JP2002014829A - Parallel processing control system, method, and medium storing program for parallel processing control - Google Patents

Abstract

(57) [Summary] [Purpose] Business data is divided in units of jobs and parallel processing is performed to reduce processing time. [Configuration] The parallel processing control program PGM2 reads parallel multiplicity, sequentially reads job definitions from the parallel processing management table ZA07, passes the job definition to the job management program K01 until the parallel multiplicity is reached, and executes the job being executed. Information about means of job management program K01
And determines whether there is a job execution unit to which job execution can be assigned. The job management program K01 to which the job definition has been passed passes the job definition to one of the job execution units that responds the fastest, activates the job execution unit, and monitors the job execution status of a plurality of job execution units. I do. The started job activating unit activates the business program PGM1, reads the business data corresponding to the job definition, and executes the business process.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention records a parallel processing control program for controlling a plurality of job execution means to process a group of business data divided into a plurality of groups in parallel using a job management program. And a computer (parallel processing control system) in which the parallel processing control program is installed, and a parallel processing control method according to the program.

[0002]

2. Description of the Related Art Programs for executing processing on receipt / shipment data, sales processing, and the like are generally referred to as business programs. However, since the amount of business data is enormous, an extremely long processing time is required. .

[0003]

DISCLOSURE OF THE INVENTION

An object of the present invention is to reduce the processing time by focusing on a plurality of functions of a business program (functions for executing a plurality of different types of jobs) and performing parallel processing separately for each function. And

Another object of the present invention is to enable the number of parallel processes (parallel multiplicity) to be changed and to easily adjust to an appropriate parallel multiplicity.

The present invention is based on the premise that business data to be processed by a business program is divided into a plurality of groups in advance according to a plurality of functions (a plurality of types of job execution functions) of the business program.

[0007] A plurality of job executing means are provided to realize parallel processing. These job execution means are generally a plurality of computers, and the same business program is installed on each computer. When each computer is given a job definition representing a job type (or specifying a job), the job of that type is executed in accordance with a business program (processing that is a unit of business, such as receipt / shipment processing and sales processing). Execute In this job execution, business data corresponding to the job is processed.

That is, there is a plurality of jobs which can be a unit of parallel processing (that is, can be executed without being affected by other jobs or other data), and a job is assigned to each job. Data is divided into multiple groups. One group of data is processed by one job. Generally, the number of divisions of business data into a plurality of groups is larger than the number of job execution units.

A plurality of job execution means can be realized by a single high-speed computer, for example, by multitask processing.

According to the present invention, job identification data representing a job for processing data of a business data group divided into a plurality of groups corresponding to job types, data representing the execution status of those jobs, and the like. Is provided.

Further, according to the present invention, the job execution status of a plurality of job execution means (executing, normal termination, abnormal termination, etc.) is monitored, and when a job definition including job identification data is given, the job definition is provided. There is a job management means (or a job management program) for giving a job execution means to execute a corresponding job. This job management means (program) may be an existing one. For example, one of the ERP (enterprise resource planning) package systems (also called integrated business package systems), "R /
A job management program which is a standard function of "3 systems" (trademark of SAP) can be used.

A parallel processing control system according to the present invention comprises:
Obtains information about the job execution status of the job execution unit from the job management unit, determines whether there is a job execution unit to which a job can be assigned, and if there is a job execution unit to which a job can be assigned. And a parallel processing control means for determining a job to be executed next with reference to the parallel processing management table and providing a job definition including identification data of the job to the job management means.

The program recording medium according to the present invention obtains information on the job execution status of the job executing means from the job management program, determines whether there is a job executing means to which a job can be assigned, and If there is a job execution means that can be assigned, the computer refers to the parallel processing management table to determine the next job to be executed, and controls the computer to provide a job definition including identification data of the job to the job management program. Is a medium storing a parallel processing control program to be executed.

The present invention also provides a method for operating a computer according to the above program.

