[Description] [Title of Invention] INFORMATION PROCESSING DEVICE [Technical Field] 5 [OOOl] The present invention relates to an information processing apparatus, and more particularly, to an information processing apparatus for performing distributed parallel processing on predetermined data. 10 [Background Art] [0002] A technique of a distributed processing system has been known, in which large-scale data are divided into multiple data units in a case where large-scale data are processed, and multiple work units respectively 15 perform processing on the divided multiple data units. [0003] For example, first, a distributed processing system divides, into data units, a processing request (job) which is given by the client and which is targeted 011 all or some of data unit groups divided large-scale 20 data. Subsequently, the distributed processing system generates a processing request (task) of a data unit for a work unit performing predetermined processing on each of the divided data units. Subsequently, in the divided processing system, each of n~ultiplew ork units performs predetermined processing on the data unit in accordance 25 with this processing request, and outputs a processing result. Then, the distributed processing system collects the processing results of the task, thus perforniing processing on the data units for the entire or a part of the large-scale data. [0004] 2 In the tecl~niqueo f such distributed processing system, it is desired to have a distributed processing system of large-scale data having a high degree of reliability capable of returning a processing result even in a case where a failure and the like occurs in a snlall number of work units (for 5 example, physical servers), and the work units goes down. In the technique of the distributed processing system, many work units are required, and therefore, it is desired to reduce the overhead due to the tasks by reducing the arnount of co~nmunicationa nd the number of times communication is perfornled between the work unit and the controller for 10 conlmanding the work unit to perfor111 the processing. [OOOS] In this case, in the distributed processing system, a technique for distributing the processing load to ~nultiplew ork units is known (for example, see PTL 1). In the technique of PTL 1, in a case where an 15 access request is given to nlultiple processors to access various kinds of data arranged in a distributed manner, a distributed-type data base management system in whicl~p rocessing according to the access request is performed by a particular processor arranged with desired data includes a processing load deviation detector and a data arrangenient change unit. 20 Then, this processing load deviation detector detects tlle processing load deviation on the basis of the system load statistics infor~nationa nd the access inforlnation for accessing the data unit, and changes the arrangement configuration of the data in accordance with the load deviation. As a result, the load of the task can be distributed. 25 [0006] In the technique of the distributed processing system, a redundant arrangenlent technique of data for arranging the same data unit in a distributed manner to each of the inultiple work units so as not to lose the data unit is known (for example, see PTL 2). The technique of PTL 2 3 includes means for classifying the physical node of the storage into groups, and means for allocating data so that distributed data and copied data of the distributed data do not exist in the group. The distributed processing system having such configuration arranges copied data to multiple 5 different groups, thus capable of maintaining redundancy of the data. [Citation List] [Patent Document] [0007] 10 PTL 1: Japanese Patent Laid-Open No. H9-218858 PTL 2: International Publication No. 2008/114441 [Summary of Invention] [Technical Proble~n] 1s [0008] However, in the above redundant arrangenlent technique, the data units are arranged in a distributed manner only in view of the redundancy, and therefore, in a case where each of ~nultipiew ork units performs processing on the data unit, it takes a long time to complete all the 20 processing. [0009] For example, in a case wl~eree ach of the two work units includes many identical data units, and a failure and the like occurs in one of the work units, one of the work units does not perform processing on the data 25 unit, and instead, the other of the work units performs processing on the data unit. Therefore, the number of data units processed by the other of the work units increases. Other work units that do not hold the same data unit as the data unit held by the one of the work units in which a failure and the like occors cannot perform processing on the data unit instead of 4 the one of the work units. As a result, the nuniber of work units performing processing at a time decreases in the entire distributed processing system, and therefore, there is a problenl in that it takes a long time to complete the processing on all of the data units, and the throughput 5 decreases. [OO lo] Therefore, it is an object of the present invention to provide an information processing apparatus capable of solving the problem of the reduction of the throughput which is the problem explained above. 