suppressing the calculation execution speed regardless of the amount of calculation processing or the content of the processing. However, since it is impossible to adjust the power consumption according to the content of processing, such as processing at high speed in accordance with the importance of processing, it has been difficult to apply to grid computing technology. In addition, the above technique for leveling power consumption may cause a large amount of power loss due to charging of storage batteries or long-distance transportation of electricity.

  According to the first aspect of the present invention, a plurality of calculation processing devices that respectively execute a plurality of partial calculations constituting a predetermined calculation processing, a calculation requesting device that requests partial calculations to a plurality of calculation processing devices, Provided corresponding to each of a lower level management device provided corresponding to each of a plurality of processing device groups into which a plurality of calculation processing devices are classified, and corresponding to each of a plurality of management device groups classified into a plurality of lower level management devices. A power management index information acquisition means for acquiring power processing index information indicating power consumption per predetermined calculation processing amount for each of a plurality of calculation processing devices, and a plurality of parts For each calculation, the power consumption index information indicating the amount of power required for the partial calculation compared to the case where another partial calculation is processed in the same calculation processing device. Power condition indicator information generating means for generating power, condition information acquisition means for acquiring power amount condition information indicating a condition to be satisfied by the amount of power consumed when a plurality of calculation processing devices execute a plurality of partial calculations, and a plurality of Corresponding to each of the partial calculations, a calculation processing device that satisfies the power amount condition information when executing the partial calculation among a plurality of calculation processing devices is selected based on the power processing index information and the power consumption index information A selection means, a partial calculation request means for causing each of the plurality of partial calculations to be executed by the calculation processing device selected in association with the partial calculation by the selection means, and a result of the partial calculation received from the plurality of calculation processing devices. Processing result output means for outputting a set as a result of calculation processing, and each of the plurality of calculation processing devices processes a partial calculation requested by the calculation requesting device. A fraction calculation processing means, a processing result sending means for sending the processing result of the partial calculation to the calculation requesting device, and the total power consumption of the corresponding processing device group as a predetermined reference among the maximum suppliable power in the processing device group When it is determined that the ratio has been reached, the partial calculation is transmitted to another calculation processing device provided in a region different from the region in which the calculation processing device is provided via the lower layer management device and the upper layer management device. A distributed computing system having partial calculation reassignment means is provided.

  The above summary of the invention does not enumerate all the necessary features of the present invention, and sub-combinations of these feature groups can also be the invention.

  According to the present invention, power can be used effectively.

  Hereinafter, the present invention will be described through embodiments of the invention. However, the following embodiments do not limit the invention according to the scope of claims, and all combinations of features described in the embodiments are included. It is not necessarily essential for the solution of the invention.

  FIG. 1 shows a block diagram of a distributed computing system 10. The distributed computing system 10 causes the calculation processing devices 30-1 to 30-N to execute a plurality of partial calculations constituting the predetermined calculation processing 700 and reduces the power consumption of the calculation processing devices 30-1 to 30-N. With the goal.

  The distributed computing system 10 includes a calculation requesting device 20 that requests the calculation processing devices 30-1 to 30 -N to perform partial calculation processing, a calculation processing devices 30-1 to 30 -N that respectively execute a plurality of partial calculations, and a calculation processing device. The lower tier management devices 40-1 to 40-M that manage 30-1 to N and the upper tier management devices 50-1 to 50L that manage the lower tier management devices 40-1 to M are provided.

  The calculation requesting device 20 includes, for each of the calculation processing devices 30-1 to 30-N, power processing index information indicating power consumption per predetermined calculation processing amount and stability indicating stability of power supplied to the calculation processing device. Information is acquired from the lower layer management devices 40-1 to 40-M and the upper layer management devices 50-1 to 50-L. Then, the calculation requesting device 20 causes each of the plurality of partial calculations to be executed by any of the calculation processing devices 30-1 to 30-N based on the power processing index information and the stability information.

  Here, the calculation process or the partial calculation is an application program that outputs a desired calculation result by being executed by the information processing apparatus, for example. Alternatively, the calculation process or partial calculation may be a subroutine, a function, a procedure, or a method.

  The lower layer management devices 40-1 to 40-M are provided corresponding to each of the plurality of processing device groups into which the calculation processing devices 30-1 to 30-N are classified, and the power consumption index information of each calculation processing device in the processing device group. Manage. For example, the lower layer management device 40-1 is provided corresponding to the processing device group 35-1 in which the calculation processing device 30-1 and the calculation processing device 30-2 are classified, and each calculation in the processing device group 35-1 is performed. It manages the power consumption index information of the processing device. Further, when the amount of power consumed in the corresponding processing device group exceeds a predetermined reference amount, the lower layer management devices 40-1 to 40-M notify the management processing devices to that effect.

  The upper layer management devices 50-1 to 50-L are provided corresponding to the plurality of management device groups into which the lower layer management devices 40-1 to 40-M are classified, and each lower layer management device in the management device group is provided. It manages power processing comprehensive information that is a value between the maximum value and the minimum value of the power consumption index information to be managed. For example, the upper layer management device 50-1 is provided corresponding to the management device group 45-1 in which the lower layer management device 40-1, the lower layer management device 40-2, and the like are classified, and the lower layer management device 40-1 is provided. A value between the maximum value and the minimum value of the power consumption index information managed by the server and a value between the maximum value and the minimum value of the power consumption index information managed by the lower layer management apparatus 40-2 Manage as.

  As an example, the upper tier management apparatus 50-1 is associated with each of the lower tier management apparatus 40-1 and the lower tier management apparatus 40-2, and is a power processing index managed by the lower tier management apparatus as power processing comprehensive information. An average value of information may be managed. Instead, the upper layer management apparatus 50-1 may manage the dispersion value or the standard deviation of the power processing index information as power processing comprehensive information, or may manage statistical information such as a frequency distribution. .

  FIG. 2 shows a block diagram of the calculation requesting device 20. The calculation requesting device 20 includes a power processing index information acquisition unit 200, a stability information acquisition unit 205, a computer environment database 210, a partial calculation request unit 215, a request history information storage unit 220, and a power consumption index information generation unit. 225, power consumption index information storage means 230, calculation content analysis means 240, condition information acquisition means 250, area information acquisition means 260, power charge information generation means 270, selection means 275, and processing result output means. 280.

  The power processing index information acquisition unit 200 transmits power processing index information indicating power consumption per predetermined calculation processing amount for each of the calculation processing devices 30-1 to 30-N via the upper layer management devices 50-1 to 50-L. Acquired from the lower layer management apparatuses 40-1 to 40 -M and stored in the computer environment database 210. Furthermore, when the power processing index information acquisition unit 200 receives the power processing index for any of the calculation processing devices 30-1 to 30 -N from the processing result output unit 280, the power processing index information acquisition unit 200 stores the power processing index information in the computer environment database 210. Update information. Further, the power processing index information acquisition unit 200 acquires the power processing comprehensive information from the upper layer management devices 50-1 to 50 -L and stores it in the computer environment database 210.

  The stability information acquisition unit 205 acquires, for each of the calculation processing devices 30-1 to 30 -N, stability information indicating the stability of the power supplied to the calculation processing device, and stores it in the computer environment database 210. For example, the stability information acquisition unit 205 receives, from the region information acquisition unit 260, regional information indicating in which of the plurality of predetermined regions each calculation processing device is provided, and is supplied to the regional information. Information indicating the stability of electricity is received from the upper level management devices 50-1 to 50 -N as stability information and stored in the computer environment database 210.

  Instead of this, the power processing index information acquisition unit 200 and the stability information acquisition unit 205 receive power processing index information and information from each of the calculation processing devices 30-1 to 30-N without going through the lower layer management device and the upper layer management device. Each of the stability information may be acquired.

  The request history information storage unit 220 stores information in which information for identifying a set of calculation processing devices that have performed the calculation processing is associated with the type of calculation processing that has already been performed by the calculation requesting device 20. For example, the type of calculation process is information indicating the nature of the calculation performed by the calculation process, and is determined by an API of an operating system that is called when the calculation process is executed.

  When the partial calculation request unit 215 obtains the calculation process 700 to be executed from the user, the partial calculation request unit 215 obtains information for identifying a set of calculation processing devices associated with the type of the calculation process 700 based on the type of the calculation process 700. The request history information storage means 220 is searched. If the search is successful, the partial calculation requesting unit 215 causes the searched set of calculation processing devices to perform the calculation process 700. If the search is not successful, the partial calculation request unit 215 sends a message to that effect to the power consumption index information generation unit 225.