According to the present invention, an enormous amount of business data can be divided and processed by a plurality of job executing means for each job, so that business data can be processed efficiently and the processing time can be shortened. .

In particular, when the job management means (program) has a function of sequentially selecting the job execution means having the least load among the plurality of job execution means and providing the job definition, the job management means (program) is most efficient. Since the job is executed, the processing time can be further reduced.

The parallel processing control means (program) updates the parallel processing management table based on the acquired information on the job execution status.

In a preferred embodiment, the parallel processing control means (program) refers to a given parallel multiplicity (for example, preset in a table), and determines whether the number of job executing means during job execution is parallel. If the number is smaller than the number represented by the multiplicity, it is determined that there is a job execution unit to which a job can be assigned.

The user can arbitrarily set the parallel multiplicity or change the parallel multiplicity to a value suitable for the business process at the start of the business process. The number of job execution units corresponding to the number represented by the set parallel multiplicity operates. The most efficient setting of the parallel multiplicity can be easily realized.

In another preferred embodiment, a business data table storing parallel processing keys corresponding to business data groups to be processed by each job is further provided. The parallel processing management table stores parallel processing keys related to business data groups to be processed by the job corresponding to each job identification data. The parallel processing key is a code for identifying a group of business data processed by one job.

The above parallel processing control means (program)
A parallel processing key corresponding to the job definition is given to the job management means (program). The job management means (program) provides the job execution means with a parallel processing key corresponding to the job definition. As a result, the job execution means refers to the business data table, and stores the business data corresponding to the given parallel processing key.
Process group data.

If an easy-to-understand number or the like is used as the parallel processing key, the handling becomes easier than the job identification data.

In still another embodiment, the job management means (program) is provided with a job status table. The job status table stores, in correspondence with the job identification data, the identification code of the job execution means executing the job and data representing the job execution status. The parallel processing control means (program) acquires information on the job execution status of the job execution means with reference to the job status table.

The most preferred embodiment of the present invention can be described as follows from the viewpoint of the operation of the database server according to the program.

The database server has a job management program having a function of monitoring the job execution status of a plurality of job execution means and passing the job definition to the job execution means when a job definition is given (for example, "job management program"). R / 3 background job management ”).

The business data to be processed is divided into a plurality of groups, and each group is given an identification number called a parallel processing key so that the processing according to the job definition can be executed in parallel by a plurality of job executing means. . The job executing means executes a job according to the job definition in units of the parallel processing key.

First, the parallel processing control program is started in the database server, and the parallel multiplicity is read.

The parallel processing control program sequentially reads the jobs to be executed from the parallel processing management table, and passes the job definition and the parallel processing key to the job management program until the number of executed processing reaches the parallel multiplicity. Thereafter, the parallel processing control program obtains information on the job executing means that is being executed (or has been executed) from the job management program, and determines whether there is a job executing means to which job execution can be assigned.

The job management program to which the job definition has been transferred passes the job definition and the parallel processing key to any one of the job execution units that responded the fastest, activates the job execution unit, and executes a plurality of job execution units. Monitor job execution status.

The started job executing means reads the business data corresponding to the job definition using the parallel processing key, and executes the processing according to the business program.

In this way, compared with the case where the business data is not divided, the business data can be remarkably efficiently processed by the plurality of job executing means, and the processing time can be reduced. .

In a system such as an R / 3 system, which can handle a plurality of job executing means for executing a job definition and has a load distribution function for transferring the job definition to any of the quickest responding job executing means. Can be more effective.

[0033]

FIG. 1 shows the overall configuration of a parallel processing control system (for example, the above-mentioned R / 3 system).
In this system, multiple application servers SV
2, SV3... SVnn (the number, that is, nn + 1 depends on the system configuration) and one database server SV1 are communicably connected via a network. When referring to general application servers, SVjj (j ≦
n).

A business program PGM1 is set on the application server SVjj. Business program PGM1
Is started according to the job definition given from the job management program K01 executed on the database server SV1, and processes the business data corresponding to the passed parallel processing key.