10 [Solution to Problem] [OOll] An information processing apparatus according to an aspect of the present invention for solving the above object, includes, 15 data unit output control means for outputting, to a plurality of data unit processors, copied data units obtained by respectively copying a plurality of data units for a number that is set in advance, . the data unit output control Ineans being configured to respectively allocate output destinations of a plurality of identical copied data units to 20 different data tunit processors, for each combination of at least two data unit processors that is set in advance, obtain a number of the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors made into a combination, and 25 respectively allocate the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the number of copied data unit groups obtained for each co~nbinationo f the data unit processors becomes smaller. [OO 121 j, 5 An information processing method according to another aspect of the present invention, includes, while outputting, to a plurality of data miit processors, copied data units obtained by respectively copying a plurality of data units for a 5 number that is set in advance, allocati~lgo utput destinations of a plurality of identical copied data units to different data unit processors, for each combination of at least two data unit processors that is set in advance, obtaining a number of the copied data unit groups in which the 10 identical copied data units are respectively allocated to the data unit processors made into a combination, and respectively allocati~lgth e output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the number of copied data unit groups obtained for each c o ~ n b i ~ l a t ioof~ i 15 the data unit processors becomes smaller. [00 131 A program, according to another aspect of the present invention, for causing an i~lformationp rocessing apparatus to realize data unit output control mealis for outputting, to a plurality of data unit processors, copied 20 data units obtained by respectively copying a plurality of data units for a lumber that is set in advance, wherein the data unit output co~ltrolm eans respectively allocates output destinations of a plurality of identical copied data units to different data unit processors, and for each combinatio~lo f at least two data unit 25 processors that is set in advance, the data unit output control means obtains a number of the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors made into a combination, and the data unit output co~ltroml eans respectively allocates the output desti~latio~oisf the plurality of copied data units to the data unit 6 processors, so that the degree of variation of the number of copied data unit groups obtained for each combination of the data unit processors becomes smaller. [Advantageous Effects of Invention] 5 [0014] The present invention is configured as described above, and can irnprove the throughput of the processing performed on the data unit. [Brief Description of Drawings] [00 151 10 [Fig. I] Fig. 1 is a figure represe~iting a configuration of a distributed processing system accordiilg to a first exemplary embodiment of the present invention. [Fig. 21 Fig. 2 is a figure for explaining an exa~npleo f a work unit storage. 15 [Fig. 31 Fig. 3 is a figure for explaining an exa~npleo f a master data storage. [Fig. 41 Fig. 4 is a figure for explaining allocation of data units. [Fig. 51 Fig. 5 is a figure for explaining allocatio~ol f data units. [Fig. 61 Fig. 6 is a figure for explaining a mirror table value. 20 [Fig. 71 Fig. 7 is a figure for explaining an example of a data arrangelnellt storage. [Fig. 81 Fig. 8 is a figure illustrating an example of a data arrangement ~natrix. [Fig. 91 Fig. 9 is a figure illustrating an exaniple of a standard 25 deviation of a mirror table value. [Fig. 101 Fig. 10 is a figure illustrating an example of a data arrangelnent matrix. [Fig. 111 Fig. 11 is a figure illustrating an example of a standard deviation of a mirror table value. 7 [Fig. 121 Fig. 12 is a figure for explairiing an example of a total summation of a mirror table value. [Fig. 131 Fig. 13 is a figure for explaining an example of a job storage. [Fig. 141 Fig. 14 is a figure for explaining an example of a task storage. [Fig. 151 Fig. 15 is a figure for explaining an example of a copied data storage. [Fig. 161 Fig. 16 is a figure for explaining an exanlple of a task 10 storage. [Fig. 171 Fig. 17 is a figure for explaining an example of a task storage. [Fig. 181 Fig. 18 is a figure for explaining an example of a task storage. 1.5 [Fig. 191 Fig. 19 is a figure for explaining an example of a job storage. [Fig. 201 Fig. 20 is a figure for explaining timing up to a job completion. [Fig. 211 Fig. 21 is a figure for explaining tinling up to a job 20 completion. [Fig. 221 Fig. 22 is a figure for explaining an operation of an infornlatio~pl rocessing apparatus. [Fig. 231 Fig. 23 is a figure for explaining an operation of a data arrangement management module. 25 [Fig. 241 Fig. 24 is a figure for explaining an operation of a data arrangement planning module. [Fig. 251 Fig. 25 is a figure for explaining an operation of a task management module. [Fig. 261 Fig. 26 is a figure for explaining an operation of a work 8 unit module. [Fig. 271 Fig. 27 is a figure for explaitling an operati011 of a task scheduler. [Fig. 281 Fig. 28 is a figure for explaiuing an operation of a task 5 collector. [Fig. 291 Fig. 29 is a block diagram illustrati~lgf unctio~lso f an information processor according to a second exemplary embodiment of the present invention. 10 [Description of Embodiments] [00 161 Hereinafter, an exemplary embodiment of a distributed processing system according to the present invention will be explained with reference to Figs. 1 to 28. Figs. 1 to 21 are figures for explaining a configuration 15 of a distributed processing system. Figs. 22 to 28 are a figure for explai~linga n operation of the distributed processing system. [00 171 (Configuration) 20 As shown in Fig. 1, a distributed processing syste~tl1 according to a first exenlplary embodi~llento f the present i~lve~ltioi~nl cludesa n infornlation processing apparatus 11, a job client 12, and a systenl managetnent client 13. The job client 12 feeds a job (processing request) which is to be executed by the information processing apparatus 11. The 25 systenl management client 13 lnallages each of multiple work units (work unit modules 25 explained later) executing a