  Then, the partial calculation request unit 215 transmits each of the plurality of partial calculations to the calculation processing device so that the selection processing unit 275 causes the calculation processing device selected in association with the partial calculation to execute. Instead, the partial calculation requesting unit 215 transmits each of the plurality of partial calculations to each of the calculation processing devices via the lower layer management devices 40-1 to M and the upper layer management devices 50-1 to L. May be.

  Further, the partial calculation requesting unit 215 sends, to each of the plurality of partial calculations, information for identifying a calculation processing device that executes the partial calculation, to the processing result output unit 280. Further, the partial calculation request unit 215 stores the types of calculation processes constituted by a plurality of partial calculations in the request history information storage unit 220 in association with a set of calculation processing devices that execute a plurality of partial calculations.

  The power consumption index information storage unit 230 associates the type of partial calculation with power consumption index information indicating the amount of power required for the partial calculation compared with the case where another partial calculation is processed in the same calculation processing device. Store.

  When the partial calculation requesting unit 215 cannot search the request history information storage unit 220 for information for identifying a set of calculation processing devices, the power consumption index information generating unit 225 executes the calculation processing 700 acquired from the user by a plurality of calculation processes 700. Divide into partial calculations. For each of the partial calculations, when the type of the partial calculation is stored in the power consumption index information storage unit 230, the power consumption index information generation unit 225 uses the power consumption associated with the type of the partial calculation. The index information is selected from the power consumption index information storage unit 230.

  On the other hand, when the type of the partial calculation is not stored in the power consumption index information storage unit 230, the power consumption index information generation unit 225 sends the partial calculation to the calculation content analysis unit 240. Then, the power consumption index information generation unit 225 generates power consumption index information based on the analysis result received from the calculation content analysis unit 240 and sends it to the selection unit 275. Further, the power consumption index information generating unit 225 stores the type of partial calculation in the power consumption index information storage unit 230 in association with the generated power consumption index.

  The calculation content analysis unit 240 analyzes the processing content of the partial calculation received from the power consumption index information generation unit 225 and returns it to the power consumption index information generation unit 225. Further, the calculation content analysis unit 240 generates critical path information obtained by analyzing whether or not the partial calculation is included in the critical path of the calculation process, and sends the critical path information to the selection unit 275. Here, the critical path refers to a group of partial calculations having the longest time required for the calculation among a plurality of partial calculations whose execution order is fixed and cannot be executed in parallel.

  The condition information acquisition unit 250 includes power amount condition information that should be satisfied when the calculation processing devices 30-1 to 30-N execute a plurality of partial calculations, and the calculation processing devices 30-1 to 30-N include a plurality of portions. Stability condition information indicating the condition of the stability of electricity received when the calculation is executed is acquired from the user and sent to the selection means 275.

  The area information acquisition unit 260 receives area information indicating which of the plurality of predetermined areas each of the calculation processing apparatuses 30-1 to 30-N is provided from the upper layer management apparatuses 50-1 to 50-N. The information is acquired and sent to the stability information acquisition unit 205, the power rate information generation unit 270, and the selection unit 275.

  The power charge information generating unit 270 is configured to calculate, for each of the calculation processing devices 30-1 to 30 -N, the power charge information indicating the power charge per power consumption of the calculation processing device from the regional information acquisition unit 260. And is sent to the selection means 275.

  The selection unit 275 associates each of the plurality of partial calculations with a calculation processing device that satisfies the power amount condition information and the stability condition information when executing the partial calculation among the calculation processing devices 30-1 to 30-N. Selection is made based on the power processing index information, stability information, power consumption index information, critical path information, area information, and power charge information, and the selection result is sent to the partial calculation request means 215.

More specifically, the selection unit 275 uses the power processing index information and the power processing general information managed by the lower layer management devices 40-1 to M and the upper layer management devices 50-1 to L according to the following procedure. A computing device may be selected. First, the selection unit 275 selects an upper layer management device, for example, the upper layer management device 50-1, in which any of the power processing comprehensive information to be managed satisfies the power amount condition information and the stability condition information. And the selection means 275 is a lower hierarchy management apparatus in which any of the power processing index information to be managed satisfies the power amount condition information and the stability condition information among the lower hierarchy management apparatuses managed by the upper hierarchy management apparatus 50-1. For example, the lower layer management apparatus 40-1 is selected. Further, the selection unit 275 selects a calculation processing device that satisfies the power amount condition information and the stability condition information, for example, the calculation processing device 30-1 among the calculation processing devices managed by the lower layer management device 40-1. .
Thereby, even if it is a case where there are a huge number of calculation processing devices 30-1 to 30-N, it is possible to efficiently search for an appropriate calculation processing device.

  The processing result output means 280 receives from the partial calculation request means 215 information for identifying a calculation processing device that executes a plurality of partial calculations. Then, the processing result output means 280 outputs a set of partial calculation results received from these calculation processing devices to the user as a result of the calculation processing 700. In addition to this, the processing result output means 280 may output the result of further calculation using the received set of partial calculation results as the result of the calculation processing 700.

  Further, the processing result output means 280 associates the actual power consumption information indicating the amount of power actually consumed when each partial calculation is processed with the information for identifying the calculation processing device that has executed the partial calculation. Are acquired from the calculation processing devices 30-1 to 30 -N and sent to the power processing index information acquisition means 200. In addition, the processing result output unit 280 may output the actual power consumption information to the user in association with information for identifying the calculation processing device. In this case, the user can know the total amount of power used in a predetermined area, for example. As a result, the user can determine whether or not the content of the contract with the electric power company is appropriate as compared with the current total amount of power, and can change the contract as necessary.

  FIG. 3 is a block diagram of the calculation processing device 30-1. The calculation processing device 30-1 includes a partial calculation processing unit 300, a processing result transmission unit 310, an actual consumption transmission unit 320, and a partial calculation reassignment unit 330. Since each of the calculation processing devices 30-2 to 30-N has substantially the same configuration as the calculation processing device 30-1, only the differences will be described.

  The partial calculation processing unit 300 processes the partial calculation requested from the calculation requesting device 20 and sends the processing result to the processing result transmission unit 310. In response to this, the processing result transmission unit 310 transmits the processing result of the partial calculation to the calculation requesting device 20. Further, the actual consumption amount transmitting unit 320 generates power actual consumption amount information indicating the amount of power consumed when the partial calculation processing unit 300 processes the partial calculation by analyzing the processing content of the partial calculation processing unit 300. Then, it is transmitted to the calculation requesting device 20 in association with the information for identifying the partial calculation.

  An example of processing for generating actual power consumption information is shown below. First, the actual consumption amount transmitting unit 320 measures the power consumption amount of the calculation processing device 30-1 in the period in which the partial calculation is executed by a predetermined program for measuring the power consumption amount. Then, the actual consumption transmission means 320 refers to the statistical information of the operating system, and thereby the ratio of the time when the CPU is used by the partial calculation to the time when the CPU is used by the processing of the entire calculation processing device 30-1. Is identified. As a result, the actual consumption transmission means 320 can estimate the power consumption used by the partial calculation and generate it as the actual power consumption information.

  Moreover, the said measuring method about electric power consumption is an example, You may use another method. For example, the actual consumption transmission means 320 may generate the actual power consumption information using an API defined in the operating system, or generate the actual power consumption information using a system call defined in the BIOS. May be. Further, each of the calculation processing devices 30-2 to 30-N may generate the actual power consumption information by a method different from that of the calculation processing device 30-1.

  In response to the notification from the lower layer management apparatus 40-1, the partial calculation reassignment unit 330 determines in advance the total power consumption in the area where the calculation processing apparatus 30-1 is provided among the maximum suppliable power in the area. It is determined whether or not the specified standard ratio has been reached. When the partial calculation reassignment unit 330 determines that the total power consumption of the area has reached the reference ratio, the partial calculation processing unit 300 causes the partial calculation processing unit 300 to stop the partial calculation processing, and the partial calculation processing unit 300 processes the partial calculation. Is obtained from the partial calculation processing means 300. Then, the partial calculation reassignment unit 330 executes the partial calculation in another calculation processing device provided in a region different from the region, for example, the calculation processing device 30-N. 1 to send. In response to this, the lower hierarchy management device 40-1 transmits the partial calculation to the calculation processing device 30-N via the upper hierarchy management device 50-1 and the lower hierarchy management device 40-2.

