JP5566936B2 - Job execution system and program - Google Patents

Description

  Embodiments described herein relate generally to a job execution system and a program.

  In general, a unit of work processed by a computer is called a job. For example, the unit of job is a process of collecting data, a process of generating a report by analyzing the collected data, and a process of printing the report. In this example, since the execution order of the three types of jobs has a dependency relationship, it is necessary to control the job execution order according to the dependency relationship.

  If there is no dependency among a plurality of jobs, the jobs may be executed in parallel in order to increase processing efficiency. For example, when collecting multiple types of data independently, a data collection job may be executed in parallel for each data resource.

  When there are a plurality of computing resources (CPU, data, etc.) and a plurality of jobs, it is necessary to appropriately schedule the computing resources to be used and the job execution order. Normally, job execution order is scheduled under the constraint of efficiently using limited computing resources.

  In general, a job net is used as a technique for instructing a computer to execute a plurality of jobs. In a job net, you can define “sequential execution”, “parallel execution”, “conditional branching”, etc., for jobs without dependencies. In addition, control is performed such that an individual job is mounted so that it can be rolled back, and if a job fails to be executed, the job is retried.

  21 and 22 are schematic diagrams showing examples of graphical representation and data structure of job net definition information, and FIG. 23 is a schematic diagram showing an example of job execution when the second condition is hit in a conditional branch. .

  First, the head job job0 is executed in accordance with the job net definition information 1 shown in FIG. When the execution of the first job job0 is completed, the first to third jobs job1, job2, job3 are executed in parallel. When execution of the first to third jobs job1, job2, and job3 is completed, the process proceeds to the next conditional branch.

  In conditional branching, a job that hits a condition is selectively executed. In the job net definition information 1 shown in FIG. 22, when the return value of the second job job2 is 1, the fourth job job4 is executed, and when the return value of the second job job2 is 2, the fifth job job5 is executed. In other cases, the sixth job job6 is executed.

  In this example, it is assumed that the return value of the second job job2 is 2, and the fifth job job5 is executed. When the execution of the fifth job job5 is completed, the final job job7 is executed. When the execution of the final job job7 is completed, the execution of the entire job net is completed.

  The first to third jobs job1, job2, and job3 are not limited to being executed on the same computer resource, but may be executed in parallel on a plurality of computer resources.

  In the above example, a job net expressing three types of execution order such as “sequential execution”, “parallel execution”, and “conditional branch” is used, but the job net is not necessarily limited thereto. For example, there may be only two types of “sequential execution” and “parallel execution”, or there may be an iterative loop in addition to “sequential execution” / “parallel execution” / “conditional branch”.

  As such parallel / distributed processing, for example, a load distribution technology using a load balancer, a distributed processing technology using MapReduce, or the like may be used in addition to a technology that explicitly instructs parallel execution using a job net or the like.

  On the other hand, parallel processing of instruction sequences is also executed inside the CPU. For example, by speculative execution, before the preceding instruction is completed, the subsequent instruction is provisionally executed while predicting the result of the preceding instruction, and if the prediction is different, the subsequent instruction is redone.

JP-A-9-218861

  As described above, normally, the job execution order is scheduled under the constraint of efficiently using limited computing resources.

  Here, according to the study of the present inventor, with the widespread use of virtualization technology and cloud computing technology, it is possible to create an environment where surplus resources can be flexibly used, and jobs that sufficiently use surplus resources and reduce processing time Execution scheduling may be requested.

  In such job execution scheduling, for example, a job is pre-read even when executing a job net, such as speculative execution at the CPU / machine language level, and the entire job net (a collection of jobs with restrictions on the execution order). It is desirable to reduce the time required for this process.

  However, according to the study of the present inventor, there are the following inconveniences when prefetching and executing a job under a job net. In other words, the above-described job net definition information cannot fully express the presence or absence of dependency between jobs. For this reason, in the job scheduling based on the job net definition information, it is not possible to determine whether or not the job can be read ahead.

  An object of the embodiment disclosed in the present specification is to provide a job execution system and a program that can determine whether or not a job can be prefetched and reduce the time required to execute the entire job net.

  The job execution system according to the embodiment includes a job activation control device and a job execution device that can communicate with each other.

  The definition information storage means of the job activation control device describes job IDs for identifying the jobs arranged in the order of execution of the jobs, and branches several jobs of the jobs. In the case of executing in parallel, the job net definition information further describing the branch condition is stored in association with the job IDs of the respective jobs.

  The dependent content storage unit of the job activation control device includes the job ID, the dependent job ID for identifying the dependent job on which the job with the job ID depends, and the dependent content of the dependent job. Dependent content indicating that it is an execution process or a return value is stored in association with each other.

  The job correction information storage unit of the job activation control device indicates the job ID, rollback information indicating whether or not the job of the job ID can be rolled back, whether or not argument correction of the job is possible, and a default value of the argument. The argument correction information is stored in association with each other.

  The execution status storage unit of the job activation control device includes the job ID, a job instance ID for identifying a job instance executed as a job of the job ID, and the status of the job instance is executing, completed or failed. An execution status indicating that the execution of the job instance is a main execution according to the execution order or a pre-read execution executed before the execution order of the main execution, and the execution state is completed. The return value of the job instance in the case of indicating is stored in association with each other.

  The first activation instruction transmission unit of the job activation control device stores the definition information that matches the job ID associated with both the execution type indicating the main execution and the execution state indicating the completion in the execution state storage unit. Among the job IDs in the means, the job ID in the last execution order is extracted in the definition information storage means.

  The first activation instruction transmission unit of the job activation control device selects a job ID arranged in the execution order next to the execution order of the extracted job ID.

  The first activation instruction transmission unit of the job activation control device includes an execution type indicating pre-read execution and an execution state indicating execution in progress for the job ID in the execution state storage unit that matches the selected job ID. When the two are not associated, a job instance ID for identifying a job instance to be executed as a job of the selected job ID is designated.

  The first activation instruction transmitting means of the job activation control apparatus transmits a first activation instruction including the selected job ID and the designated job instance ID to the job execution apparatus.

  The job activation control device associates a job instance ID in the first activation instruction, an execution state indicating execution, and an execution type indicating actual execution with the job ID in the first activation instruction. Write to the execution state storage means.

  The pre-reading candidate selection unit of the job activation control device is a job ID different from a job ID associated with an execution state indicating execution or completion in the execution state storage unit, and the dependency content storage unit includes the job ID. A job ID that is not associated with the dependency content indicating that it is an execution process is selected as a pre-read execution candidate job ID.