task (divided processiug request) generated on the basis of a job. Then, the information processing apparatus 11 generates multiple tasks on the basis of the job fed fro111 the job client 12, and causes the ~liultiplew ork units to execute the 9 generated ~nultipleta sks. For example, the job client 12 outputs any given character string, as the content of the fed job. Then, the information processing apparatus 11 compares any given character string, which is output from the job client 12, and the character string stored in 5 advance, and executes processing for outputting the nuniber of matching characters. [0018] First, the configuration of the information processing apparatus 11 will be explained. The inforniation processing apparatus 11 includes a 10 work unit managenlent tnodule 21, a task managenlent module 22, a task collector 23, a data arrangement managenlent module 24, multiple work unit niodules 25A to 25N (hereinafter described as work unit niodules 25 when each of the multiple work unit modules 25A to 25N are not distinguished from each other). For exanlple, the function parts 15 constituting the infor~nationp rocessing apparatus 11 are arranged on different physical server. In this case, each function part constituting the infor~nationp rocessing apparatus 11 includes a central processing unit (CPU), a RAM (Random Access Memory), storage, and a network interface card (NIC), and can operate with a control of the operating system (0s). 20 Then, each server constituting the work unit management nodule 21, the task management nodule 22, the task collector 23, the data arrangelnent ~nanagementm odule 24, and multiple work unit nlodules 25 can be connected with each other via a network and the like (for example, TCP (Transmission Control Protocol)/IP (Internet Protocol)). 25 [0019] Each of data unit allocation processing, job registration processing, and job execution processing in the distributed processing systeni 1 will be hereinafter explained in order. Tlte data unit is divided predeter~nined data (for example, large-scale data) into multiple pieces. The data unit 10 allocation processing is processing for respectively allocating data units to ~nultiplew ork unit modules 25 (data unit processors). The job registration processing is processing for generating a task divided the job fed from the job client 12 in association with the data units. Then, in the 5 job execution processing, the work unit n~odule2 5 executes predetermined processing on the allocated data unit on the basis of the task. [0020] It slioold be noted that the data unit allocation processing is perforilled once at tlie first and the job registration processing and the job 10 execution processing are executed repeatedly. When the configuration of the distributed processing system 1 is changed, e.g., there is a change in the number of work unit nlodules 25 to which tlie data units are allocated, the data unit allocation processing is perforined again. [0021] 15 Explained below is a case where the number of work unit modules 25 (tlie number of work units) w is three, the number of data units d is four, and a redundancy degree number r of data unit is two. It is to be understood that the number of work units w, the number of data units d, and the redundancy degree number r are not limited thereto, and each of 20 them may be an integer equal to or more than two. The redundancy degree number r is a value that does not change for each data unit. The identical tnultiple copied data units that obtained by copying a data unit are respectively arranged in (allocated to) multiple different work unit modules 25. At this occasion, in the case where the number of work units 25 w is less than tlie redundancy degree number r holds, the degree of redundancy cannot be satisfied, and therefore, l Subsequently, a distributed processing system according to the second exemplary enlbodiment o f the present invention will be explained with reference to Fig. 29. A distributed processing system 201 according to the second exemplary embodiment includes a data unit output controller 21 1 (data unit output control means) for outputting, to ~nultipled ata unit processors 212, copied data units obtained by respectively copying multiple data units for a number that is set in advance, 10 wherein the data unit output controller 21 1 respectively allocates the output destinations of the identical multiple copied data units to different data unit processors 212, and for each conlbination o f at least two data unit processors 212 that is set in advance, the number of the identical copied data unit groups in 15 which the identical copied data units are allocated to the data unit processors 212 made into a conlbination is obtained, and the output destinations o f the multiple copied data units are respectively allocated to the data unit processors 212, so that the degree of variation of the number of copied data unit groups obtained for each combination o f the data unit 20 processors 212 becomes smaller. [0 1091 According to the above configuration, in a case where the data unit output controller 211 outputs the copied data units, which are obtained by copying each o f multiple data units for the number of times that is set in 25 advance, to ~nultipled ata unit processors 212, for exatnple, the multiple copied data units of which identification information of the data units is the same are respectively allocated to different data unit processors 212. For each combination o f at least two data unit processors 212 that is set in advance, the data unit output controller 21 1 obtains the number of the 47 identical copied data unit groups (mirror table values) in which the identical copied data units are allocated to the data unit processors made into the combination. Further, the data unit output controller 21 1 respectively allocates the output destinations of multiple