The job definition stores a job name, a job NO (not necessarily required because it corresponds to the job name), a business program name, and a variant (information (processing date, etc.) necessary for operating the business program PGM1). Data indicating the location where the user is performing the job. The content of the variant is set corresponding to the parallel processing key, and there are as many as the number of parallel processing keys.

A business data table creation program PGM3, a parallel processing control program PGM2, and a job management program K01 (R / 3 system standard function) are set on the database server SV1. On the database server SV1, a job status summary table TBTCO,
Parallel processing management table ZA07, parallel multiplicity management table ZA
06 and business data tables ZM01 to ZMmm.

The job management program K01 monitors a job execution status executed on a plurality of application servers SVjj. When the job management program KO1 receives the job definition and the parallel processing key from the parallel processing control program PGM2, the job management program KO1 also passes the job definition and the parallel processing key to the application server SVjj that responded the fastest. Start PGM1. The job management program K01 monitors the job execution status of the business program PGM1.

The job status summary table TBTCO (see FIG. 6) is managed by the job management program K01.
Stores the execution status of the started job. That is, when the job management program KO1 starts the business program PGM1, the job name, the job NO, and the started application server SVjj (target system) are written into the job status summary table TBTCO. The job status is written according to the job execution status of the business program. The job status symbols are F for normal termination, R for processing, and A for abnormal termination.

Parallel multiplicity management table ZA06 (see FIG. 8)
Describes the job net classification set by the user, the parallel multiplicity, and the identification code of the parallel processing management table to be referred to. Parallel processing control program PGM2
Reads the number of jobs to be executed in parallel (parallel multiplicity) from the parallel multiplicity management table ZA06 when started, and the number of jobs running on the application server SVjj becomes the number represented by the parallel multiplicity. Until the job definition and parallel processing key to be processed, the job management program K0
Pass to 1. The parallel processing control program PGM2 monitors the job status summary table TBTCO and determines whether there is an executable application server SVjj.

Parallel processing management table ZA07 (see FIG. 7)
Is used by the parallel processing control program PGM2, and is stored in a state where the parallel processing key, the job name, and the job number correspond in advance. The job status is updated each time by the parallel processing control program PGM2.

The business data table ZM01 describes the correspondence between business data groups divided according to jobs and parallel processing keys (ie, jobs). The business data table ZM01 contains data identification of business data and record No. , The parallel processing key, the processing status data, the processing date, and the business data to be processed (or the root of the storage location of the business data).
Those having the same parallel processing key form a business data group to be processed by one type of job. The processing status is updated by the business program PGM1, where Y indicates normal termination, H indicates suspension, E indicates an error, and is unprocessed or is being processed.
The processing date is a date to be processed. Business program PGM1
Executes the process using the business data of the group corresponding to the processing key passed from the job management program K01.

The user prepares the parallel processing management table in advance according to the business data table creation program PGM3.
ZA07 (excluding job status) and business data table
Create ZM01 (excluding the processing status). Also,
The parallel processing control program PGM is stored in the database server SV1.
Prior to the start of 2, the business program name is set.

FIGS. 2 to 5 show the parallel processing control program PG.
It shows the operations in M2, the job management program K01, and the application servers SV2 and SV3 in association with each other. In this embodiment, the number of divisions of business data to be processed is three. Accordingly, the parallel processing keys are set to 1 to 3. The parallel multiplicity is 2.

The parallel processing control program PGM2 is started (step 1). The started parallel processing control program PG
M2 is obtained from the parallel multiplicity management table ZA06 from the parallel multiplicity (=
2) is read (step 2). Also, this table ZA06
Describes the parallel processing management table ZA07 to be referred to, so that the program PGM2 executes the parallel processing management table ZA07.
, The first parallel processing key to be processed (= 1),
And the job name corresponding to this parallel processing key (= JOB
1) and the job number (= 00000001) are read (step 3).