  The second activation instruction transmission unit of the job activation control device can perform rollback information or argument correction indicating that rollback is possible in the job correction information storage unit among the job IDs of the selected prefetch execution candidates. The job ID associated with the argument correction information indicating “” is selected.

  The second activation instruction transmission unit of the job activation control device includes the job ID arranged in the earliest execution order among the job IDs in the definition information storage unit that match the selected job ID, and the job A second start instruction including a job instance ID for identifying a job instance to be executed as a job with an ID, and a default value of an argument associated with the job ID in the job correction information storage unit that matches the job ID Is transmitted to the job execution apparatus.

  The job activation control device associates the job instance ID in the second activation instruction, the execution state indicating execution, and the execution type indicating prefetch execution with the job ID in the second activation instruction. Write to the execution state storage means.

  When the job activation control device receives an execution result including a job ID, a job instance ID, and a return value from the job execution device, the job activation control device associates the return value with the job ID and job instance ID in the execution result. The execution state associated with the job ID and job instance ID is updated to completion while writing to the storage means.

  The job activation control device includes a job ID in the definition information storage unit that matches a job ID associated with both an execution type indicating the main execution and an execution state indicating completion in the execution state storage unit. The job ID of the last execution order is extracted in the definition information storage unit.

  The job activation control apparatus selects a job ID arranged in the execution order next to the execution order of the extracted job ID, and matches the job ID in the next execution order with the job ID in the execution state storage unit Is associated with the job ID in the job correction information storage means that matches the job ID of the next execution order when the execution type indicating pre-reading execution and the execution state indicating completion are associated with The value is compared with the return value associated with the execution type indicating the pre-read execution, and it is determined whether or not they match.

  In the job activation control device, when the result of this determination is that they do not match, the rollback information associated with the default value of the argument used for the determination in the job correction information storage means indicates whether rollback is possible. When the argument correction information associated with the rollback information in the job correction information storage means indicates argument correction or not, the job ID associated with the rollback information and the job ID that matches the job ID A rollback request including a job instance ID associated with the job ID in the execution state storage unit and an argument indicating a return value used for the determination is sent out.

  The rollback instruction transmission means of the job activation control apparatus transmits the sent rollback request to the job execution apparatus as a rollback instruction.

  When the job activation unit of the job execution device receives the first or second activation instruction from the job activation control device, the job activation unit activates a job based on the job ID and job instance ID in the activation instruction.

  The job execution device returns an execution result of the activated job to the job activation control device.

  When the rollback execution means of the job execution apparatus receives the rollback instruction from the job activation control apparatus, the rollback execution means executes a job rollback based on the rollback instruction.

  The job execution device returns the execution result of the rollback again to the job activation control device.

It is a schematic diagram which shows an example of the job dependence relationship information for outline description common to each embodiment. It is a schematic diagram which shows an example of the job prefetch execution information for outline description common to each embodiment. It is a time chart which shows the prefetch execution example for the outline | summary description common to each embodiment. It is a time chart which shows the prefetch execution example for outline description common to each embodiment compared with the conventional job execution example. It is a mimetic diagram showing an example of composition of a job execution system concerning a 1st embodiment. It is a schematic diagram which shows an example of the graphical representation of the job dependence information in the same embodiment. It is a schematic diagram which shows an example of the data structure of the job dependence information in the embodiment. It is a schematic diagram which shows an example of the data structure of the job prefetch execution information in the same embodiment. It is a schematic diagram which shows an example of the data structure of the job execution state table in the same embodiment. It is a schematic diagram for demonstrating the operation | movement in the embodiment. FIG. 6 is a schematic diagram for explaining a job activation operation by a job net analysis unit and a job activation unit in the embodiment. It is a schematic diagram for demonstrating operation | movement of the job prefetch execution in the same embodiment. It is a flowchart for demonstrating the detailed operation | movement of step ST11 in the embodiment. It is a schematic diagram for demonstrating the detailed operation | movement of step ST11 in the embodiment. It is a schematic diagram for demonstrating the detailed operation | movement of step ST11 in the embodiment. 6 is a flowchart for explaining an operation of correcting a pre-executed job in the embodiment. It is a schematic diagram which shows an example of the graphical representation of the job net definition information used for the job execution system which concerns on 2nd Embodiment. It is a schematic diagram which shows an example of the data structure of the job dependence information in the embodiment. It is a schematic diagram which shows an example of the data structure of the job prefetch execution information in the same embodiment. 6 is a time chart showing a pre-read execution example in the embodiment in comparison with a conventional job execution example. It is a schematic diagram which shows an example of the graphical representation of general job net definition information. It is a schematic diagram which shows an example of the data structure of general job net definition information. It is a time chart which shows the example of conventional job execution.

  Each embodiment will be described below with reference to the drawings, but before that, an outline common to each embodiment will be described.

  As a common outline, job dependency information related to job prefetch execution and job prefetch execution information are used in addition to the job net definition information 1 shown in FIG. 22 as information for defining the job execution order. FIG. 1 and FIG. 2 show examples of job dependency relationship information 2 and job prefetch execution information 3.

  In the job dependency relationship information 2, it is defined whether or not execution of a certain job depends on the execution result of another job. In the example illustrated in FIG. 2, the execution of the second job job2 depends on the execution result of the top job job0. This is the case when the first job job0 and the second job job2 are accessing the same data resource, and there is a possibility that the first job job0 may update the data. The execution of the third job job3 also depends on the execution result of the top job job0, but the third job job3 depends only on the return value of the top job job0. In the above example, it depends on the execution result of job0 and the return value of job3, but it is not necessarily limited to these. For example, an example such as “a third job job3 cannot be executed unless the first job job0 is executed once or more” is also conceivable. Further, the job dependency relationship information 2 may be called an inter-job dependency graph.

  The job prefetch execution information 3 describes information for determining whether or not each job can be prefetched. For example, the fourth job job4 can be rolled back, and the third job job3 can be corrected. Specifically, the correctable third job job3 is executed with the default value of the argument, and may be corrected if the argument value changes.

  FIG. 3 shows a time chart of a prefetch execution example using the job dependency relationship information 2 and the job prefetch execution information 3. In this example, as in FIG. 23, the second condition is hit in the conditional branch.

  According to the job net definition information 1, the first job job0 is executed first. In parallel with the execution of the first job job0, the first job job1 and the third job job3 are also executed. This is because the first job job1 and the third job job3 use the return value from the first job job0, but can be corrected after receiving the return value by executing with the default value of the argument instead of the return value. On the other hand, since the second job job2 depends on the execution result of the top job job0 in addition to the return value, it is not started until the top job job0 is completed.