  As described above, as illustrated in FIGS. 1 to 3, the distributed computing system 10 allows each of the computing devices that execute each of the plurality of partial calculations constituting the predetermined calculation process to be performed per predetermined calculation processing amount. Selection is made so as to satisfy the power processing index information indicating the power consumption and the stability information indicating the stability of the power supplied to the calculation processing device. Thereby, while satisfy | filling the conditions of power consumption, a calculation process can be performed at high speed and highly efficient.

  FIG. 4 shows an operation flow in which the distributed computing system 10 executes calculation processing. The power processing index information acquisition unit 200 acquires power processing index information for each of the calculation processing devices 30-1 to 30-N (S400). For example, the power processing index information acquisition unit 200 uses the power consumption per predetermined processing amount for at least one of the CPU, the hard disk drive, and the input / output device of the calculation processing apparatus as the power processing index information of each calculation processing apparatus. Get the performance information shown.

  Specifically, the power processing index information acquisition unit 200 may acquire information indicating the thermal design power consumption (TDP: Thermal Design Power) of the CPU and the type of CPU as performance information. Thereby, the calculation request | requirement apparatus 20 can estimate the power consumption per CPU 1 clock as power process parameter | index information.

  For example, in the Pentium (registered trademark) 4 processor (Northwood) of Intel Corporation, a plurality of types of processors having different operating frequencies have different TDPs. As an example, in a processor operating at 3.0 Ghz, the TDP is 81.9 W, and in a processor operating at 2.4 Ghz, the TDP is 57.8 W. Therefore, in the 3.0 Ghz operation processor, the power consumption per 1 Ghz is 27.3 W, and in the 2.4 Ghz operation processor, the power consumption per 1 Ghz is 24.1 W. As described above, the power processing index information acquisition unit 200 acquires the CPU TDP and type set as the power processing index information, so that the calculation requesting device 20 can estimate the power consumption per clock.

  In addition, the power processing index information acquisition unit 200 uses the power consumption of the calculation processing apparatus and the calculation processing apparatus when the calculation processing apparatuses 30-1 to 30-N perform a predetermined amount of calculation as the power processing index information. You may acquire the sum total of the power consumption of the network apparatus to which is connected. That is, the power processing index information is not limited to the power consumption of only the calculation processing device, but is the power consumption in a series of processes in which the calculation processing device performs the calculation requested from the calculation requesting device 20 and transmits the calculation result.

  The stability information acquisition unit 205 acquires stability information for each of the calculation processing devices 30-1 to 30-N (S410). For example, the stability information acquisition unit 205 acquires a value obtained by multiplying the average number of instantaneous interruption occurrences per unit time by the average instantaneous interruption time from the upper hierarchy management devices 50-1 to 50 -N as the stability information. Further, the stability information acquisition unit 205 stabilizes information indicating the power stability in the area where each of the calculation processing devices 30-1 to 30 -N is provided based on the area information acquired from the area information acquisition unit 260. It may be acquired as sex information.

  Furthermore, the stability information acquisition unit 205 may acquire information indicating whether or not the calculation processing device is managed by an uninterruptible power supply using a battery or a generator as stability information in each calculation processing device. . In this case, the selection means 275 is a calculation processing device with high power stability compared to a calculation processing device managed by the uninterruptible power supply device compared to a calculation processing device not managed by the uninterruptible power supply device. Judgment can be made.

  When the partial calculation requesting unit 215 obtains the calculation process 700 to be executed from the user (S420), the partial calculation requesting unit 215 identifies a set of calculation processing devices associated with the type of the calculation process 700 based on the type of the calculation process 700. Information to be searched is retrieved from the request history information storage means 220 (S430).

  When the search is successful, that is, when the type of the calculation processing 700 is a type that has already been processed by the calculation requesting device 20 (S440: YES), the partial calculation requesting unit 215 determines that the searched calculation processing device The set is subjected to the calculation process 700 (S450), and the process is terminated. As a result, the calculation requesting apparatus 20 can efficiently operate by omitting the process of selecting a calculation processing apparatus on which partial calculation is to be performed when repeatedly performing the same type of calculation processing. For example, when a plurality of calculation processes in which only some of the parameters are different from each other are repeated, for example, when performing Monte Carlo simulation, it is efficient.

  When the search cannot be performed, that is, when the type of the calculation process 700 is a type that has not yet been processed by the calculation requesting device 20 (S440: NO), the power consumption index information generation unit 225 executes the calculation process 700. Dividing into a plurality of partial calculations (S460). Subsequently, the calculation requesting apparatus 20 executes the following processing for each of the plurality of partial calculations (S470).

When the partial calculation type is stored in the power consumption index information storage unit 230 (S480: YES), the power consumption index information generation unit 225 stores the consumption stored in association with the partial calculation type. The power index information is acquired from the power consumption index information storage unit 230 (S490). On the other hand, the power consumption index information generation means 225 generates power consumption index information when the partial calculation type is not stored in the power consumption index information storage means 230 (S480: NO) (S485). Details will be described later.
The calculation requesting device 20 executes the above processing for each of the plurality of partial calculations (S495).

  FIG. 5 shows an operation flow subsequent to FIG. 4 in the distributed computing system 10. The condition information acquisition unit 250 includes power amount condition information that should be satisfied when the calculation processing devices 30-1 to 30-N execute a plurality of partial calculations, and the calculation processing devices 30-1 to 30-N include a plurality of portions. Stability condition information indicating the condition of the stability of electricity received when the calculation is executed is acquired from the user (S500).

  For example, the condition information acquisition unit 250 acquires a power upper limit value that defines an upper limit of power consumed per unit time in a plurality of calculation processing devices provided in a predetermined region as the power amount condition information. Instead of this, the condition information acquisition unit 250 may acquire the upper limit value of the total power consumption consumed from the start to the end of the calculation process as the power amount condition information. In addition, the condition information acquisition unit 250 may acquire the power amount condition information that sets the power consumption to a predetermined amount or less under the condition that the execution time of the critical path of the calculation process is within a predetermined time, You may acquire the electric energy condition information made into below predetermined.

  The area information acquisition unit 260 acquires area information indicating which of the plurality of predetermined areas each of the calculation processing devices 30-1 to 30-N is provided (S510). For example, the regional information acquisition unit 260 includes, as regional information in each of the calculation processing devices 30-1 to 30-N, information indicating the country in which the calculation processing device is provided, and a power company that supplies power to the calculation processing device. Information indicating a network domain to which the calculation processing device belongs, information indicating a building in which the calculation processing device is provided, or information indicating a floor or a room in which the calculation processing device is provided may be acquired. .

  The power charge information generation unit 270 generates power charge information indicating the power charge per power consumption for each of the calculation processing devices 30-1 to 30-N based on the regional information (S520). For example, based on the time zone in which the calculation process is performed and the time zone of the region determined by the region information, the power rate information generation unit 270 determines whether the time zone in which the calculation process is performed is daytime or nighttime in the region. The electricity bill information is generated in association with each of the calculation processing devices 30-1 to 30-N. More specifically, the power rate information generation unit 270 may acquire the current time acquired from the NTP server as a time zone during which the calculation process is performed. In addition, the power rate information generation unit 270 may determine a time zone of a region determined by the region information based on a database in which a time difference is recorded in association with each country or region.

  The selection unit 275 associates each of the plurality of partial calculations with a calculation processing device that satisfies the power amount condition information and the stability condition information when executing the partial calculation among the calculation processing devices 30-1 to 30-N. A selection is made based on power processing index information, stability information, power consumption index information, area information, and power rate information (S525). An example of the selection method is shown below.

  When the condition information acquisition unit 250 acquires the power upper limit value in a plurality of calculation processing devices provided in a predetermined region as the power amount condition information, the selection unit 275 indicates that the plurality of calculation processing devices in the predetermined region When the power consumed per unit time by processing a plurality of partial calculations is less than or equal to the power upper limit value acquired by the condition information acquisition unit 250 (for example, a value specified by wattage), the plurality of partial calculations are performed. A plurality of calculation processing devices in the predetermined area may be selected in association with each of the above.