copied data units 5 to the data unit processors 212 so that the degree of variation of the nlirror table values obtained for each co~ubinationo f the data unit processors 212 beconles smaller. Therefore, the data unit processors 212 holding copies of the identical data unit are allocated in a distributed manner without being concentrated on a single data unit processor 212. Therefore, even 10 in a case where a failure and the like occurs in the predetermined data unit processor, each data unit processor holding the same copied data unit as the copied data unit held by the data unit processor 212 in which a failure and the like occurs can perform predetermined processing on the copied data units on behalf of the data unit processor 212. As a result, the 15 processing on all the data units can be done quickly, and the throughput can be improved. [ O l lo] The invention of the present application has been hereinabove explained with reference to the above exemplary embodiments, but the 20 invention of the present application is not limited to the above exe~nplary e~nboditnents. Various changes that can be understood by a person skilled in the art within the scope of the invention of the present application can be made in the configuration and the details of the invention of the present application. 25 [ O l l l ] Some or all of the above exemplary embodiments may be described as shown in the following Supplemental note s, but are not limited thereto. [0112] 48 (Supplemental note 1) An information processing apparatus includes, data unit output control means for outputting, to a plurality of data unit processors, copied data units obtained by respectively copying a 5 plurality of data units for a nunlber that is set in advance, the data unit output control means being configured to respectively allocate output destinations of a plurality of identical copied data units to different data unit processors, for each combination of at least two data unit processors that is set 10 in advance, obtain a nunlber of the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors nlade into a combination, and respectively allocate the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation 15 of the nunlber of copied data unit groups obtained for each colnbination of the data unit processors beconles smaller. [0113] According to the above configuration, in a case where the data unit output control means outputs, to the plurality of data unit processors, the 20 copied data units obtained by respectively copying the plurality of data units for the number that is set in advance, for example, the data unit output controller respectively allocates a plurality of copied data units of which identification information of data units is the salne to different data unit processors. For each combination of at least two data unit 25 processors that is set in advance, a number of the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors made into a conibination (mirror table value) is obtained. The data unit output controller respectively allocates the output destinations of the plurality of copied data units to the data unit 49 processors, so that the degree of variation of the number of the mirror table values obtained for each co:nbinatioli of the data unit processors becomes sn~aller. Therefore, the data unit processors holding copies of the identical data unit are allocated in a distributed manner without being 5 concentrated on a single data unit processor. Therefore, even in a case where a failure and the like occurs in the predetermined data unit processor, each data unit processor holding the same copied data uuit as the copied data unit held by the data unit processor in which a failure and the like occurs can perforrn predetermined processing on the copied data units on 10 behalf of the data unit processor. As a result, the processing on all the data units can be done quickly, and the throughput can be improved. [0114] (Supplelnental note 2) The inforlnation processing apparatus according to Supplelnental 15 note 1, wherein, further, the data unit output control ~neansre spectively allocates the output destinations of the plurality of copied data ullits to the different data unit processors, so that the degree of variation of the nuntber of copied data units allocated for each data unit processor beconles smaller. 20 [0115] According to the above configuration, the data unit output control tneans respectively allocates the output destinations of the plurality of copied data units to the different data unit processors, so that the degree of variation of the nu~nbero f copied data units allocated for each data unit 25 processor beconles smaller. Therefore, the copied data units are not output in such a lnanner that the copied data units are concentrated on a single data unit processor, and the copied data units are respectively output in a distributed nlanller to the plurality of data unit processors. As a result, even in a case where a failure and the like occurs in the 50 predetermined data unit processor, another data unit processor can perform predetermined processi~lgo n behalf of the original data unit processor. As a result, the processing 011 all the data units can be done quickly, and the througllput can be improved. 5 [0116] (Supplemental note 3) The information processing apparatus according to Supplemental note 1 or 2, wherein the data unit output control means calculates the degree of variation, with respect to a predetern~inedr eference value, of the 10 number of copied data unit groups allocated with the identical copied data units obtained for each combination of the data unit processors, and respectively allocates the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the number of copied data unit groups thus calculated becomes smaller. 