  Further, the fifth job job5 included in the conditional branch option is also executed in parallel. Conditional branching options job4 to job6, job4, job5, and job6 have no conditional branching conditions immediately after the job net is started, and it is uncertain which job should be executed. However, since the fifth job job5 does not depend on the previous job, it is speculatively executed immediately after the start of the job net. Since the fourth and sixth jobs job4 and job6 are individually dependent on the first and second jobs job1 and job2, they are not executed at the first time point. When the first job job0 is completed, the second job job2 is started, and when the second job job2 is completed, the sixth job job6 is started.

  When the first job job0 is completed, rollback or correction of the job that has been pre-read is executed according to the execution result of the first job job0. For example, when the execution result of the first job “job0” is “1”, the first job job1 prefetched with the default value “0” of the argument is rolled back and reexecuted with the argument value “1”. Since the third job job3 is prefetched with the default value “1” of the argument, it is not corrected in this example.

  When the execution of the first to third jobs job1, job2, and job3 is completed, the condition branching condition is determined. In this example, the second condition is hit, and the fifth job job5 is executed. However, the fifth job job5 is completed by prefetching execution. Therefore, the final job job7 is immediately executed.

  Here, the time chart of the conventional job execution example shown in FIG. 23 and the time chart of the prefetch execution example shown in FIG. 3 are shown side by side in FIG. 4 (a) and FIG. 4 (b).

In the prefetch execution example, the execution time of the entire job net is the total time of the execution time of the first job job0, the rollback and re-execution time of the first job job1, and the execution time of the final job job7 (t _job0 + t _{job1-rollback} + t _job1 + t _job7 ).

On the other hand, in the case of the conventional execution example, the execution time of the entire job net is the total time of the execution time of the first job job0 and the execution times of the third, fifth and seventh jobs job3, job5, job7 (t _job0 + t _job3 + _tjob5 + _tjob7 ).

Therefore, in the _pre- read execution example, the execution time of the entire job net can be shortened by the difference between the two (t _job3 + t _job5- (t _{job1 -rollback} + t _job1 )).

<First Embodiment>
FIG. 5 is a schematic diagram illustrating a configuration example of the job execution system according to the first embodiment, and FIGS. 6 to 9 are schematic diagrams illustrating an example of information used in the system. The job execution system includes a job activation control device 10 and a job execution device 20 that can communicate with each other. Each of the devices 10 and 20 can be implemented by either a hardware configuration or a combined configuration of hardware resources and software for each of the devices 10 and 20. As the software of the combined configuration, a program that is installed in advance on the computers of the corresponding apparatuses 10 and 20 from a network or a storage medium and that realizes the functions of the corresponding apparatuses 10 and 20 is used.

  Here, the job activation control device 10 includes a storage unit 11 and a job activation control unit 12.

  The storage unit 11 is a storage device that can be read / written from each of the units 12 to 14. For example, the storage unit 11 stores job net definition information 1, job dependency relationship information 2, job prefetch execution information 3, and job execution state table 4. Yes. Each information 1 to 3 is written in advance by a CPU (not shown) in accordance with an operation of a user terminal (not shown) such as a job net editor by the user.

  In the job net definition information 1, for example, as shown in FIG. 22, job IDs for identifying each job are described in the order of execution of each job. Are branched and executed in parallel, branch conditions are further described in association with the job IDs of the respective jobs.

  The job dependency relationship information 2 includes a job ID, a job ID of a dependency source that identifies a job of a dependency source on which the job of the job ID depends, and the content of the dependency is an execution process or a return value of the job of the dependency source. The dependency contents indicating that there is a description are described in association with each other.

  An example of the graphical representation and data structure of the job dependency relationship information 2 is shown in FIGS. In the graph notation, the dependency relationship is expressed by an arrow between jobs, and the dependency relationship is stored in a table format inside the system.

  The job dependency relationship information 2 defines, for example, the following two dependency relationships (2-1) to (2-2).

  (2-1) Execution of a certain job depends on the execution process of another job.

  (2-2) Execution of a certain job depends only on the return value of another job.

  However, the job dependency relationship information 2 is not limited to these two types of dependency relationships (2-1) to (2-2).

  In the example of FIG. 7, the first job job1 depends only on the return value of the first job job0. For example, the first job job1 does not refer to the data resource updated by the first job job0, but the first job job1 uses a return value obtained as a result of the execution of the first job job0. In the figure notation, it is defined as a broken line arrow, and “return value” in the data structure table. The second job job2 depends on the execution process of the first job job0. For example, the case where the data resource with the first job job0 is updated and the second job job2 refers to the data resource is applicable. In the figure notation, it is defined as a solid line arrow and “execution process” in the data structure table.

  As shown in FIG. 8, the job prefetch execution information 3 includes a job ID, rollback information indicating whether or not the job with the job ID can be rolled back, and load, and whether or not the argument correction for the job is possible and the load and argument. The argument correction information indicating the default value is described in a table format in association with each other. Note that the rollback load and the argument correction load may be omitted. The job prefetch execution information 3 is not limited to these rollback information and argument correction information.

  In the example of FIG. 8, if the third job job3 becomes unnecessary after the start of execution, the execution result can be rolled back. The load applied to it is “40”. In rollback, the unit of load may be determined for each application, and the value may not be a constant but a function. The third job job3 is executed with the default value as an argument, and the execution result can be corrected if the actual argument value changes later. The load applied to it is “10”. The unit of the argument correction load may be determined for each application, and the value may be a function instead of a constant.

  As shown in an example in FIG. 9, the job execution state table 4 includes a job ID, a job instance ID for identifying a job instance executed as a job of the job ID, and the job instance state being executed, completed or failed. The execution status indicating that the job instance is executed, the execution type indicating whether the execution of the job instance is the main execution according to the execution order or the pre-read execution executed before the execution order of the main execution, and the execution state indicates completion The return values of the job instances are described in a table format in association with each other.

  However, the job execution status table 4 includes (4-1) ID of actually executed job instance, (4-2) Job execution status (execution, completion, failure, etc.), (4-3) Job net It is not limited to four types of information such as the distinction between the main execution based on the definition information 1 and the job prefetching execution, and (4-4) the return value of the job execution result.

  In the example of FIG. 9, the first job job0 has completed this execution, and its return value is “3”. The first to third jobs job1, job2, job3 are being executed as the main execution. The fifth job job5 is being executed as prefetching execution. The fourth, sixth, and final jobs job4, job6, and job7 are not started.

  The job activation control unit 12 plays a central role in job execution scheduling, and includes a job net analysis unit 13 and a job prefetch execution analysis unit 14.