  In this case, preferably, the condition information acquisition unit 250 sets a value obtained by subtracting a predetermined amount from the power upper limit value instead of the power upper limit value as a power amount that can be consumed in the predetermined area. That is, when the power consumed per unit time is equal to or less than a value obtained by subtracting a predetermined amount from the power upper limit value, the selection unit 275 preferably selects a plurality of calculation processing devices in the predetermined area. Thereby, even if the power processing index information or the power consumption index information has an error compared with the actual power consumption, it is possible to prevent the actual power consumption from exceeding the power upper limit value.

  As another example, the selection unit 275 is a calculation processing device that executes each of a plurality of partial calculations. In the area where the calculation processing device is provided, the calculation processing in which the partial calculation request unit 215 requests partial calculation is nighttime. A device is selected in preference to a calculation processing device whose time is daytime. This makes it possible to effectively use the surplus power at night and to save the electricity bill in the distributed computing system 10 as a whole. Here, the night refers to a time zone in which a night electricity charge is applied, which is determined by a company that supplies power. In addition, daytime means a time zone in which daytime electricity charges are applied. In addition, the distinction between nighttime and daytime is an example, and the selection unit 275 is configured so that the calculation processing device in the time zone in which power consumption in the predetermined area is smaller than the calculation processing device in the time zone in which power consumption is higher. You should select with priority.

  As yet another example, the selection unit 275 compares the calculation processing device that satisfies the power amount condition information and the stability condition information with a calculation processing device that selects and matches the partial calculation that is not included in the critical path. A calculation processing device having a larger power consumption and a faster processing speed indicated by the power processing index information is selected in association with the partial calculation included in the critical path of the calculation processing. As a result, the calculation in the critical path can be quickly processed even when the power consumption is large, so that the processing speed can be increased while satisfying the power amount condition information and the stability condition information.

  In addition, the selecting unit 275 associates the stability information with the partial calculation included in the critical path of the calculation processing device, as compared with the calculation processing device selected in association with the partial calculation not included in the critical path. Select a computing device that exhibits greater stability. Thereby, it is possible to prevent the completion of the calculation process from being delayed due to the interruption or recalculation of the calculation due to an instantaneous power interruption or the like.

  Subsequently, the partial calculation request unit 215 transmits an instruction to start execution of the partial calculation for each of the plurality of partial calculations to the calculation processing device selected by the selection unit 275 in association with the partial calculation ( S530). For example, the partial calculation request unit 215 transmits the partial calculation to the calculation processing device 30-1 selected in association with the partial calculation. Instead, the partial calculation request unit 215 may transmit the partial calculation to the upper hierarchy management device selected by the selection unit 275, for example, the upper hierarchy management device 50-1. In this case, the upper hierarchy management apparatus 50-1 transfers the partial calculation to the lower hierarchy management apparatus 40-1 that manages the calculation processing apparatus that executes the partial calculation. Furthermore, the lower layer management apparatus 40-1 transfers the partial calculation to the selected calculation processing apparatus. That is, the partial calculation request unit 215 may transmit the partial calculation to the calculation processing device via the upper hierarchy management device and the lower hierarchy management device.

  Here, when a plurality of calculation processing devices in the same region are selected as calculation processing devices for processing a plurality of partial calculations, preferably, the partial calculation requesting means 215 is assigned to one calculation processing device in the same region. The timing for starting execution of each of the plurality of partial calculations is adjusted so that the time at which the partial calculation is started and the time at which the other calculation processing device in the region starts the partial calculation are different. As a result, when the power consumption becomes higher at the start of the calculation than after the start of the calculation, for example, even when a hard disk drive or the like is driven at the start of the calculation, the instantaneous maximum power consumption in the same region can be reduced. it can.

  More preferably, the partial calculation requesting unit 215 starts the partial calculation that requires a longer processing time earlier than the partial calculation that requires a shorter processing time. Thereby, the time required from the start of calculation to the end of each of the plurality of partial calculations can be shortened.

  Subsequently, the partial calculation requesting unit 215 stores the types of calculation processing constituted by a plurality of partial calculations in the request history information storage unit 220 in association with a set of calculation processing devices that execute the plurality of partial calculations ( S533).

  When receiving an instruction to start execution of partial calculation (S535: YES), the partial calculation processing means 300 executes partial calculation (S540). The partial calculation reassignment unit 330 determines whether or not the total power consumption of the area where the calculation processing device 30-1 is provided has reached a predetermined reference ratio among the maximum suppliable power in the area ( S545).

  When the partial calculation reassignment unit 330 determines that the total power consumption of the area has reached the reference rate (S545: YES), the partial calculation processing unit 300 causes the partial calculation processing unit 300 to stop the partial calculation processing, and the partial calculation processing unit 300 The partial calculation to be processed in step S3 is acquired from the partial calculation processing unit 300 and transmitted to another calculation processing device (S550). Thereby, the calculation processing apparatus 30-1 can prevent the concentration of power consumption in a predetermined area autonomously regardless of the processing of the calculation requesting apparatus 20.

  When it is determined that the total power consumption in the area has not reached the reference ratio (S545: NO), it is determined whether or not the partial calculation is completed (S555). When the partial calculation is not completed (S555: NO), the calculation processing device 30-1 returns the process to S540 and continues the partial calculation process.

  On the other hand, when the partial calculation is completed (S555: YES), the processing result transmitting unit 310 transmits the processing result of the partial calculation to the calculation requesting apparatus 20 (S560). Further, the actual consumption amount transmitting unit 320 generates power actual consumption amount information indicating the amount of power consumed when the partial calculation processing unit 300 processes the partial calculation by analyzing the processing content of the partial calculation processing unit 300. Then, it is transmitted to the calculation requesting apparatus 20 in association with the information for identifying the partial calculation (S575).

  When the processing result output unit 280 receives processing results for each of the plurality of partial calculations from the calculation processing devices 30-1 to 30 -N (S 565: YES), the processing result output unit 280 As a result, it outputs to the user (S570). In addition, the power processing index information acquisition unit 200 uses the actual power consumption information indicating the amount of power actually consumed when each partial calculation is processed as information for identifying the calculation processing device that has executed the partial calculation. Correspondingly, it is acquired from the calculation processing devices 30-1 to 30-N via the processing result output means 280 (S580). Then, the power processing index information acquisition unit 200 stores the received actual power consumption as power processing index information in the computer environment database 210 (S590).

  FIG. 6 shows the details of the operation flow in S485. The calculation content analysis means 240 analyzes the number of instructions included in each partial calculation for each of the plurality of partial calculations (S600). For example, the calculation content analysis unit 240 may analyze the number of execution steps in the execution path predicted to be executed in the partial calculation as the number of instructions, or may analyze the number of instructions in the entire partial calculation. . Moreover, the calculation content analysis means 240 may analyze the number of instructions using the induction variable in the loop processing.

  In addition, the calculation content analysis unit 240 analyzes the number of times the input / output device is accessed in the partial calculation processing for each of the plurality of partial calculations (S610). For example, the calculation content analysis unit 240 may analyze the number of accesses based on the number of API calls used to access the input / output device.

  Furthermore, the calculation content analysis means 240 analyzes the size of the transfer data transferred by the input / output device in the partial calculation process for each of the partial calculations (S620). For example, the calculation content analysis means 240 may analyze the size of transfer data based on the value of an API argument used to access the input / output device.

  In addition, the calculation content analysis unit 240 analyzes, for each of the plurality of partial calculations, whether or not the partial calculation is included in the critical path in the calculation process (S630). Further, the calculation content analysis means 240 executes, for each of the plurality of partial calculations, a benchmark program that is another process that is the same type of process as the partial calculation and has a shorter processing time than the partial calculation ( S640).

  Subsequently, the power consumption index information generating unit 225 generates power consumption index information by the following processing based on the analysis result (S650). For example, the power consumption index information generating unit 225 compares each partial calculation with the other partial calculations when the partial calculation has a larger number of instructions than the other partial calculations. Power consumption index information indicating that the power consumption is large is generated.

  In addition, the power consumption index information generating unit 225 compares each partial calculation with the other partial calculations when the number of accesses is larger than that of the other partial calculations. To generate a power consumption index indicating that the power consumption is large. In addition, the power consumption index information generation unit 225 compares each partial calculation with another partial calculation when the size of the transfer data is larger than that of the other partial calculation. Then, a power consumption index indicating that the power consumption is large is generated.

  For example, the power consumption index information generation unit 225 may calculate the amount of CPU computation required for the partial calculation compared to the case where each partial calculation is processed in the same calculation processing device for each partial calculation. Processing content information indicating at least one of the access amount to the device and the usage amount of the input / output device may be generated as power consumption index information. The power consumption index information generating unit 225 may generate power consumption index information based on the amount of power consumed as a result of executing the benchmark program.