The parallel processing control program PGM2 passes the job definition (job name = JOB1, business program name = PGM1) and the parallel processing key (= 1) to the job management program K01.
The parallel processing management table ZA07 is updated (R: processing is entered as the job status corresponding to JOB1). That is, in the parallel processing management table ZA07, the record with the parallel processing key = 1, the job N0 = 00000001, and the job status =
R (under processing) (step 4).

The job management program K01 issues a request to the application server SVjj. The application server that has the most margin among the application servers SV2 to SVnn responds quickly. Application server
Assuming that SV2 responds fastest, the job management program K01 passes the business program name (= PGM1) and the parallel processing key (= 1) to the application server SV2, and sends the job JOB1 to the job status summary table TBTCO.
(Job name = JOB1, job NO = 00000001, job status = R), and the end monitoring of job JOB1 is started (steps 5, 7, 8).

The application server SV2 starts business processing according to the business program PGM1. The business program PGM1 of the application server SV2 is the business data table Z
The business data corresponding to the parallel processing key (= 1) is read from M01 and processed (step 6).

Next, the parallel processing control program PGM2 obtains, from the parallel processing management table ZA07, the parallel processing key to be processed next (= 2) and the corresponding job name (= JOB2).
And job NO (= 00000002) are read (step 9).

The parallel processing control program PGM2 creates a job definition using the read data, and the job definition (job name = JOB2, job NO = 00000002, business program name = PGM1) and the parallel processing key (= 2) is passed to the job management program K01, and the parallel processing management table ZA07 is updated. In the parallel processing management table ZA07, the record with the parallel processing key = 2, the job number = 00000002 and the job status = R (step 10).

The job management program K01 issues a request to the application server SVjj and sends the business program name (= P
GM1) and the parallel processing key (= 2), and register the information of job JOB2 (job name = JOB2, job NO = 00000002, job status = R (running)) in the job status summary table TBTCO, Monitoring of the end of the process is started (steps 11, 13, and 14). Application server SV3
Starts the business program PGM1. Business program PGM1
Reads the business data corresponding to the parallel processing key (= 2) from the business data table ZM01 and performs processing (step 1).
2).

Since the number of jobs being processed has reached the parallel multiplicity (= 2), the parallel processing control program PGM2 retrieves job status data from the job status summary table TBTCO for the jobs being processed (JOB1, JOB2) at intervals of 10 seconds. And monitors the end of each job (step 15).

[0052] The job management program K01 is
Monitor the end of 1 and JOB2 (step 16).

When the business program PGM1 being processed on the application server SV3 ends, the job management program K01 detects the end of the corresponding job JOB2 (steps 17, 18). The job management program K01 updates the job status data of the job JOB2 in the job status summary table TBTCO (step 19). That is, in the job status summary table TBTCO, job name = JOB
In the records 2 and 3, the job status = F (normal end) in the record of job NO = 00000002.

The parallel processing control program PGM2 executes the job JOB2 based on the contents of the job status summary table TBTCO.
And ends the parallel processing management table ZA07 (steps 20 and 21). That is, the parallel processing management table
In ZA07, the target record is a record with the parallel processing key = 2, and the job status = F (normal end).

In the parallel processing control program PGM2, since the number of jobs being processed has become less than the parallel multiplicity (= 2), the parallel processing key (= 3) to be processed next and the job name (=
JOB3) and job NO (= 00000003) are read from the parallel processing table ZA07. (Step 22).

The parallel processing control program PGM2 passes the job definition (job name = JOB3, job No. = 00000003, business program name = PGM1) and the parallel processing key (= 3) to the job management program K01, and the parallel processing management table ZA07. Is updated (step 23). In the parallel processing table ZA07,
Record with parallel processing key = 3, job NO = 00000003,
Job status = R (processing).