  The job net analysis unit 13 has the following functions (f13-1) to (f13-6).

  (f13-1) Of the job IDs in the job net definition information 1 that matches the job ID associated with both the execution type indicating the main execution and the execution state indicating completion in the job execution state table 4, the job The job ID of the last execution order in the net definition information 1 is extracted, the job ID arranged in the execution order next to the execution order of the extracted job ID is selected, and the job execution that matches the selected job ID A job that identifies a job instance to be executed as a job of the selected job ID when both the execution type indicating pre-read execution and the execution state indicating execution are not associated with the job ID in the state table 4 An instance ID is designated, and a first start instruction including the selected job ID and the designated job instance ID is sent to the job execution device. The first startup instruction transmission function for transmitting to 20.

  (f13-2) A job execution state in which the job instance ID in the first start instruction, the execution state indicating execution, and the execution type indicating main execution are associated with the job ID in the first start instruction Function to write to table 4.

  (f13-3) When an execution result including a job ID, job instance ID, and return value is received from the job execution device 20, the return value is associated with the job ID and job instance ID in the execution result and the job execution state table 4 And updating the execution state associated with the job ID and job instance ID to completion.

  (f13-4) Of the job IDs in the job net definition information 1 that matches the job ID associated with both the execution type indicating the main execution and the execution state indicating completion in the job execution state table 4, the job The job ID of the last execution order is extracted from the net definition information 1, and the job ID arranged in the execution order next to the execution order of the extracted job ID is selected, and matches the job ID of the next execution order When an execution type indicating pre-reading execution and an execution state indicating completion are associated with the job ID in the job execution state table 4 to be executed, the job pre-reading execution information 3 in the job pre-reading execution information 3 that matches the job ID in the next execution order is associated. A function that compares the default value of the argument associated with the job ID with the return value associated with the execution type indicating the prefetch execution and determines whether or not they match. .

  (f13-5) If the result of this determination is that they do not match, the rollback information associated in the job prefetch execution information 3 with the default value of the argument used for the determination indicates that rollback is possible. When the argument correction information associated with the back information in the job prefetch execution information 3 indicates whether or not argument correction is performed, the job ID associated with the rollback information and the job execution state table 4 matching this job ID A function of sending a rollback request including a job instance ID associated with a job ID in the job and an argument indicating a return value used for the determination to the job prefetch execution analysis unit 14.

  (f13-6) As a result of determination, when the two do not match, and the argument correction information including the default value of the argument used for the determination indicates that argument correction is possible, the job ID associated with the argument correction information A correction request including a job instance ID associated with the job ID in the job execution state table 4 that matches the job ID and an argument indicating the return value used for the determination to the job prefetch execution analysis unit 14. Function to send out.

  The job prefetch execution analysis unit 14 has the following functions (f14-1) to (f14-6).

  (f14-1) Dependence indicating that the job ID is different from the job ID associated with the execution state indicating execution or completion in the job execution state table 4 and that the job dependency information 2 indicates an execution process A prefetch candidate selection function that selects a job ID that is not associated with the content as a job ID of a prefetch execution candidate.

  (f14-2) Among the job IDs of the selected prefetch execution candidates, the job ID associated with the rollback information indicating that rollback is possible or the argument correction information indicating that argument correction is possible in the job prefetch execution information 3 is The job ID selected in the job net definition information 1 that matches the selected job ID is identified in the earliest execution order, and the job instance to be executed as the job with this job ID is identified. A second start instruction for transmitting a second start instruction including a job instance ID and a default value of an argument associated with the job ID in the job prefetch execution information 3 that matches the job ID to the job execution apparatus 20 Send function.

  (f14-3) The job execution state in which the job instance ID in the second start instruction, the execution state indicating execution, and the execution type indicating pre-read execution are associated with the job ID in the second start instruction Function to write to table 4.

  (f14-4) When an execution result including a job ID, job instance ID, and return value is received from the job execution device j20, the return value is associated with the job ID and job instance ID in the execution result and the job execution status table 4 And updating the execution state associated with the job ID and job instance ID to completion.

  (f14-5) A rollback instruction transmission function for transmitting the sent rollback request to the job execution apparatus 20 as a rollback instruction.

  (f14-6) A correction instruction transmission function for transmitting the sent correction request to the job execution apparatus 20 as a correction instruction.

  On the other hand, the job execution device 20 includes a storage unit 21 and a job execution unit 22.

  The storage unit 21 stores job programs corresponding to each job in advance. Each job has one function of a CPU (not shown) that executes a job program corresponding to each job, and is executed when the CPU that has received an activation instruction executes the job program.

  The job execution unit 22 includes a job activation unit 23 and a job execution result correction unit 24.

  The job activation unit 23 has the following functions (f23-1) to (f23-2).

  (f23-1) A job activation function that activates a job based on the job ID and job instance ID in the activation instruction when the first or second activation instruction is received from the job activation control device 10.

  (f23-2) A function of returning the execution result of the activated job to the job activation control device 10.

  The job execution result correction unit 24 has the following functions (f24-1) to (f24-4).

  (f24-1) A rollback execution function that, when receiving a rollback instruction from the job activation control device 10, executes a rollback of a job based on the rollback instruction.

  (f24-2) A function of returning the execution result of the rollback again to the job activation control device 10.

  (f24-3) A job execution result correction function that, upon receiving a correction instruction transmitted from the job activation control device 10, corrects the job execution result based on the job ID, job instance ID, and argument in the correction instruction.

  (f24-4) A function of returning the corrected execution result to the job activation control device 10.

  Next, the operation of the job execution system configured as described above will be described with reference to FIGS. In the drawings to be referred to, the storage units 11 and 21 are not shown, for example, as shown in FIG.

  As in the past, in order to execute a job along the job net definition information 1, only the job net analysis unit 13 of the job activation control unit 12 and the job activation unit 23 of the job execution unit 22 need be provided. In order to perform prefetching, the job prefetching execution analysis unit 14 of the job activation control unit 12 and the job execution result correction unit 24 of the job execution unit 22 are required.

  FIG. 11 is a schematic diagram for explaining the job starting operation by the job net analyzing unit 13 and the job starting unit 23.

  The job net analysis unit 13 refers to the job net definition information 1 and selects a job to be started next (ST1). Specifically, the job net analysis unit 13 matches the job ID in the job net definition information 1 that matches the job ID associated with both the execution type indicating the main execution and the execution state indicating the completion in the job execution state table 4. Among the IDs, the job ID of the last execution order is extracted in the job net definition information 1, and the job ID arranged in the execution order next to the execution order of the extracted job ID is selected.

