JPH03226856A - Program distributed arranging and executing method - Google Patents

Abstract

PURPOSE:To efficiently design the system by placing a module on an optimal computer by a directory being independent from a program. CONSTITUTION:In the computer system, plural host computers 1, 8 are connected through a communication circuit, and at least one set of plural host computers 1, 8 is provided with a program module 2 of a subroutine unit, and a directory 3 for showing to which computer the subroutine is loaded on and executed. In such a state, in accordance with a distributed arrangement of the program module shown in the directory 3, the computer generates automatically a commu nication processing program, and transfers each subroutine to the execution computer designated by the directory 3. In such a way, by the directory 3 instead of the program, the module is placed on the optimal computer, therefore, the system is designed efficiently.

Description

Detailed Description of the Invention [Industrial Application Field] The present invention provides a system in which a plurality of host computers are connected via a communication line, in which a certain host computer takes out a module and stores it in the host according to the program execution order. The present invention also relates to a method for executing distributed program placement by communicating with a host module and executing the program while transferring control.

[Prior Art] Conventionally, in a multiprocessor system in which multiple processors are connected on a LAN, each processor node is equipped with a storage management unit, and the memory distributed in each node can be used as shared memory by multiple processors. A method has been proposed (for example, Kuchigaru Electronics 1988.10.3 (No. 457) pp. 94
-95). The above-mentioned techniques do not consider arranging programs in separate memories.

Regarding the decentralization of program placement, the issue of which computers on the network should be placed on each program is being considered (for example, 'SAA distributed processing' A, L, 5cherr, [
rIBM SYSTEM JOUNA J G32
1-0091-00, 1988).

However, the technology described above does not consider the method of distributing the program in module units, and the program creator must be aware of the distributed arrangement in advance.
Generally, a method of programming using a communication interface or the like has been considered.

[Problem to be solved by the invention] As mentioned above, when databases are distributed, the amount of data accessed to the database becomes extremely large compared to the amount of data handed over between modules even for one business. , it is more efficient to execute on a different computer. In the conventional method, when creating a program, it is necessary to set in advance which modules will be placed where, and then perform programming with these placements in mind.

However, when considering using a program consisting of multiple modules in various types of networks, it is necessary to create the program without being aware of where each module will be placed. This is desirable.

The purpose of the present invention is to solve such conventional problems, to be able to place program module calls in an optimal computer-1-, and to allow program creators to do so without being aware of where each module is placed. An object of the present invention is to provide an efficient program distributed execution method that allows programming to be performed in a timely manner and at the same time suppresses the amount of data transfer.

[Means for Solving the Problems] In order to achieve the two-pronged objective, the program distributed execution method of the present invention provides a program module in subroutine units to at least one of a plurality of host computers, and a subroutine on which computer. The computer automatically generates a communication processing program according to the distributed arrangement of program modules indicated in the directory, and executes each subroutine on the execution computer specified in the directory. It is characterized by transfer.

[For production]

In the present invention, each system designer determines in advance on which computer in each system each module of a program should be placed, and stores this information in a directory. In other words, the directory maintains a table that indicates which host computer in the network should load and execute each subroutine, and each host computer loads the module from the directory before executing the program. Reads the location information and transfers the module to the computer specified in that directory. Further, a communication function is added to the module as necessary.

As a result, program creators do not need to be aware of where each module is located; in other words, program creators can simply think of each module as if it were all being executed on a single computer. A third party, such as a system designer, specifies computer threads to be placed for each module by directing outside the program, and the program is distributed and executed on optimal computers.

As a result, the amount of data communication can be reduced and efficient system design is possible.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a distributed computer system showing an embodiment of the present invention.

In FIG. 1, ■ is a computer (A) that has a function of transferring modules to other computers, and 2 is a modular program to be distributed, which is modularized in subroutine units. 3 is a directory that holds a table indicating on which computer each module is to be executed, and 4 is a directory that recognizes a group of modules to be transferred together from the information held by program 2 and BUILEF I.S3.
a program distribution unit that determines where to transfer the module group and the structure of the variable list used therein;
5 is a program modification section within the own computer that changes the local CALL part of program 2 to remote data transmission/reception; 6 is a program modification section within the computer that acquires actual variable values from computer 1 to the first and last parts of the module to be transferred. 5END/RE
A communication program addition unit that adds a CV program to the program 2; 7 a program transfer unit that sends the structure of a module group or a variable list to be passed to the module; 8 a module transfer unit that does not have the program 2 or the program distribution unit 4; General computers subject to (B)
, 9 is a program transfer unit provided in the general computer 8 and receives the module sent from the computer 1.

Among these, directory 3 is a table in memory, and other programs 2, program distributed allocation unit 4,
The internal computer program modification section 5 and the communication program generation section 6 are program modules having respective functions, and their functions are executed by being executed by the processing device.