15 [0117] According to the above configuration, the data unit output control means calculates the degree of variatiou (for example, a standard deviation), with respect to a predetermined reference value, of the number of mirror table values obtained for each combination of the data unit 20 processors. The data unit output control means respectively allocates the atitput destinations of the plurality of copied data units to the data unit processors, so that the standard deviation thus calculated becomes smaller. As described above, by reducing the degree of variation of the number of identical copied data units, the number of identical copied data units for 25 each colnbiuation of the data unit processors can be equalized. As a result, the multiple copied data units can be more quickly and reliably distributed to n~ultipled ata unit processors. [OllS] (Supplemental note 4) 5 1 The information processing apparatus according to any one of Supplenlental notes 1 to 3, wherein the data unit output cor~trolm eans respectively allocates the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the 5 number of copied data unit groups allocated with the identical copied data units obtained for each co~nbinationo f the data unit processors becomes smaller, and a total su~n~natioonf the nuniber of copied data unit groups obtained for each combination of the data unit processors is minimized. [0119] 10 According to the above configuration, the data unit output control nleans respectively allocates the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the mirror table values obtained for each conlbination of the data unit processors beconles smaller, and a total summation of the nlirror table 15 values obtained for each conlbination of the data unit processors is minimized. As described above, in a case where there is an allocation pattern of data units in which the degree of variation of the nunlber of copied data unit groups is the same, the number of redundant data units (copied data units) for the combination of the data unit processors can be 20 reduced. Therefore, the processing requests for the data units are not output in snch a lnanner that the processing requests are concentrated 011 a single data unit processor, and the processing requests are respectively output in a distributed manner to the plurality of data unit processors. As a result, the processing on all the data units can be done quickly, and the 25 throughput can be improved. [0120] (Supplemental note 5) The information processing apparatus according to any one of Supplenlental note s 1 to 4, wherein the data unit output control means 52 generates a data unit output table in which the data unit processor serving as the output destination and the copied data nnit allocated to the data unit processor are associated with each other, and outputs the plurality of copied data units to the plurality o f data unit processors on the basis o f the 5 generated data unit output table. [O121] According to the above configuration, the data unit output control means generates a data unit output table (for example, a data arrangement matrix) in which the data unit processor and the data unit which is output 10 to the data unit processor are associated with each other in order to allocate the copied data unit and the data unit processor. Therefore, the matlagenlent and the control of the copied data units beco~nee asier, and the ~nultiplec opied data units call be more quickly and reliably output to data unit processors. 15 [O122] (Supplemental note 6 ) The information processing apparatus according to any one o f Supplemental notes 1 to 5, further includes, processing request control lllealls for obtaining a processing request for the plurality o f data units, and 20 generating a divided processing request divided the obtained processing request in association with each o f the data units, and in a case where the processing request co~ltroll nealls obtains identificatioll inforlllatio~flo r identifying the copied data unit allocated to the data unit processor from the data unit processor, the processing request 25 control lnealls outputs, to the data unit processor, the divided processing request corresponding to the copied data unit identified by the obtained identification information. [0 1231 According to the above configuration, the processing request 53 control means obtains a processing request (for example, a job) for the plurality of data units, and generates a divided processing request (a task for each data unit) divided the obtained processing request in association with each of the data units. Then, in a case where the processing request 5 control means obtains identification information for identifying the copied data unit allocated to the data unit processor from the data unit processor, the processing request control means obtains the task associated with the copied data unit allocated to the data unit processor, and outputs the task to the data unit processor. Therefore, for example, a processing request 10 (job) for large-scale data is efficiently divided, and can be output to the data unit processors. As a result, the processing on the large-scale data can be done quickly, and the throughput can be improved. [0 1241 (Snpplen~entaln ote 7) 15 The infor~nationp rocessing apparatus according to Suppleniental note 6, wherein the processing request control nleans outputs divided processing requests to the data unit processor, the divided processing requests having not yet output to another 20 data unit processor chosen from anlong the divided processing requests corresponding to the copied data unit identified by the obtained identification information, and being as many divided processing requests as a number equal to or less than an average value obtained by calculating a total summation of the nu~nberso f copied data units allocated to the data 25 unit processors and dividing the calculated total summation by the nunlber of data