  The job net analysis unit 13 refers to the job execution state table 4 and confirms the execution state of the job (ST2). Specifically, the job net analysis unit 13 confirms that an execution state indicating that the job is being executed is not associated with the job ID in the job execution state table 4 that matches the job ID selected in step ST1.

  If the job has not been activated yet, the job net analysis unit 13 transmits a job activation instruction to the job activation unit 23 (ST3) and records the fact that the activation instruction has been transmitted in the job execution state table 4. . Specifically, the job net analysis unit 13 specifies a job instance ID for identifying a job instance to be executed as a job having the job ID selected in step ST1 after the confirmation in step ST2, and selects the selected job ID and the job ID. A first activation instruction including the designated job instance ID is transmitted to the job execution apparatus 20. In addition, the job net analysis unit 13 associates the job instance ID in the first start instruction, the execution state indicating execution, and the execution type indicating main execution with the job ID in the first start instruction. To the job execution state table 4.

  On the other hand, when the job execution unit 22 receives the first activation instruction from the job activation control device 10, the job execution unit 22 activates the job based on the job ID and job instance ID in the first activation instruction (ST4).

  When the job execution is completed, the job returns an execution result including the job ID, job instance ID, and return value to the job execution unit 22 (ST5).

  The job execution unit 22 transmits the execution result to the job net analysis unit 13 (ST6).

  The job net analysis unit 13 records the execution result in the job execution state table 4 (ST7). Specifically, when the job net analysis unit 13 receives an execution result including a job ID, job instance ID, and return value from the job execution apparatus 20, the job net analysis unit 13 associates the return result with the job ID and job instance ID in the execution result. The value is written in the job execution state table 4 and the execution state associated with the job ID and job instance ID is updated to complete.

  FIG. 12 is a schematic diagram for explaining the operation of job prefetch execution with the job prefetch execution analysis unit 14 and the job execution result correction unit 24 added.

  The job prefetch execution analysis unit 14 refers to the job execution state table 4, the job dependency relationship information 2, and the job prefetch execution information 3, and extracts a job that can be prefetched (ST11). For example, the job prefetch execution analysis unit 14 has a job ID that is different from the job ID associated with the execution state indicating execution or completion in the job execution state table 4, and in the execution process within the job dependency relationship information 2. A job ID that is not associated with the dependency content indicating that it is present is selected as the job ID of the prefetch execution candidate. Next, the job prefetch execution analysis unit 14 uses the job ID of the selected prefetch execution candidate as rollback information indicating that rollback is possible in the job prefetch execution information 4 or argument correction information indicating that argument correction is possible. The associated job ID is selected, and the job ID arranged in the earliest execution order is extracted from the job IDs in the job net definition information 1 that match the selected job ID.

  Details of step ST11 according to a specific example will be described later.

  The job prefetch execution analysis unit 14 refers to the job execution state table 4 and confirms the execution state of a job that can be prefetched (ST12).

  If the job has not been started yet, the job prefetch execution analysis unit 14 transmits a start instruction for the job to be prefetched to the job starter 23 (ST13), and the job execution state table 4 indicates that the start instruction has been transmitted. To record. At the time of recording, it is clearly indicated that the pre-read execution instruction is not a normal start instruction. Specifically, the job prefetch execution analysis unit 14 confirms step ST12 and then extracts the job ID extracted in step ST11, the job instance ID for identifying the job instance to be executed as the job of this job ID, and the job A second start instruction including the default value of the argument associated with the job ID in the job prefetch execution information 3 that matches the ID is transmitted to the job execution apparatus 20. In addition, the job prefetch execution analysis unit 14 sets the job instance ID in the second start instruction, the execution state indicating the execution in progress, and the execution type indicating the prefetch execution as the job ID in the second start instruction. Correlate and write to the job execution state table 4.

  On the other hand, when the job execution unit 22 receives the second activation instruction from the job activation control device 10, the job execution unit 22 activates the job based on the job ID, the job instance ID, and the default value of the argument in the second activation instruction. (ST14).

  When the job execution is completed, the job returns an execution result including the job ID, job instance ID, and return value to the job execution unit 22 (ST15).

  The job execution unit 22 transmits the execution result to the job prefetch execution analysis unit 14 (ST16).

  The job prefetch execution analysis unit 14 records the execution result in the job execution state table 4 (ST17). Specifically, when the job prefetch execution analysis unit 14 receives an execution result including a job ID, job instance ID, and return value from the job execution device 20, the job prefetch execution analysis unit 14 associates the return result with the job ID and job instance ID in the execution result. The value is written in the job execution state table 4 and the execution state associated with the job ID and job instance ID is updated to complete. It should be noted that this execution result record is clearly indicated by the execution type written after step ST13, not the regular execution result but the pre-read execution result.

  Next, details of step ST11 described above will be described with reference to FIGS.

  First, as shown in FIGS. 13 and 14, the job execution state table 4 is referred to extract a job being executed or completed (ST11-1). In this example, the first job job0 is being executed.

  Next, job dependency information 2 is referenced to extract job candidates that can be pre-read (ST11-2). In this example, the first, third, and fifth jobs job1, job3, and job5 are candidates for prefetching execution. The second, fourth, sixth, and final jobs job2, job4, job6, and job7 are not candidates for prefetching execution because they depend on the execution process of other jobs job1, job2, and job5 that have not been completed yet.

  Finally, as shown in FIG. 15, the job pre-read execution information 3 is referred to determine whether pre-read execution is possible, and a pre-read executable job is extracted (ST11-3). In this example, the first, third, and fifth jobs job1, job3, and job5 are all prefetchable. For example, since the third job job3 is “possible” for both rollback and correction, it can be read ahead. The first job job1 is rollback “impossible” and correction is “possible”. However, since the first job job1 is in parallel execution from the job net definition information 1 and the execution itself is not canceled, the prefetch execution is possible. Become.

  The above is the detailed description of step ST11.

  Next, FIG. 16 is a schematic diagram for explaining a correction operation for a job that has been pre-read. Supplementally, the job activation instruction by the job net analysis unit 13 and the job prefetch execution analysis unit 14 described with reference to FIGS. 11 and 12 normally operates independently. However, when the job net analysis unit 13 tries to transmit a job start instruction, if the job pre-read execution analysis unit 14 has already transmitted a pre-read execution start instruction, correction of job execution is required. The interaction between the job net analysis unit 13 and the job prefetch execution analysis unit 14 will be described with reference to FIG.

  As in step ST1, the job net analysis unit 13 refers to the job net definition information 1 and selects a job to be started next (ST21).