The program transfer unit 7 includes hardware and a program for transmitting and receiving program data.

In this computer system, a plurality of host computers 1 and 8 are connected via communication lines, and execute computer information on a subroutine unit module specified by the user, for example, information on which computer the module is executed on. Based on this, the program distributed allocation unit 4 individually extracts the subroutine modules of the program 2 from the program library, transfers the program 2 to the own computer 1 or another computer 8 according to the contents of the directory 3, and determines the execution order of the program 2. Accordingly, communication is performed with the module in the computer 1 or 8, and execution is performed while transferring control.

The directory 3 stores in advance, by each system designer, a table indicating on which computer in each system each module of a program should be placed. Here, it is shown that module 1 should be placed in computer Kl, module 2 in computer 2, . . . module N should be placed in computer KN, respectively.

The computer 1 includes a program library consisting of a plurality of programs 2 and a directory 3, and after adding a communication program, transfers the corresponding module to the computer specified by the directory.

Program 2 here involves inputting a message, calling module l using variable list 1,
In the same way, calling modules 2-N using variable lists 2-N and outputting a message are shown.

An example of such a distributed program deployment method is a counter service at a telephone office that checks equipment inventory, confirms open construction dates, notifies customers of construction openings, and manages statistical information. etc. In this business, inventory database,
When the construction schedule data bus and statistical database are distributed on different computers, a module is executed on the computer where the database used by that module exists, and the takeover information is transferred to the next module.

In addition to the above-mentioned tasks, there are many other examples in a distributed environment.

FIG. 2 is a processing flowchart of the computer in FIG. 1.

The program distributed arrangement unit 4 is executed and managed as one application of the O8 (operating system) of each computer.

First, when an operator command or the like is input, the program distributed placement unit 4 is activated (step 1).
). Next, a variable 1 representing which module placement is currently being determined is initialized (step 2). Next, the execution computer Kl of the j-th module is acquired from directory 3 (step 3). By referring to the directory 3, J (≧0) such that the first to i+jth computers are all execution computers Ki is acquired (step 4). That is, modules from module i to module (i+j) are successively transferred to computer Ki. Next, it is determined whether the computer Ki is the computer A (step 5). If it is computer A, there is no need to change the corresponding CALL portion of program 2, so the process jumps to step 10. Also, if the computer is not A, step 6
Proceed to. Here, from module i to module (i+
The structure of the variable list used by j) is acquired from the program and the transfer order is defined (step 6).

Up to this point, the processing has been performed by the program distribution arrangement unit 4.
Next, the communication program generation unit 6 creates a transfer program to the computer 1 (step 7). That is, an instruction to RECEIVE the value of the variable list l□(i, j) sent from computer A at the time of execution is added to the beginning of module (a). (b) At the end of module (i+j),
Set the value of variable list (i,,i) to 5EN for computer A.
Add the function of D.

Next, the program transfer unit 7 transfers the program to the computer 1 (step 8). That is, the following (a) (b)
) is transferred from computer A to computer K], and
] Create a program on (a) Transfer the variable list structure and transfer order. (1) Transfer module 1 to module (10J).

Next, the local computer program modification unit 5 modifies the program 2 of the computer A (step 9).

That is, the CAT, L module of program 2 (
REC to remove up to 1+, +) and take the value of the variable list for the corresponding program with 1 in H1' calculator.
E TVl et al command [1. And (10J)
It is determined whether the value of has become N, that is, whether all module placement has been completed. If YES, the process ends; if NO, i −+−, i + ] and returns to step 3 (step 1.0).

In this way, in this embodiment, by specifying the names of the computers placed in the module 4U in a directory outside the program, the computers are distributed and placed in the optimal i'' computers.
Since the program is executed, the program creator does not need to be aware of where each module is placed, and the amount of data communication can be reduced.

〔Effect of the invention〕

As explained above, according to the present invention, modules are placed on the optimal computer using a directory independent of the program, so the program creator can program without being aware of where each module is placed. At the same time, it is possible to design an efficient system that suppresses the amount of pigs transferred.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a computer in which programs are distributed and arranged, showing one embodiment of the present invention, and FIG. 2 is a processing flowchart of the computer in FIG. 1. IL 8: Computer, one program from 2 Niprogram module group, 3: Directory, 4 Niprogram program distribution section, 5: Program modification section in own computer, 6:
Communication program generation section, 7°9 Niprogram transfer section.

Claims (1)

[Claims] (1) In a multiprocessor system having a plurality of host computers connected via a communication line, at least one of the plurality of host computers is loaded with a subroutine program module, and on which computer the subroutine is loaded. , a directory indicating whether to execute the subroutine, and the computer executes the program according to the distributed arrangement of the program modules indicated in the directory.
A program distributed execution method characterized in that a communication processing program is automatically generated and each of the above subroutines is transferred to an execution computer specified by the above directory.