The job management program K01 starts up for the job JOB3, passes the business program name (= PGM1) and the parallel processing key (= 3) to the application server SV3 that responded the fastest, and sends the job JOB3 to the job status summary table TBTCO. Information (job name = JOB3, job N
O = 00000003, job status = R) is registered, and job JOB3
Of the end of the process is started (steps 24, 26, 27). The application server SV3 starts the business program PGM1. That is, the business program PGM1 reads the business data corresponding to the parallel processing key (= 3) from the business data table ZM01 and performs processing (step 25).

Since the parallel processing control program PGM2 has passed all the jobs to be processed to the job management program K01, the parallel processing control program PGM2 acquires job information from the job status summary table TBTCO at 10-second intervals for the jobs being processed (JOB1, JOB3). The end is monitored (step 28).

The job management program K01 is a job
1. Monitor the end of JOB3 (step 29).

When the business program PGM1 being processed on the application server SV2 ends, the job management program K01 detects the end of the corresponding job JOB1 (steps 30 and 31). The job management program K01 updates the job information of the job JOB1 in the job status summary table TBTCO. Update contents are: job name = JOB1, job NO =
In the record of 00000001, the job status = F (normal termination) (step 32).

The parallel processing control program PGM2 is a job JOB1.
And the parallel processing management table ZA01 is updated. For the record with the parallel processing key = 1, the job status = F. (Steps 33, 34).

The job management program K01 is OB3 at regular intervals.
Is monitored (step 35).

The parallel processing control program PGM2 acquires job information from the job status summary table TBTCO at 10-second intervals for the job being processed (JOB3) and monitors the end (step 36).

When the business program PGM1 being processed on the application server SV3 ends, the job management program K01 detects the end of the corresponding job JOB3 (steps 37 and 38). The job management program K01 updates the job information of the job JOB3 in the job status summary table TBTCO. Update contents are: job name = JOB3, job
In the record of NO = 00000003, the job status = F (normal end) (step 39).

The parallel processing control program PGM2 is the job JOB3
And the parallel processing management table ZA07 is updated. With the record of the parallel processing key = 3, the job status = F (steps 40 and 41).

The parallel processing control program PGM2 ends because it detects that all the jobs to be processed have been completed (step
42).

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a parallel processing control system.

FIG. 2 is a flowchart illustrating a flow of parallel processing control.

FIG. 3 is a flowchart illustrating a flow of parallel processing control.

FIG. 4 is a flowchart illustrating a flow of parallel processing control.

FIG. 5 is a flowchart illustrating a flow of parallel processing control.

FIG. 6 shows an example of the contents of a job status summary table TBTCO.

FIG. 7 shows an example of the contents of a parallel processing management table ZA07.

FIG. 8 shows an example of the content of a parallel multiplicity management table ZA06.

FIG. 9 shows an example of the contents of a business data table ZM01.

[Explanation of symbols]

 SV1 database server SV2 application server SV3 application server SVjj application server PGM1 business program PGM2 parallel processing control program PGM3 business data table creation program K01 job management program TBTCO job status summary table ZA06 parallel multiplicity management table ZA07 parallel processing management table ZMnn business data table

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Ayako Shimada 16th Ichibancho, Chiyoda-ku, Tokyo Inside the Japan Research Institute, Inc. (72) Inventor Masataka Yahagi 16th Ichibancho, Chiyoda-ku, Tokyo Japan Comprehensive Research Inc. (72) Inventor Junichi Kobashi 16th Ichibancho, Chiyoda-ku, Tokyo F-term in Japan Research Institute, Limited (Reference) 5B045 GG04 GG09 GG11 JJ02 KK06 5B098 AA10 GA03 GA07 GB01 GD02 GD14

Claims (10)