  As in step ST2, the job net analysis unit 13 refers to the job execution state table 4 and confirms the execution state of the job (ST22).

  If the job is prefetched as a result of step ST22, the job net analysis unit 13 issues a correction request for the job to the job prefetch execution analysis unit 14 (ST23), indicating that the job has been executed. Record in the job execution state table 4.

  Specifically, as a result of step ST22, when the job type in the job execution state table 4 matching the job ID selected in step ST21 is associated with the execution type indicating pre-read execution and the execution state indicating completion, The job net analysis unit 13 compares the default value of the argument associated with the job ID in the job prefetch execution information 3 that matches the selected job ID and the return value associated with the execution type indicating the prefetch execution. Then, it is determined whether or not they match. If the result of this determination shows that the two do not match and the argument correction information including the default value of the argument used for the determination indicates that argument correction is possible, the job net analysis unit 13 is associated with the argument correction information. A job prefetch execution analysis unit including a correction request including the job ID, the job instance ID associated with the job ID in the job execution state table matching the job ID, and an argument indicating the return value used for the determination. 14 to send. In addition, the job net analysis unit 13 associates the job instance ID in the correction request, the execution state indicating execution, and the execution type indicating main execution with the job ID in the correction request, and executes the job execution state table 4. Write to.

  The job prefetch execution analysis unit 14 transmits the correction request received in step ST23 to the job execution result correction unit 24 as a correction instruction (ST24).

  Upon receiving the transmitted correction instruction, the job execution result correction unit 24 corrects the job execution result based on the job ID, job instance ID, and argument in the correction instruction (ST25).

  When the correction of the job execution result is completed, the job returns the corrected execution result to the job execution result correction unit 24 (ST26).

  The job execution result correction unit 24 transmits the execution result to the job prefetch execution analysis unit 14 (ST27).

  The job prefetch execution analysis unit 14 records the execution result in the job execution state table 4 (ST28).

  In the job execution correction in step ST25 described above, the argument value is replaced with an actually determined value for a job that has been pre-read with the default value of the argument. The method of replacing the argument value with the actual value from the default value depends on the implementation of the individual job. For example, in the case of a job in which the default value of an argument is simply stored in the DB, it may be sufficient to replace the DB value with an actually determined value. If the job is a calculation process based on the default value of the argument, recalculation may be necessary based on the determined argument value.

  When transmitting a rollback instruction instead of the correction instruction, the above-described steps ST23 to ST28 are executed as in the following steps ST23 'to ST28'.

  If the job is prefetched as a result of step ST22, the job net analysis unit 13 sends a rollback request for the job to the job prefetch execution analysis unit 14 (ST23 '), and the job is actually executed. Is recorded in the job execution state table 4. Specifically, the job net analysis unit 13 executes the above-described determination. As a result of the determination, the rollback information associated with the default value of the argument used for the determination in the job prefetch execution information 3 is the rollback. On the other hand, if the argument correction information associated with this rollback information in the job prefetch execution information 3 indicates whether or not argument correction is possible, the job ID associated with the rollback information and the job ID A rollback request including the job instance ID associated with the job ID in the matching job execution state table 4 and the argument indicating the return value used for the determination is sent to the job prefetch execution analysis unit 14 (ST23 ′). ). In addition, the job net analysis unit 13 associates the job instance ID in the rollback request, the execution state indicating execution, and the execution type indicating main execution with the job ID in the rollback request, and executes the job execution state. Write to Table 4.

  The job prefetch execution analysis unit 14 transmits the rollback request received in step ST23 'to the job execution result correction unit 24 as a rollback instruction (ST24').

  Upon receiving the transmitted rollback instruction, the job execution result correcting unit 24 executes a rollback of the job based on the rollback instruction (ST25 ').

  When the re-execution by the rollback of the job is completed, the job returns an execution result including the job ID, job instance ID, and return value to the job execution result correction unit 24 (ST26 ').

  The job execution result correction unit 24 transmits the execution result to the job prefetch execution analysis unit 14 (ST27 ').

  The job prefetch execution analysis unit 14 records the execution result in the job execution state table 4 (ST28 ').

  As described above, according to the present embodiment, the job prefetch execution analysis unit 14 has a job ID different from the job ID associated with the execution state indicating execution or completion in the job execution state table 4, and the job A job ID that is not associated with the dependency content indicating that it is an execution process in the dependency information 2 is selected as a prefetch execution candidate, and rollback is possible in the job prefetch execution information 3 among the job IDs of the prefetch execution candidates. Alternatively, a job ID associated with the possibility of argument correction is selected, and a second start instruction including the earliest job ID, job instance ID, and default value of the argument among the selected job IDs is set as the job. With the configuration of transmitting to the execution device 20, the job is pre-read and executed in consideration of the dependency between jobs when executing the job net, and the surplus resources are allocated. It is possible to reduce the time required for execution of the entire job net (a collection of in execution order constrained jobs) and efficient use.

<Second Embodiment>
Next, a job execution system according to the second embodiment will be described.

  The second embodiment is a specific example of the first embodiment, and uses a task of “sorting a lot of data into normal / abnormal and processing normal data and abnormal data separately”, An example in which the processing time of the entire job net is shortened by job pre-read execution is shown.

  17 to 19 show the job net definition information 1, job dependency relationship information 2, and job prefetch execution information 3 in this example. In this example, it is assumed that the developer sets job dependency information 2 and job prefetch execution information 3 during system development, and job net definition information 1 is created / updated as needed by the system operation department during system operation. ing.

  In the job net definition information 1, as shown in FIG. 17, a series of job definitions for “separating a large number of data into normal / abnormal and processing normal data and abnormal data separately” are described. Yes. First, all data to be processed is classified into normal data and abnormal data by the head job job0. As the execution result of the first job job0, a normal data list and an abnormal data list are output.

  Next, conditional branching is performed according to the ratio of the number of abnormal data in all data including the number of normal data in the normal data list and the number of abnormal data in the abnormal data list. If the number of abnormal data is within 5% of the total data, the process proceeds to normal processing including the first and second jobs job1 and job2. If the percentage of abnormal data exceeds 5%, the process proceeds to exception processing for the third job job3.

  In normal processing, normal data processing by the first job job1 and abnormal data processing by the second job job2 are executed in parallel.

  As shown in FIG. 18, the job dependency relationship information 2 describes that the first to third jobs job1, job2, job3 depend on the return value of the first job job0.

  As shown in FIG. 19, the job prefetch execution information 3 describes that the second to third jobs job2 and job3 cannot perform both rollback and argument correction. That is, the second to third jobs job2 and job3 are processes that cannot be prefetched.