  Subsequently, the power consumption index information generation unit 225 stores the type of partial calculation in the power consumption index information storage unit 230 in association with the generated power consumption index (S660).

  FIG. 7 shows details of the calculation process 700. The calculation process 700 includes application 1, application 2, application 3, application 4, application 5, application 6, and application 7, each of which is an example of partial calculation. A solid line arrow in this figure indicates a calling relationship between applications. For example, during processing by the application 1, the applications 2 to 4 are sequentially called and executed. Further, the power consumption index information generating unit 225 generates each of the power consumption index information A1 to A7 in association with each of the applications 1 to 7.

  Further, each of the applications includes a plurality of jobs, each of which is an example of partial calculation. For example, the application 4 includes job 1, job 2, job 3, job 4, job 5, and job 6. As in the case of the application, job 2 and job 3 are sequentially called and executed during processing by job 1. The power consumption index information generating unit 225 generates each of the power consumption index information J1 to J6 in association with each of the jobs 1 to 6.

  As described above, the calculation processing 700 is classified into a plurality of partial calculations for each unit that is called and executed, and the calculation requesting device 20 assigns each of these partial calculations to the calculation processing devices 30-1 to 30-N. Run in either. For example, in the case where the execution order relationship between the application 2, the application 3, and the application 4 is not determined, the calculation requesting device 20 performs three different calculation processes for the application 2, the application 3, and the application 4, respectively. The processing can be speeded up by executing in parallel by the apparatus.

  FIG. 8 shows an example of the hardware configuration of the calculation requesting device 20. The calculation requesting device 20 includes a CPU peripheral unit including a CPU 1000, a RAM 1020, a graphic controller 1075, and a display device 1080 connected to each other by a host controller 1082, and a communication interface 1030 connected to the host controller 1082 by an input / output controller 1084. An input / output unit having a hard disk drive 1040 and a CD-ROM drive 1060, and a legacy input / output unit having a ROM 1010, a flexible disk drive 1050 and an input / output chip 1070 connected to the input / output controller 1084 are provided.

  The host controller 1082 connects the RAM 1020 to the CPU 1000 and the graphic controller 1075 that access the RAM 1020 at a high transfer rate. The CPU 1000 operates based on programs stored in the ROM 1010 and the RAM 1020, and controls each unit. The graphic controller 1075 acquires image data generated by the CPU 1000 or the like on a frame buffer provided in the RAM 1020 and displays it on the display device 1080. Alternatively, the graphic controller 1075 may include a frame buffer that stores image data generated by the CPU 1000 or the like.

  The input / output controller 1084 connects the host controller 1082 to the communication interface 1030, the hard disk drive 1040, and the CD-ROM drive 1060, which are relatively high-speed input / output devices. The communication interface 1030 communicates with an external device via a network such as a fiber channel. The hard disk drive 1040 stores programs and data used by the calculation requesting device 20. The CD-ROM drive 1060 reads a program or data from the CD-ROM 1095 and provides it to the input / output chip 1070 via the RAM 1020.

  The input / output controller 1084 is connected to the ROM 1010 and relatively low-speed input / output devices such as the flexible disk drive 1050 and the input / output chip 1070. The ROM 1010 stores a boot program executed by the CPU 1000 when the calculation requesting device 20 is started up, a program depending on the hardware of the calculation requesting device 20, and the like. The flexible disk drive 1050 reads a program or data from the flexible disk 1090 and provides it to the input / output chip 1070 via the RAM 1020. The input / output chip 1070 connects various input / output devices via a flexible disk 1090 and, for example, a parallel port, a serial port, a keyboard port, a mouse port, and the like.

  The program provided to the calculation requesting device 20 is stored in a recording medium such as the flexible disk 1090, the CD-ROM 1095, or an IC card and provided by the user. The program is read from the recording medium via the input / output chip 1070 and / or the input / output controller 1084, installed in the calculation requesting apparatus 20, and executed.

  The programs installed and executed in the calculation requesting device 20 include a power processing index information acquisition module, a stability information acquisition module, a partial calculation request module, a power consumption index information generation module, a calculation content analysis module, a condition An information acquisition module, a region information acquisition module, a power rate information generation module, a selection module, and a processing result output module are included. The operation that each module causes the calculation requesting apparatus 20 to perform is the same as the operation of the corresponding member in the calculation requesting apparatus 20 described with reference to FIGS.

  The program or module shown above may be stored in an external storage medium. As the storage medium, in addition to the flexible disk 1090 and the CD-ROM 1095, an optical recording medium such as a DVD or PD, a magneto-optical recording medium such as an MD, a tape medium, a semiconductor memory such as an IC card, or the like can be used. Further, a storage device such as a hard disk or a RAM provided in a server system connected to a dedicated communication network or the Internet may be used as a recording medium, and the program may be provided to the calculation requesting device 20 via the network.

  As described above, as is apparent from the present embodiment, the distributed computing system 10 divides a calculation process that requires enormous time into a plurality of partial calculations, and uses the calculation processors 30-1 to 30-N for the plurality of partial calculations. Process in parallel. Thereby, calculation processing can be performed at high speed. Further, the distributed computing system 10 generates information such as the power required for the partial calculation or the importance of the partial calculation by analyzing the processing contents of the partial calculation. Furthermore, the distributed computing system 10 acquires information on the power consumed per calculation amount in each calculation processing device. Based on these pieces of information, the distributed computing system 10 selects a calculation processing device on which partial calculation is to be executed. Thereby, according to the conditions acquired from a user, electric power consumption can be reduced, nighttime electric power can be used effectively, an electricity bill can be reduced, or the stability of calculation processing can be ensured.

  As mentioned above, although this invention was demonstrated using embodiment, the technical scope of this invention is not limited to the range as described in the said embodiment. It will be apparent to those skilled in the art that various modifications or improvements can be added to the above-described embodiment. It is apparent from the scope of the claims that the embodiments added with such changes or improvements can be included in the technical scope of the present invention.

  According to the embodiment described above, the distributed computing system, the control method, the program, and the recording medium shown in the following items are realized.

(Item 1) A distributed computing system comprising: a plurality of calculation processing devices that respectively execute a plurality of partial calculations constituting predetermined calculation processing; and a calculation requesting device that requests partial calculations from the plurality of calculation processing devices. The calculation requesting device includes, for each of the plurality of calculation processing devices, power processing index information acquisition means for acquiring power processing index information indicating power consumption per predetermined calculation processing amount, and the plurality of parts For each of the calculations, power consumption index information generating means for generating power consumption index information indicating the amount of power required for the process of the partial calculation compared to the case where another partial calculation is processed in the same calculation processing device, and the plurality A condition for acquiring power amount condition information indicating a condition to be satisfied by the amount of power consumed when the plurality of calculation processing devices execute the plurality of partial calculations. A calculation processing device that satisfies the power amount condition information when executing the partial calculation among the plurality of calculation processing devices in association with each of the plurality of partial calculations. Selection means for selecting based on the information and the power consumption index information; and a partial calculation requesting means for causing the calculation processing device selected by the selection means to associate each of the partial calculations with the partial calculation. And a processing result output means for outputting a set of partial calculation results received from the plurality of calculation processing devices as a result of the calculation processing, each of the plurality of calculation processing devices from the calculation requesting device. A distributed computer comprising: a partial calculation processing means for processing the requested partial calculation; and a processing result transmission means for transmitting the result of the partial calculation to the calculation requesting device. Computing system.
(Item 2) The power processing index information acquisition unit is configured to store at least one of a CPU, a hard disk drive, and an input / output device of the calculation processing device per predetermined processing amount as the power processing index information of each calculation processing device. The performance information indicating power consumption is acquired, and the power consumption index information generating unit performs processing of the partial calculation for each of the plurality of partial calculations compared to a case where another partial calculation is processed in the same calculation processing device. The distributed computing system according to item 1, wherein processing content information indicating at least one of an amount of computation of a CPU, an access amount to a hard disk drive, and a usage amount of an input / output device required for generating is used as the power consumption index information.

