JP3494545B2 - Distributed computer system, computer and distributed processing method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed computer system, its computer, and a distributed processing method.

[0002]

2. Description of the Related Art As the performance and cost of personal computers have decreased,
Networking is progressing and resources are effectively used by sharing data.

[0003]

However, in the conventional networked computer system, non-coordinated master-slave relational computers such as a master-slave computer pair and a server-client pair are connected. , Only data is shared,
It was not possible to share the program equally among multiple computers.

In view of the above problems, an object of the present invention is to allow a plurality of computers to share a program equally and to improve the program processing speed.
( EN) Provided are a distributed computer system, a computer therefor, and a distributed processing method capable of performing the following.

[0005]

Means and effects to an aspect of the present invention
In one aspect , in a distributed computer system in which a plurality of computers are connected to each other via a network, each of the plurality of computers has a program name, a location indicating a computer that holds the program, and a program. Depends on each computer
Program execution for unit input data under unit load
Required time and data of processing result for the unit input data
In response to an undecided program execution request, the program table is referred to, the computer holding the program is inquired about the load state, and the result is obtained. And the program
A network processing unit that determines a computer to execute the program based on the line duration and the data amount, and sends a determined program execution request including a message necessary for executing the program to the determined computer. And an operating system that executes an internal program in response to the determined program execution request and transmits the result to the network processing unit that is the source of the program execution request, and the network processing unit that has received the result Transfer the result to the internal operating system.

The program is a concept including one or more related programs and jobs. According to this distributed computer system, the programs are shared equally among a plurality of computers, the program utilization efficiency is improved, the functions and performance of each computer are substantially improved, and the program processing speed is improved. Play. The configuration, action and effect of another aspect of the present invention are as follows.
Will be clear from the explanation.

[0007]

[0008]

[0009]

[0010]

[0011]

[0012]

[0013]

DETAILED DESCRIPTION OF THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. In the distributed computer system 10 shown in FIG. 1A, computers 11 having an equal relationship
The networks 15 to 14 are connected to each other. Each of the computers 11 to 14 includes an external storage device and an input / output device. The network 15 includes a communication control device corresponding to each of the computers 11 to 14 and a communication line connecting these, and data is transmitted in a packet format, for example.

FIG. 1B shows the software configuration of the computer 11 by functional blocks. The computer 11 has m programs 21 to 2 in its external storage device.
The operating system 2 which has m and executes the programs 21 to 2m to enable time-division parallel processing.
It is managed at 0. Here, the program is a concept including a single or a plurality of related programs (for example, a main program and a plurality of subprograms) and a job, and an object-oriented type is preferable, and a program execution result is used as a data message in response to a message input. Output.

For example, by clicking an icon on the display screen with a mouse, a program execution request is issued to the network processing unit 30. The network processing unit 30 refers to the program table 31 and determines one of the computers 11 to 14 that should execute the program as described later. FIG. 2 shows a configuration example of the program table 31.

In this table, the program name, the location indicating the computer holding the program, the average execution time required when the program is executed for the unit input data under the load condition of the predetermined load level 1, The amount (byte count) of the processing result data message per unit input data and the latest update date and time of the program are registered. Locations C1 to C4 indicate the computers 11 to 14 of FIG. 1A, respectively.

For example, the program PROG1 is located in locations C1 and C2, that is, computers 11 and 1.
It is held in the second storage device, and the execution time required per unit input data in the computers 11 and 12 at load level 1 is 482 sec and 320 sec, respectively, and the message amount of the processing result per unit input data is 253 kB. The latest update date of the program PROG1 is January 18, 1997. Other programs P
The same applies to ROG2 to PROGn.

In FIG. 1, the software configuration of the computers 12 to 14 is the same as that of the computer 11. The network processing units in the computers 11 to 14 communicate with each other using the same protocol. Therefore, the types of operating systems of the computers 11 to 14 may be different from each other. The program tables in the computers 11 to 14 have the same contents as each other. For example, when the program 21 in the computer 11 is updated by version upgrade, the program table 3
The update date of the program PROG1 in 1 is also updated, and the update information is transferred from the network processing unit 30 to all the other computers 12 to 14 to ensure the identity of the contents of the program tables in the computers 11 to 14. To be done. The same program 21 is stored in the computer 11 and the computer 12, and the computer 1
When only one program 21 is updated, the program 21 of the computer 11 is treated as different from the program 21 of the computer 12. Instead of handling as a different program in this way, the updated program 21 may be transferred from the computer 11 to the computer 12 to update the program 21 in the computer 12.

FIG. 3 shows the processing from the program execution request to the network processing unit 30 generated by the computer 11 to the program execution request to any operating system of the computers 11-14. Hereinafter, the numbers in parentheses are step identification numbers in FIG. (50) Network processing unit 30 in the computer 11
On the other hand, when a program execution request is issued, the network processing unit 30 refers to the program table 31 and acquires the location of the program from the program name included in this request.