  An example of job execution is shown in FIG. In this example, the ratio of abnormal data is 5% or less, and the first and second jobs job1 and job2 are processed as usual.

  20A on the left side shows a conventional job execution example according to the job net definition information 1, and FIG. 20B on the right side shows an example of prefetching execution considering the job dependency information 2.

In the conventional job execution example, normal / abnormal data is first sorted by the first job job0. Next, normal data processing by the first job job1 and abnormal data processing by the second job job2 are executed in parallel. In this case, the execution of the first job job0 and the execution of the first job job1 are performed in “sequential parallel”, and this is the execution time (t _job0 + t _job1 ) of the entire job net.

In the prefetch execution example on the right side, normal data processing by the first job job1 is started in parallel with the sorting processing of the first job job0. All data is received as the default value of the argument, and the first job job1 is executed for all data. Thereafter, when the sorting process of the first job job0 is completed, the correction process of the first job job1 and the abnormal data process of the second job job2 are executed. For this reason, the execution time of the entire job net in the prefetch execution example is the total time (t _job0 + t _job2 ) of the execution time of the first job job0 and the execution time of the second job job2.

Here, in the prefetch execution example, the ratio of abnormal data is small, and the correction process for the first job job1 and the abnormal data process for the second job job2 are short. Therefore, the execution time of the prefetch execution example can shorten the execution time of the entire job net by the difference (t _job1 −t _job2 ) between the execution times of the conventional execution examples.

  As described above, according to the present embodiment, the same effect as that of the first embodiment can be obtained even in the configuration in which the job of the first embodiment is embodied.

  According to each of the embodiments described above, it is possible to determine whether or not a job can be prefetched by the configuration including the job dependency information 2 and the job prefetch execution information 3 in addition to the job net definition information 1, and the entire job net. It is possible to reduce the time required for execution.

  Note that the methods described in the above embodiments are, as programs that can be executed by a computer, magnetic disks (floppy (registered trademark) disks, hard disks, etc.), optical disks (CD-ROMs, DVDs, etc.), magneto-optical disks. (MO), stored in a storage medium such as a semiconductor memory, and distributed.

  In addition, as long as the storage medium can store a program and can be read by a computer, the storage format may be any form.

  In addition, an OS (operating system) running on a computer based on an instruction of a program installed in the computer from a storage medium, MW (middleware) such as database management software, network software, and the like realize the above-described embodiment. A part of each process may be executed.

  Furthermore, the storage medium in each embodiment is not limited to a medium independent of a computer, but also includes a storage medium in which a program transmitted via a LAN, the Internet, or the like is downloaded and stored or temporarily stored.

  Further, the number of storage media is not limited to one, and the case where the processing in each of the above embodiments is executed from a plurality of media is also included in the storage media in the present invention, and the media configuration may be any configuration.

  The computer in each embodiment executes each process in each of the above embodiments based on a program stored in a storage medium, and a single device such as a personal computer or a plurality of devices are connected to a network. Any configuration of the system or the like may be used.

  In addition, the computer in each embodiment is not limited to a personal computer, and includes an arithmetic processing device, a microcomputer, and the like included in an information processing device, and is a generic term for devices and devices that can realize the functions of the present invention by a program. Yes.

  In addition, although some embodiment of this invention was described, these embodiment is shown as an example and is not intending limiting the range of invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 1 ... Job net definition information, 2 ... Job dependence information, 3 ... Job prefetch execution information, 4 ... Job net definition information, 10 ... Job start control device, 11, 21 ... Storage part, 12 ... Job start control part, 13 ... Job net analysis unit 14. Job prefetch execution analysis unit 20 Job execution device 22 Job execution unit 23 Job start unit 24 Job execution result correction unit

Claims (3)