(Item 3) The calculation requesting device includes: region information acquisition means for acquiring region information indicating which of the plurality of calculation processing devices is provided in a plurality of predetermined regions; For each of the calculation processing devices, it further includes power rate information generating means for generating power rate information indicating a power rate per power consumption of the calculation processing device based on the regional information, and the selection unit includes: The distributed computing system according to item 1, wherein a calculation processing device that satisfies the power amount condition information is further selected based on the power rate information.
(Item 4) The power rate information generation means associates the electricity rate information with each of the plurality of calculation processing devices based on a time zone in which the calculation process is performed and a time zone of a region determined by the region information. The selection means is a calculation processing device that executes each of the plurality of partial calculations, and the time when the partial calculation requesting means in the area where the calculation processing device is provided requests the partial calculation is nighttime. 4. The distributed computing system according to item 3, wherein the processing device is selected with priority over the computing device whose time is daytime.
(Item 5) The calculation requesting apparatus further includes area information acquisition means for acquiring area information indicating which of the plurality of calculation processing apparatuses is provided in a plurality of predetermined areas. In the case where the condition information acquisition unit acquires, as the power amount condition information, a power upper limit value that defines an upper limit of power consumed per unit time in a plurality of calculation processing devices provided in a predetermined area, the selection When the power consumed per unit time by the plurality of calculation processing devices in the predetermined area processing the plurality of partial calculations is equal to or less than the power upper limit value, each of the plurality of partial calculations The distributed computing system according to item 1, wherein a plurality of calculation processing devices in the predetermined area are selected in association with each other.
(Item 6) When a plurality of calculation processing devices in the same region are selected as the calculation processing devices for processing the plurality of partial calculations, the partial calculation requesting unit is assigned to one calculation processing device in the same region. 6. The distributed computing system according to item 5, wherein the timing for starting execution of each of the plurality of partial calculations is adjusted so that the time for starting the partial calculation differs from the time for starting another partial calculation processing device in the region. .

(Item 7) The calculation requesting device further includes calculation content analysis means for analyzing the number of instructions included in the partial calculation for each of the plurality of partial calculations, and the power consumption index information generating means For each of the partial calculations, power consumption index information indicating that the partial calculation has a larger power consumption than the other partial calculations when the partial calculation has a larger number of instructions than the other partial calculations. The distributed computing system according to item 1, wherein:
(Item 8) The calculation requesting device further includes calculation content analysis means for analyzing the number of times the input / output device is accessed in the partial calculation processing for each of the plurality of partial calculations. The information generation means, for each of the plurality of partial calculations, when the partial calculation has a higher access count compared to other partial calculations, the partial calculation consumes less power than the other partial calculations. Item 4. The distributed computing system according to item 1, which generates power consumption index information indicating a large value.
(Item 9) The calculation requesting device further includes calculation content analysis means for analyzing the size of transfer data transferred by the input / output device in the partial calculation processing for each of the plurality of partial calculations. The index information generating means consumes each partial calculation when the partial calculation is larger than the other partial calculations and the transfer data is larger in size than the other partial calculations. Item 4. The distributed computing system according to item 1, which generates power consumption index information indicating that the amount of power is large.
(Item 10) When the power consumption index information generating unit executes, for each of the plurality of partial calculations, another process that is the same type of process as the partial calculation and has a processing time shorter than the partial calculation. Item 4. The distributed computing system according to item 1, wherein power consumption index information is generated based on an amount of power consumed by the device.
(Item 11) The calculation requesting device further includes, for each of the plurality of partial calculations, a calculation content analysis unit that analyzes whether or not the partial calculation is included in a critical path in the calculation process, and the selection unit Compared with a calculation processing device selected in association with a partial calculation not included in the critical path among calculation processing devices satisfying the power amount condition information, the power consumption amount indicated by the power processing index information is more 2. The distributed computing system according to item 1, wherein a computing device that is large and has a faster processing speed is selected in association with a partial calculation included in the critical path.

(Item 12) The calculation processing device associates the actual power consumption information indicating the amount of power consumed when the partial calculation processing unit processes the partial calculation with the information for identifying the partial calculation, and calculates the calculation. The power consumption index information acquisition means further transmits the actual power consumption information per partial calculation from the actual consumption amount transmission means as the power processing index information. Item 4. A distributed computing system according to item 1.
(Item 13) The calculation requesting device further includes power consumption index information storage means for storing a type of partial calculation and power consumption index information of the partial calculation in association with each other, and for each of the plurality of partial calculations. When the partial calculation type is stored in the power consumption index information storage unit, the power consumption index information generation unit displays the power consumption index information associated with the partial calculation type as the power consumption index. The distributed computing system according to item 1, wherein the power consumption index information of the partial calculation is generated by selecting from the information storage means.
(Item 14) The request history in which the calculation requesting device stores information in which the type of the calculation processing already performed by the calculation requesting device is associated with information for identifying the set of the calculation processing devices that have performed the calculation processing The partial calculation requesting means further includes information for identifying a set of calculation processing devices associated with the type of calculation processing based on the type of calculation processing to be executed. The distributed computing system according to item 1, wherein a search is performed from an information storage unit, and the set of calculated processing devices is caused to perform the calculation processing.
(Item 15) The distributed computing system according to item 1, wherein the partial calculation request means starts a partial calculation requiring a longer processing time earlier than a partial calculation requiring a shorter processing time.
(Item 16) When the calculation processing device determines that the total power consumption of the area where the calculation processing device is provided has reached a predetermined ratio of the maximum suppliable power in the region, the calculation processing device 2. The distributed computing system according to item 1, further comprising partial calculation reassignment means for causing partial calculation processed in the calculation processing means to be executed in another calculation processing device provided in an area different from the area.

(Item 17) A lower-layer management device that is provided corresponding to each of a plurality of processing device groups into which the plurality of calculation processing devices are classified, and that manages power processing index information of each calculation processing device in the processing device group; Between the maximum value and the minimum value of the power processing index information provided corresponding to each of the plurality of management device groups into which the plurality of lower layer management devices are classified, and managed by each lower layer management device in the management device group An upper tier management device that manages power processing total information that is a value, and the power processing index information acquisition means receives the power processing total information from the upper tier management device and the power processing from the lower tier management device. Each of the index information is acquired, and the selection unit selects an upper layer management device in which any of the power processing comprehensive information to be managed satisfies the power amount condition information, Among the lower tier management devices managed by the upper tier management device, a lower tier management device in which any of the power processing index information to be managed satisfies the power amount condition information is selected and managed by the lower tier management device. The distributed computing system according to item 1, wherein a computing processing device that satisfies the power consumption condition is selected from the computing processing devices.
(Item 18) A distributed computing system comprising: a plurality of calculation processing apparatuses that respectively execute a plurality of partial calculations constituting a predetermined calculation process; and a calculation requesting apparatus that requests partial calculations from the plurality of calculation processing apparatuses. The calculation requesting device includes, for each of the plurality of calculation processing devices, stability information acquisition means for acquiring stability information indicating stability of power supplied to the calculation processing device, and the plurality of calculation processing devices. In association with each of the plurality of partial calculations, a condition information acquisition unit that acquires stability condition information indicating a condition of the stability of electricity received when the calculation processing device executes the plurality of partial calculations. Selection for selecting a calculation processing device that satisfies the stability condition information when executing the partial calculation from among the plurality of calculation processing devices based on the stability information Means, a partial calculation requesting means for causing each of the plurality of partial calculations to be performed by a calculation processing device selected in association with the partial calculation by the selection means, and a partial calculation received from the plurality of calculation processing devices. Processing result output means for outputting a result set as a result of the calculation processing, and each of the plurality of calculation processing devices includes partial calculation processing means for processing a partial calculation requested by the calculation requesting device; And a processing result transmitting means for transmitting the processing result of the partial calculation to the calculation requesting device.