(51) Inquire of the computer having the program about the current load state. For example, when the computer 11 issues an execution request for the program PROG1 to the network processing unit 30, since the locations are C1 and C2, the operating system is transferred to the operating system via the operating system 20 of the computer 11 itself and the network processing unit in the computer 12. , Inquire about the load status.

(52) From the result of this inquiry, the required execution time and result data amount registered in the program table 31, and the input data amount included in the program execution request, the network processing unit 30 Decide which computer should run the program, that is, the computer that gives the fastest results. For example, when supplying M times the unit input data to the program PROG1, when the load level of the computer 11 is A, the required execution time is 482 AM seconds, and when the load level of the computer 12 is B, the required execution time is 320 BM seconds. Become.
When the program PROG1 is executed by the computer 11, the transfer time of the result is 0 seconds, but when it is executed by the computer 12, the data transfer time of 253 M / V seconds is added to the execution time 320 BM seconds. It Here, V is a data transfer rate [kB / sec]. From these times, determine the computer that gives the fastest results.

(53) A program execution request including an input message for the program is transmitted to the computer determined in step 52. For example, when the computer 12 determines to execute the program PROG1 in the above example, the network processing unit 30 of the computer 11 transmits a program execution request to the operating system of the computer 12 via the network processing unit of the computer 12.

On the computer 12, the program PROG1
When the execution result is obtained, the operating system of the computer 12 supplies the result to the network processing unit. The network processing unit of the computer 12 transfers this to the network processing unit 30 of the computer 11.
And the network processing unit 30 supplies it to the operating system 20.

For example, in the computer 11, when the program PROG1 and the program PROG2 are desired to be executed simultaneously at the same time as fast as possible, the program PROG2 is executed by itself, and the load level of the computer 11 is increased at this time. Cause the computer 12 to execute in parallel. If an execution request for the program PROGi, for example, a translation request is issued in the computer 11, the computer 11
Computer PROGi does not exist in computer 1
Since it exists only in 3, the computer 11 causes the computer 13 to execute and receives the result as described above.

Such processing is equally performed for each of the computers 11 to 14 according to the above configuration. As a result, each program in the computers 11 to 14 is shared and effectively used, and the function and performance of each computer are substantially improved.

[Brief description of drawings]

FIG. 1A is a block diagram showing a distributed computer system according to an embodiment of the present invention, and FIG. 1B is a functional block diagram in a computer.

FIG. 2 is a diagram showing a configuration example of a program table in FIG. 1 (B).

FIG. 3 is a flowchart showing processing from a program execution request to an internal network processing unit to a program execution request to an operating system.

[Explanation of symbols]

10 distributed computer system 11-14 Computer 15 network 20 operating system 21-2m program 30 Network processing unit 31 program table

Claims (5)

(57) [Claims] 1. In a distributed computer system in which a plurality of computers are connected to each other by a network, each of the plurality of computers has a program name, a location indicating a computer having the program, and a program. Each compilation
The unit input data when the unit load is applied by the computer.
Program execution time and processing for the unit input data
A program table that has been registered and the amount of data of management results correspond, in response to a program execution request pending, by referring to the program table, query the load status to the computer for carrying the program, the results and , The program
A network processing unit that determines a computer to execute the program based on the required execution time and the amount of data, and sends a determined program execution request including a message necessary for executing the program to the determined computer. And an operating system that executes an internal program in response to the determined program execution request and sends the result to the network processing unit that is the source of the program execution request, and the network processing unit that receives the result A distributed computer system, which transfers the result to an internal operating system. 2. When the internal program table is updated, the network processing unit notifies all the external network processing units of the update, and the network processing unit that has received the notification responds to the internal distributed computer system of claim 1 Symbol placement and updates the program table. 3. A computer constituting the distributed computer system according to claim 1 . 4. A distributed computer system in which a plurality of computers are connected to each other by a network is used, and each of the plurality of computers indicates a program name and a computer having a program represented by the program name. Location and each of the programs
It corresponds to the unit input data when the unit load by the computer.
Program execution time and the unit input data
The program table registered in association with the data amount of the processing result is held, and in response to an undecided program execution request, the program table is referred to and the load state is inquired to the computer holding the program. , The result and the program
A computer to execute the program is determined based on the required execution time and the data amount, and a determined program execution request including a message necessary for executing the program is transmitted to the determined computer, and the determination is performed. A distributed processing method characterized in that an internal program is executed in response to a completed program execution request, and the result is transmitted to a computer which is a program execution request source. When 5. A program table for one of said plurality of computers is updated and, correspondingly to update the program table for all remaining computers, according to claim 4, wherein the Distributed processing method.