A job execution system including a job activation control device and a job execution device that can communicate with each other,
The job activation control device
Definition information storage means for storing job net definition information defining job execution order for each job, in which job IDs for identifying each job are described in the order of execution of the jobs;
Dependency content storage means for storing dependency relationships between the jobs;
Job correction information that associates and stores the job ID, rollback information indicating whether or not the job with the job ID can be rolled back, and argument correction information indicating whether or not argument correction of the job is possible and a default value of the argument. Storage means;
The job ID, a job instance ID for identifying a job instance executed as a job of the job ID, an execution state indicating that the state of the job instance is being executed, completed or failed, and execution of the job instance Is associated with the execution type indicating whether the execution is in accordance with the execution order or the pre-read execution executed before the execution order of the main execution and the return value of the job instance when the execution state indicates completion Execution state storage means for storing
Among the job IDs in the definition information storage unit that match the job IDs associated with both the execution type indicating the main execution and the execution state indicating completion in the execution state storage unit, in the definition information storage unit The job ID in the last execution order is extracted, the job ID arranged in the execution order next to the execution order of the extracted job ID is selected, and the job in the execution state storage unit that matches the selected job ID Specifying a job instance ID for identifying a job instance to be executed as a job of the selected job ID when both the execution type indicating pre-read execution and the execution state indicating execution are not associated with the ID, A first start instruction including the selected job ID and the specified job instance ID is transmitted to the job execution apparatus. And the dynamic instruction transmission means,
The job instance ID in the first start instruction, the execution state indicating execution, and the execution type indicating main execution are written in the execution state storage unit in association with the job ID in the first start instruction. Means,
The job ID is different from the job ID associated with the execution state indicating execution or completion in the execution state storage unit, and is associated with the dependency content indicating the execution process in the dependency content storage unit. Prefetch candidate selection means for selecting a job ID that has not been used as a job ID of a prefetch execution candidate;
Among the selected prefetch execution candidate job IDs, select a job ID associated with rollback information indicating whether rollback is possible or argument correction information indicating whether argument correction is possible in the job correction information storage unit, Among the job IDs in the definition information storage means matching the selected job ID, the job ID arranged in the earliest execution order and the job instance ID for identifying the job instance executed as the job of this job ID And a second start instruction transmitting means for transmitting to the job execution apparatus a second start instruction including a default value of an argument associated with the job ID in the job correction information storage means matching the job ID When,
The job instance ID in the second start instruction, the execution state indicating execution, and the execution type indicating pre-read execution are written in the execution state storage unit in association with the job ID in the second start instruction. Means,
Upon receiving an execution result including a job ID, job instance ID, and return value from the job execution device, the return value is written in the execution state storage means in association with the job ID and job instance ID in the execution result, and Means for updating the execution state associated with the job ID and job instance ID to complete;
Among the job IDs in the definition information storage unit that match the job IDs associated with both the execution type indicating the main execution and the execution state indicating completion in the execution state storage unit, in the definition information storage unit The execution status storage unit that extracts the job ID of the last execution order, selects the job ID arranged in the execution order next to the execution order of the extracted job ID, and matches the job ID of the next execution order Is associated with the job ID in the job correction information storage unit that matches the job ID of the next execution order. Means for comparing the default value of the argument and the return value associated with the execution type indicating the prefetch execution, and determining whether or not both match,
As a result of the determination, when the two do not match, the rollback information associated with the default value of the argument used for the determination in the job correction information storage means indicates that rollback is possible, while the rollback information If the argument correction information associated in the job correction information storage means indicates argument correction or not, the job ID associated with the rollback information and the job in the execution state storage means that matches this job ID Means for sending a rollback request including a job instance ID associated with the ID and an argument indicating a return value used for the determination;
Rollback instruction transmission means for transmitting the sent rollback request to the job execution apparatus as a rollback instruction;
With
The job execution device includes:
Job starting means for starting a job based on the job ID and job instance ID in the start instruction when the first or second start instruction is received from the job start control device;
Means for returning the execution result of the activated job to the job activation control device;
When the rollback instruction is received from the job activation control device, rollback execution means for executing a rollback of the job based on the rollback instruction;
Means for returning the execution result of the rollback again to the job activation control device;
A job execution system comprising: The job execution system according to claim 1,
The job activation control device
As a result of the determination, if the two do not match, if the argument correction information including the default value of the argument used for the determination indicates that argument correction is possible, the job ID associated with the argument correction information and this job Means for sending a correction request including a job instance ID associated with the job ID in the execution state storage means matching the ID, and an argument indicating a return value used for the determination;
Correction instruction transmission means for transmitting the sent correction request to the job execution apparatus as a correction instruction;
Further comprising
The job execution device includes:
Job execution result correction means for correcting the job execution result based on the job ID, job instance ID, and argument in the correction instruction when receiving the transmitted correction instruction;
Means for returning the corrected execution result to the job activation control device;
A job execution system further comprising: A program that is communicable with a job execution device and that is used in a job activation control device that includes definition information storage means, dependency content storage means, job correction information storage means, and execution state storage means,
The job activation control device;
Job IDs for identifying the respective jobs are described in the order of execution of the respective jobs, and in the case where some of the jobs are branched and executed in parallel, Means for writing job net definition information further describing the branching condition in association with the job ID of each job in the definition information storage means;
Dependent content indicating that the job ID, the dependent job ID for identifying the dependent job on which the job with the job ID depends, and that the dependent content is an execution process or a return value of the dependent job And means for writing to the dependence content storage means in association with each other,
The job correction information is stored in association with the job ID, rollback information indicating whether or not the job with the job ID can be rolled back, and argument correction information indicating whether or not argument correction of the job is possible and a default value of the argument. Means to write means,
The job ID, a job instance ID for identifying a job instance executed as a job of the job ID, an execution state indicating that the state of the job instance is being executed, completed or failed, and execution of the job instance Is associated with the execution type indicating whether the execution is in accordance with the execution order or the pre-read execution executed before the execution order of the main execution and the return value of the job instance when the execution state indicates completion Means for writing to the execution state storage means,
Among the job IDs in the definition information storage unit that match the job IDs associated with both the execution type indicating the main execution and the execution state indicating completion in the execution state storage unit, in the definition information storage unit The job ID in the last execution order is extracted, the job ID arranged in the execution order next to the execution order of the extracted job ID is selected, and the job in the execution state storage unit that matches the selected job ID Specifying a job instance ID for identifying a job instance to be executed as a job of the selected job ID when both the execution type indicating pre-read execution and the execution state indicating execution are not associated with the ID, A first start instruction including the selected job ID and the specified job instance ID is transmitted to the job execution apparatus. Dynamic instruction transmission means,
The job instance ID in the first start instruction, the execution state indicating execution, and the execution type indicating main execution are written in the execution state storage unit in association with the job ID in the first start instruction. means,
The job ID is different from the job ID associated with the execution state indicating execution or completion in the execution state storage unit, and is associated with the dependency content indicating the execution process in the dependency content storage unit. Prefetch candidate selection means for selecting a job ID that has not been used as a prefetch execution candidate job ID;
Among the selected prefetch execution candidate job IDs, select a job ID associated with rollback information indicating whether rollback is possible or argument correction information indicating whether argument correction is possible in the job correction information storage unit, Among the job IDs in the definition information storage means matching the selected job ID, the job ID arranged in the earliest execution order and the job instance ID for identifying the job instance executed as the job of this job ID And a second start instruction transmitting means for transmitting to the job execution apparatus a second start instruction including a default value of an argument associated with the job ID in the job correction information storage means matching the job ID ,
The job instance ID in the second start instruction, the execution state indicating execution, and the execution type indicating pre-read execution are written in the execution state storage unit in association with the job ID in the second start instruction. means,
Upon receiving an execution result including a job ID, job instance ID, and return value from the job execution device, the return value is written in the execution state storage means in association with the job ID and job instance ID in the execution result, and Means for updating the execution state associated with the job ID and job instance ID to completion,
Among the job IDs in the definition information storage unit that match the job IDs associated with both the execution type indicating the main execution and the execution state indicating completion in the execution state storage unit, in the definition information storage unit The execution status storage unit that extracts the job ID of the last execution order, selects the job ID arranged in the execution order next to the execution order of the extracted job ID, and matches the job ID of the next execution order Is associated with the job ID in the job correction information storage unit that matches the job ID of the next execution order. Means for comparing the default value of the argument and the return value associated with the execution type indicating the pre-read execution, and determining whether or not both match,
As a result of the determination, when the two do not match, the rollback information associated with the default value of the argument used for the determination in the job correction information storage means indicates that rollback is possible, while the rollback information If the argument correction information associated in the job correction information storage means indicates argument correction or not, the job ID associated with the rollback information and the job in the execution state storage means that matches this job ID Means for sending a rollback request including a job instance ID associated with the ID and an argument indicating a return value used for the determination;
Rollback instruction transmission means for transmitting the sent rollback request to the job execution apparatus as a rollback instruction;
Function as
As the job execution device,
When the first or second activation instruction is received from the job activation control device, the job is activated based on the job ID and job instance ID in the activation instruction, and the execution result of the job is displayed on the job activation control device. Reply to
A program that, when receiving the rollback instruction from the job activation control apparatus, executes a rollback of the job based on the rollback instruction, and returns the execution result of the rollback again to the job activation control apparatus.

Images