(Item 19) The stability information acquisition unit acquires, as the stability information in each calculation processing device, information indicating whether the calculation processing device is managed by an uninterruptible power supply using a battery or a generator. Item 19. A distributed computing system according to Item 18.
(Item 20) The calculation requesting apparatus further includes area information acquisition means for acquiring area information indicating which of the plurality of calculation processing apparatuses is provided in a plurality of predetermined areas. 19. The distributed computing according to item 18, wherein the stability information acquisition means acquires, as the stability information in each calculation processing device, information indicating the stability of electricity supplied to an area where the calculation processing device is provided. system.
(Item 21) The calculation requesting apparatus further includes power processing index information acquisition means for acquiring power processing index information indicating power consumption per predetermined calculation processing amount for each of the plurality of calculation processing apparatuses, The condition information acquisition unit further acquires power amount condition information indicating a condition to be satisfied by an amount of power consumed when the plurality of calculation processing devices execute the plurality of partial calculations, and the selection unit further includes the power 19. The distributed computing system according to item 18, wherein a computing processing device that further satisfies the power amount condition information is selected based on processing index information.
(Item 22) A calculation requesting apparatus for causing a plurality of calculation processing apparatuses to execute a plurality of partial calculations constituting a predetermined calculation process, and for each of the plurality of calculation processing apparatuses, a predetermined calculation processing amount Power processing index information acquisition means for acquiring power processing index information indicating the power consumption per unit, and the partial calculation compared with the case where each of the plurality of partial calculations is processed with another partial calculation in the same calculation processing device Power consumption index information generating means for generating power consumption index information indicating the amount of power required for the processing, and a condition to be satisfied by the amount of power consumed when the plurality of calculation processing devices execute the plurality of partial calculations Condition information acquisition means for acquiring power amount condition information, and the partial calculation among the plurality of calculation processing devices in association with each of the plurality of partial calculations. A selection unit that selects a calculation processing device that satisfies the power amount condition information when performing calculation based on the power processing index information and the power consumption index information, and each of the plurality of partial calculations is the selection unit. A processing result for outputting a set of partial calculation results received from the plurality of calculation processing devices as a result of the calculation processing and a partial calculation requesting unit to be executed by the calculation processing device selected in association with the partial calculation A calculation requesting device comprising output means.

(Item 23) A calculation requesting apparatus that causes a plurality of calculation processing apparatuses to execute a plurality of partial calculations constituting a predetermined calculation process, and for each of the plurality of calculation processing apparatuses, Stability information acquisition means for acquiring stability information indicating the stability of supplied power, and conditions for the stability of electricity received when the plurality of calculation processing devices execute the plurality of partial calculations Condition information acquisition means for acquiring stability condition information and a calculation process that satisfies the stability condition information when executing the partial calculation among the plurality of calculation processing devices in association with each of the plurality of partial calculations A selection means for selecting a device based on the stability information, and a calculation process in which each of the plurality of partial calculations is selected in association with the partial calculation by the selection means A calculation requesting device comprising: partial calculation requesting means to be executed by the apparatus; and processing result output means for outputting a set of partial calculation results received from the plurality of calculation processing devices as a result of the calculation processing.
(Item 24) A control method for causing a computer to operate as a calculation requesting device that causes a plurality of calculation processing devices to execute a plurality of partial calculations constituting predetermined calculation processing, respectively, wherein the plurality of calculations are performed by the computer. For each of the processing devices, a power processing index information acquisition stage for acquiring power processing index information indicating power consumption per predetermined calculation processing amount, and for each of the plurality of partial calculations, in the same calculation processing device, other parts A power consumption index information generation stage for generating power consumption index information indicating the amount of power required for the partial calculation process compared to the case where the calculation is processed, and when the plurality of calculation processing devices execute the plurality of partial calculations A condition information acquisition stage for acquiring power amount condition information indicating a condition to be satisfied by a power amount to be consumed, and the partial calculations The calculation processing device that satisfies the power amount condition information when the partial calculation is executed among the plurality of calculation processing devices is selected based on the power processing index information and the power consumption index information. A selection step, a partial calculation request step for causing each of the plurality of partial calculations to be performed by the calculation processing device selected in association with the partial calculation in the selection step, and a portion received from the plurality of calculation processing devices A control method comprising: a process result output stage for outputting a set of calculation results as a result of the calculation process.

(Item 25) A control method for operating a computer as a calculation requesting device for causing a plurality of calculation processing devices to execute a plurality of partial calculations constituting a predetermined calculation processing, respectively, comprising: For each, a stability information acquisition stage for acquiring stability information indicating the stability of power supplied to the calculation processing device, and supply when the plurality of calculation processing devices execute the plurality of partial calculations In the condition information acquisition stage for acquiring the stability condition information indicating the condition of the electrical stability, in association with each of the plurality of partial calculations, when executing the partial calculation among the plurality of calculation processing devices A selection step of selecting a calculation processing device that satisfies the stability condition information based on the stability information, and each of the plurality of partial calculations in the selection step A processing result output for outputting a partial calculation request stage to be executed by the calculation processing device selected in association with the partial calculation and a result of the partial calculation received from the plurality of calculation processing devices as a result of the calculation processing A control method comprising:
(Item 26) A program for causing a computer to function as a calculation requesting device that causes a plurality of calculation processing devices to execute a plurality of partial calculations constituting a predetermined calculation processing, respectively, the computer including the plurality of calculation processing Power processing index information acquisition means for acquiring power processing index information indicating power consumption per predetermined calculation processing amount for each of the devices, and another partial calculation in the same calculation processing device for each of the plurality of partial calculations When the plurality of calculation processing devices execute the plurality of partial calculations, and the power consumption index information generating means for generating power consumption index information indicating the amount of power required for the partial calculation processing compared to Condition information acquisition means for acquiring power amount condition information indicating a condition to be satisfied by power consumption, and the plurality of partial calculations In association with each of the plurality of calculation processing devices, a calculation processing device that satisfies the power amount condition information when executing the partial calculation is selected based on the power processing index information and the power consumption index information. A selection means, a partial calculation request means for causing the calculation processing device selected in association with the partial calculation by the selection means to perform each of the partial calculations, and the partial calculation received from the plurality of calculation processing devices. A program that causes a set of the results to be output as a processing result output means for outputting the result set as a result of the calculation processing.

(Item 27) A program that causes a computer to function as a calculation requesting device that causes a plurality of calculation processing devices to execute a plurality of partial calculations that constitute a predetermined calculation processing, respectively, the computer including the plurality of calculation processing Stability information acquisition means for acquiring stability information indicating the stability of power supplied to the calculation processing device for each of the devices, and supplied when the plurality of calculation processing devices execute the plurality of partial calculations A condition information acquisition means for acquiring stability condition information indicating a condition of the stability of electricity received, and executing the partial calculation among the plurality of calculation processing devices in association with each of the plurality of partial calculations Selection means for selecting a calculation processing device satisfying the stability condition information based on the stability information, and each of the plurality of partial calculations, A set of partial calculation request means for causing the calculation processing device selected in association with the partial calculation by the selection means and a result of partial calculation received from the plurality of calculation processing devices to be output as a result of the calculation processing A program that functions as a processing result output means.
(Item 28) A recording medium on which the program according to item 26 or 27 is recorded.

FIG. 1 shows a block diagram of a distributed computing system 10. FIG. 2 shows a block diagram of the calculation requesting device 20. FIG. 3 is a block diagram of the calculation processing device 30-1. FIG. 4 shows an operation flow in which the distributed computing system 10 executes calculation processing. FIG. 5 shows an operation flow subsequent to FIG. 4 in the distributed computing system 10. FIG. 6 shows the details of the operation flow in S485. FIG. 7 shows details of the calculation process 700. FIG. 8 shows an example of the hardware configuration of the calculation requesting device 20.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Distributed computing system 20 Calculation request | requirement apparatus 30 Calculation processing apparatus 35 Processing apparatus group 40 Lower hierarchy management apparatus 45 Management apparatus group 50 Upper hierarchy management apparatus 200 Power processing index information acquisition means 205 Stability information acquisition means 210 Computer environment database 215 part Calculation request means 220 Request history information storage means 225 Power consumption index information generation means 230 Power consumption index information storage means 240 Calculation content analysis means 250 Condition information acquisition means 260 Area information acquisition means 270 Power charge information generation means 275 Selection means 280 Processing results Output means 300 Partial calculation processing means 310 Processing result transmission means 320 Actual consumption transmission means 330 Partial calculation reassignment means 700 Calculation processing

Claims (17)