unit processors. [0125] According to the above configuration, the processing request control nieans outputs, to the data unit processor, divided processing 54 requests that has not yet output to another data unit processor, wherein the divided processing requests are as many as the number equal to or less than an average value obtained by calculating a total summation of the numbers o f copied data units allocated to the data unit processors and dividing the 5 calculated total sunlnlation by the number o f data unit processors. Therefore, the number of copied data units treated by the data unit processor at a time is reduced, so that the number o f divided processing requests that have not yet output to another data unit processor is increased. Therefore, all o f the divided processing requests for the data units held by 10 the predetermined data unit processor are less likely to be output to another data unit processor, and the nunlber o f data unit processors that cannot perform processing on the divided processing request is reduced. As a result, the processing for all the divided processing requests (data units) can be done quickly, and the throughput can be improved. 15 [0126] (Supplemental note 8) An information processing method includes, while outputting, to a plurality o f data unit processors, copied data units obtained by respectively copying a plurality of data units for a nu~nberth at 20 is set in advance, allocating output destinations o f a plurality o f identical copied data units to different data unit processors, for each conlbination o f at least two data unit processors that is set in advance, obtaining a number o f the copied data unit groups in which the 25 identical copied data units are respectively allocated to the data unit processors made into a combination, and respectively allocating the output destinations of the plurality of copied data units to the data unit processors, so that the degree o f variation of the number o f copied data unit groups obtained for each co~libinationo f 55 the data unit processors becomes smaller. [0 1 271 (Supplen~entaln ote 9) An infornlation processing nlethod according to claim 8, further 5 includes, respectively allocating the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the number of copied data units allocated for each data unit processor becomes smaller. [0128] 10 (Supplemental note 10) A program for causing an information processing apparatus to realize data unit output control means for outputting, to a plurality of data unit processors, copied data units obtained by respectively copying a plurality of data units for a number that is set in advance, 15 wherein the data unit output control means respectively allocates output destinations of a plurality of identical copied data units to different data unit processors, and for each combination of at least two data unit processors that is set in advance, the data unit output control nleans obtains a number of the copied data unit groups in which the identical copied data 20 units are respectively allocated to the data unit processors made into a combination, and the data unit output control means respectively allocates the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the nulnber of copied data unit groups obtained for each combination of the data unit processors 25 becomes s~naller. [0129] It should.be noted that the progranls described in each of the above exenlplary e~nbodinlentsa nd Suppleniental notes are stored in a storage device, or recorded in a computer-readable recording medium. For 56 example, the recording mediurn is a medium having portability such as a flexible disk, an optical disk, a magneto-optical disk, a sen~iconductor memory, and the like. [0130] 5 The invention of the present application has been hereinabove explained with reference to each of the above exemplary embodiments, but the invention of the present application is not limited to the exelnplary embodinlents described above. Various changes that can be understood by a person skilled in the art within the scope of the invention of the 10 present application can be made in the configuration and the details of the invention of the present application. [0131] It should be noted that the present invention enjoys the benefits of the priority claim based on the patent application of Japanese Patent 15 Application No. 2013-0273 12 filed on February 15, 20 13, in Japan, and all the contents described in the patent application is deemed to be included in this Description. [Reference Signs List] [0132] 20 1 distributed processing system 11 infornlation processing apparatus 12 job client 13 system management client 21 work unit nlanagenient module 25 22 task lnanagement module 23 task collector 24 data arrangement management nlodule 25 work unit module 3 1 work unit storage 41 job receiver 42 task generator 43 task scheduler 44 job storage 5 45 task storage 5 1 task result receiver 52 job result output module 61 data arrangement planning nodule 62 data arrangement execution tnodule 10 63 master data storage 64 data arratigelneut storage 71 task request module 72 task executior~~ llodule 73 copied data storage 15 201 distributed processing system 21 1 data unit output co~ltroller 212 data unit processor 58 [CLAIMS] [Claim 11 An illformation processing apparatus comprising: data unit output co~ltroln leatls for outputting, to a plurality of data 5 unit processors, copied data units obtained by respectively copying a plurality of data units for a number that is set in advance, the data unit output control nleans being configured to respectively allocate output destinations of a plurality of identical copied data units to different data unit processors, 10 for each combi~latio~ofl at least two data unit processors that is set in advance, obtain a number of the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors made into a combination, and respectively allocate the output destinations of the plurality of 15 copied data units to the data unit processors, so that the degree of variation of the nulnber of copied data uuit groups obtained for each colllbitlation of the data unit processors becomes smaller. [Claim 21 20 The infor~nationp rocessing apparatus accordi~lgto clailll 1, wherein, further, the data unit output coutrol means respectively allocates tile output destinations of the plurality of copied data units to the different data unit processors, so that the degree of variation of the uu~nbero f copied data 25 units allocated for each data unit processor becomes s~llaller. [Claim 31 The infornlation processi~lga pparatus according to clai~n1 or 2, wherein 59 the data unit output control means calculates the degree of variation, with respect to a predetermined reference value, of the number of copied data unit groups allocated with the identical copied data units obtained for each colnbination of the data unit processors, and respectively allocates the 5 output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the number of copied data unit groups thus calculated becollies smaller. [Clainl 41 10 The information processing apparatus according to any one of claims 1 to 3, wherein the data unit output control means respectively allocates the output destinations of the plurality of copied data units to the data unit processors. so that the degree of variation of the number of copied data unit groups 15 allocated with the identical copied data units obtained for each combination of the data unit processors beco~i~esms aller, and a total sulnniatioii of the nulnber of copied data unit groups obtained for each con~binationo f the data unit processors is minimized. 20 [Claim 51 The inforn~ationp rocessing apparatus according to any one of claims 1 to 4, wherein the data unit output control nleans genel.ates a data unit output table in which the data unit processor serving as the output destination and the 25 copied data unit allocated to the data unit processor are associated with each other, and outputs the plurality of copied data units to the plurality of data unit processors on the basis of the generated data unit output table. [Claim 61 60 The infor~nationp rocessing apparatus according to any one of claims 1 to 5, further comprising, processing request control means for obtaining a processing request for the plurality of data units, and generating a divided processing request 5 divided the obtained processing request in association with each of the data units, and in a case where the processing request control means obtains identification information for identifying the copied data unit allocated to the data unit processor from the data unit processor, the processing request 10 control means outputs, to the data unit processor, the divided processing request corresponding to the copied data unit identified by the obtained identification information. [Claim 71 15 The infor~nationp rocessing apparatus according to claim 6, wherein the processing request control means outputs divided processing requests to the data unit processor, the divided processing requests having not yet output to another data unit processor chosen from among the divided processing requests 20 corresponding to the copied data unit identified by the obtained identification information, and being as many divided processing requests as a nunlber equal to or less than an average value obtained by calculating a total summation of the nu~llberso f copied data units allocated to the data unit processors and dividing the calculated total sumnlation by the nulrtber 25 of data unit processors. [Claim 81 An information processing method comprising, while outputting, to a plurality of data unit processors, copied data units 6 1 obtained by respectively copying a plurality o f data units for a number that is set in advance, allocating output destinations o f a plurality o f identical copied data units to different data unit processors, 5 for each conlbi~~atioonf at least two data unit processors that is set in advance, obtaining a number o f the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors made into a combination, and respectively allocating the output destinations of the plurality o f 10 copied data units to the data unit processors, so that the degree o f variation o f the number o f copied data unit groups obtained for each combination o f the data unit processors becomes smaller. [Claim 91 15 An information processing method according to claim 8, further comprising, respectively allocating the output destinations o f the plurality o f copied data lullits to the data lullit processors, so that the degree o f variation o f the number o f copied data units allocated for each data unit processor 20 becomes smaller. [Claim 101 A program for causing an information processing apparatus to realize data unit output control means for outputting, to a plurality o f data 25 unit processors, copied data units obtained by respectively copying a plurality o f data units for a number that is set in advance, wherein the data unit output control means respectively allocates output destinations o f a plurality o f identical copied data units to different data unit processors, and for each con~binationo f at least two data unit 62 processors that is set in advance, the data unit output control means obtains a number of the copied data unit groups in which the identical copied data units are respectively allocated to the data unit processors made into a combination, and the data unit output colitrol means respectively allocates 5 the output destinations of the plurality of copied data units to the data unit processors, so that the degree of variation of the number of copied data unit groups obtained for each combination of the data unit processors becomes smaller.