[Claims] A job execution status of a plurality of job execution means is monitored, and when a job definition including job identification data is given, the job definition is given to any one of the job execution means to execute a corresponding job. Job management means for storing job identification data representing a job for processing data of a business data group divided into a plurality of groups corresponding to job types and data representing the execution status of those jobs Obtains information on the job execution status of the job execution means from the parallel processing management table and the job management means, determines whether there is a job execution means to which the job can be assigned, and assigns the job. Job that should be executed next by referring to the parallel processing management table Determined, the parallel-processing control system including a parallel-processing control unit, to provide a job definition including identification data of the job to the job management unit. 2. The method according to claim 1, wherein said parallel processing control means refers to a given parallel multiplicity and allocates a job if the number of job executing means during job execution is smaller than the number represented by the parallel multiplicity. 2. The parallel processing control system according to claim 1, wherein it is determined that there is a possible job execution unit. 3. A job data table storing a parallel processing key corresponding to a job data group to be processed by each job, and the parallel processing management table includes a job data table corresponding to each job identification data. A parallel processing key associated with a business data group to be processed by a job is stored, and the parallel processing control means supplies a parallel processing key corresponding to the job definition to the job management means, and the job management means The job execution unit is provided with the parallel processing key corresponding to the job definition together with the job definition, whereby the job execution unit refers to the business data file and generates the business data group corresponding to the given parallel processing key. 3. The parallel processing control system according to claim 1, wherein the parallel processing control system processes data. 4. The job management means according to claim 1, wherein
The job status table stores an identification number of the job execution means executing the job and data representing the job execution status in correspondence with the job identification data. The parallel processing control system according to any one of claims 1 to 3, wherein the means acquires information about a job execution status of the job execution means by referring to the job status table. 5. The parallel processing control system according to claim 1, wherein said parallel processing control means updates the parallel processing management table based on the acquired information on the job execution status. 6. A job execution status of a plurality of job execution units is monitored, and when a job definition including job identification data is given, the job definition is given to one of the job execution units to execute the corresponding job. This is a control method for causing a plurality of job execution units to process data of a business data group divided into a plurality of groups in parallel by using a job management unit for causing the plurality of business data groups to be processed. A parallel processing management table for storing job identification data representing a job for processing data and data representing the job execution status is provided, and information on the job execution status of the job execution means is obtained from the job management means. And determine whether there is a job execution means that can assign the job,
If there is a job execution unit to which a job can be assigned, the next job to be executed is determined by referring to the parallel processing management table, and a job definition including the identification data of the job is given to the job management unit. Control method. 7. A job execution unit to which a job can be assigned if the number of job executing means during job execution is smaller than the number represented by the given parallel multiplicity with reference to the given parallel multiplicity. 7. The parallel processing control method according to claim 6, wherein it is determined that there is a means. 8. A job data table storing a parallel processing key corresponding to a job data group to be processed by each job, and the parallel processing management table includes a job data table corresponding to each job identification data. The parallel processing key associated with the business data group to be processed by the job is stored, and the parallel processing key corresponding to the job definition is provided to the job management means, so that the job management means stores the job definition together with the job definition. 8. The job execution unit is provided with a corresponding parallel processing key, and the job execution unit refers to the business data table and processes the data of the business data group corresponding to the given parallel processing key. 3. The parallel processing control method according to 1. 9. A job execution status of a plurality of job execution units is monitored, and when a job definition including job identification data is given, the job definition is given to one of the job execution units to execute the corresponding job. A control program for causing the plurality of job execution means to process data of the business data group divided into a plurality of groups in parallel by using the job management program to be executed. Information about the job execution status of each job, determine whether there is a job execution unit that can assign jobs, and if there is a job execution unit that can assign jobs, Job identification data representing jobs for processing each data and the execution status of those jobs A parallel processing control program that determines a job to be executed next with reference to the parallel processing management table storing the data representing the job and a job definition including the identification data of the job is provided to the job management program. The stored medium. 10. Referring to a given parallel multiplicity, if the number of job executing means during job execution is less than the number represented by the given parallel multiplicity, job execution capable of assigning a job is performed. The medium according to claim 9, wherein the parallel processing control program for controlling a computer to determine that there is a means is stored.