A plurality of calculation processing devices that respectively execute a plurality of partial calculations constituting a predetermined calculation process;
A calculation requesting device for requesting partial calculation to the plurality of calculation processing devices;
A lower layer management device provided corresponding to each of a plurality of processing device groups into which the plurality of processing devices are classified;
An upper layer management device provided corresponding to each of a plurality of management device groups into which the plurality of lower layer management devices are classified;
With
The calculation requesting device is:
Power processing index information acquisition means for acquiring power processing index information indicating power consumption per predetermined calculation processing amount for each of the plurality of calculation processing devices;
For each of the plurality of partial calculations, power consumption index information generating means for generating power consumption index information indicating the amount of power required for the partial calculation compared with the case where another partial calculation is processed in the same calculation processing device When,
Condition information acquisition means for acquiring power amount condition information indicating a condition to be satisfied by the amount of power consumed when the plurality of calculation processing devices execute the plurality of partial calculations;
Corresponding to each of the plurality of partial calculations, a calculation processing device that satisfies the power amount condition information when executing the partial calculation among the plurality of calculation processing devices, the power processing index information and the power consumption index A selection means for selecting based on the information;
A partial calculation requesting unit that causes each of the plurality of partial calculations to be performed by the calculation processing device selected in association with the partial calculation by the selection unit;
Processing result output means for outputting a set of partial calculation results received from the plurality of calculation processing devices as a result of the calculation processing;
Each of the plurality of calculation processing devices includes:
Partial calculation processing means for processing the partial calculation requested from the calculation requesting device;
Processing result transmitting means for transmitting the processing result of the partial calculation to the calculation requesting device ;
When it is determined that the total power consumption of the corresponding processing device group has reached a predetermined reference ratio of the maximum suppliable power in the processing device group, the partial calculation is performed using the lower layer management device and the upper layer. A distributed computing system comprising: partial calculation reassignment means for transmitting to another calculation processing device provided in a region different from a region provided with the calculation processing device via a management device .   The lower tier management device, when the amount of power consumed in the corresponding processing device group exceeds a predetermined reference value, notifies that to each of the calculation processing devices to be managed,
  In response to the notification from the lower layer management device, the partial calculation reassignment means reaches the predetermined reference ratio of the maximum suppliable power in the processing device group when the total power consumption of the corresponding processing device group is reached. When it is determined that the partial calculation is transmitted to the other calculation processing device provided in a region different from the region in which the calculation processing device is provided via the lower layer management device and the upper layer management device. Do
  The distributed computing system of claim 1.   The lower layer management device manages power processing index information of each calculation processing device in the processing device group,
  The upper tier management device manages power processing general information that is a maximum value and a minimum value of power index information managed by each lower tier management device in the management device group,
  The power processing index information acquisition unit acquires the power processing comprehensive information from the upper tier management apparatus, and acquires the power processing index information from the lower tier management apparatus, respectively.
  The selection unit selects an upper tier management device in which any of the managed power processing comprehensive information satisfies the power amount condition information, and manages the power processing to be managed among the lower tier management devices managed by the upper tier management device A lower tier management device that satisfies any of the power amount condition information is selected, and a calculation processing device that satisfies the power amount condition is selected from among the calculation processing devices managed by the lower tier management device.
  The distributed computing system of claim 1. The power processing index information acquisition unit indicates power consumption per predetermined processing amount for at least one of the CPU, the hard disk drive, and the input / output device of the calculation processing device as the power processing index information of each calculation processing device. Get performance information
The power consumption index information generating means, for each of the plurality of partial calculations, to the hard disk drive, the amount of computation of the CPU required for the partial calculation compared with the case where another partial calculation is processed in the same calculation processing device The distributed computing system according to claim 1, wherein processing content information indicating at least one of an access amount and an input / output device usage amount is generated as the power consumption index information. The calculation requesting device is:
Area information acquisition means for acquiring area information indicating which of the plurality of calculation processing devices is provided in a plurality of predetermined areas;
For each of the plurality of calculation processing devices, it further comprises power rate information generating means for generating power rate information indicating a power rate per power consumption of the calculation processing device based on the regional information,
The distributed computing system according to claim 1, wherein the selection unit further selects a calculation processing device that satisfies the power amount condition information based on the power rate information. The power rate information generation means generates the electricity rate information in association with each of the plurality of calculation processing devices based on a time zone in which the calculation process is performed and a time zone of the region determined by the region information,
The selection means is a calculation processing device that executes each of the plurality of partial calculations, a calculation processing device in which the partial calculation requesting means in the area where the calculation processing device is provided requests the partial calculation at night, The distributed computing system according to claim 5, wherein the time is selected with priority over a computing device that is daytime. The calculation requesting device further includes region information acquisition means for acquiring region information indicating which of the plurality of calculation processing devices is provided in a plurality of predetermined regions,
In the case where the condition information acquisition means acquires a power upper limit value that defines an upper limit of power consumed per unit time in a plurality of calculation processing devices provided in a predetermined area as the power amount condition information.
The selection unit is configured to output the plurality of partial calculations when a plurality of calculation processing devices in the predetermined area process the plurality of partial calculations and power consumed per unit time is equal to or less than the power upper limit value. The distributed computing system according to claim 1, wherein a plurality of computing devices in the predetermined area are selected in association with each of the plurality of computing devices. When a plurality of calculation processing devices in the same region are selected as the calculation processing devices for processing the plurality of partial calculations, the partial calculation requesting unit starts partial calculation in one calculation processing device in the same region. 8. The distributed computing system according to claim 7, wherein the timing for starting execution of each of the plurality of partial calculations is adjusted so that the time for starting and the time for starting the partial calculation for another calculation processing device in the region are different. The calculation requesting device further includes, for each of the plurality of partial calculations, a calculation content analysis unit that analyzes the number of instructions included in the partial calculation.
The power consumption index information generating means, for each of the plurality of partial calculations, when the partial calculation has a larger number of instructions than the other partial calculations, the partial calculation is consumed in comparison with the other partial calculations. The distributed computing system according to claim 1, wherein power consumption index information indicating that the amount of power is large is generated. The calculation requesting device further includes, for each of the plurality of partial calculations, a calculation content analysis unit that analyzes the number of accesses to which the input / output device is accessed in the partial calculation processing.
The power consumption index information generating means, for each of the plurality of partial calculations, when the partial calculation is larger in number of accesses compared to other partial calculations, the partial calculation is compared with other partial calculations. The distributed computing system according to claim 1, wherein power consumption index information indicating that the power consumption is large is generated. The calculation requesting device further includes, for each of the plurality of partial calculations, a calculation content analysis unit that analyzes a size of transfer data transferred by the input / output device in the partial calculation processing.
The power consumption index information generating unit compares the partial calculation with each of the plurality of partial calculations when the size of the transfer data is larger than that of the partial calculation. The distributed computing system according to claim 1, wherein power consumption index information indicating that the power consumption is large is generated. The power consumption index information generating means consumes power when each of the plurality of partial calculations is executed in the same type of process as the partial calculation and having a shorter processing time than the partial calculation. The distributed computing system according to claim 1, wherein power consumption index information is generated based on the amount. The calculation requesting apparatus further includes, for each of the plurality of partial calculations, a calculation content analysis unit that analyzes whether the partial calculation is included in a critical path in the calculation process,
The selection means is a power consumption indicated by the power processing index information as compared with a calculation processing device selected in association with a partial calculation not included in the critical path among the calculation processing devices satisfying the power amount condition information. The distributed computing system according to claim 1, wherein a computing device having a larger amount and a faster processing speed is selected in association with a partial calculation included in the critical path. The calculation processing device transmits actual power consumption information indicating the amount of power consumed when the partial calculation processing means has processed the partial calculation to the calculation requesting device in association with information for identifying the partial calculation. And further having an actual consumption amount transmitting means,
The distributed computing system according to claim 1, wherein the power processing index information acquisition unit acquires the actual power consumption information per partial calculation from the actual consumption transmission unit as the power processing index information. The calculation requesting device further includes power consumption index information storage means for storing the type of partial calculation and the power consumption index information of the partial calculation in association with each other,
For each of the plurality of partial calculations, when the type of the partial calculation is stored in the power consumption index information storage unit, the power consumption index information generation unit is configured to use the consumption associated with the type of partial calculation. The distributed computing system according to claim 1, wherein the power consumption index information of the partial calculation is generated by selecting power index information from the power consumption index information storage unit. The calculation requesting device includes request history information storage means for storing information in which information for identifying a set of calculation processing devices that have performed the calculation processing is associated with a type of calculation processing that has already been performed by the calculation requesting device. In addition,
The partial calculation requesting unit searches the request history information storage unit for information for identifying a set of calculation processing devices associated with the type of calculation processing based on the type of calculation processing to be executed. Causing the set of calculation processing devices to perform the calculation processing,
The distributed computing system of claim 1. The distributed computing system according to claim 1, wherein the partial calculation request unit starts a partial calculation that requires a longer processing time earlier than a partial calculation that requires a shorter processing time.

Images