JPH0267622A - Disk input/output system - Google Patents

Abstract

PURPOSE:To uniformly distribute files to plural disks and to easily shorten and secure a response by having access to a file distributing table by an argument following a disk input/output request at the time of generating the request to have access to the disk concerned. CONSTITUTION:When an I/O request is outputted by a working program, a key value and file identifier in a file distribution table 8 are transferred to a fire distribution table access means 7 as arguments. The means 7 receives a disk identifier DSK 1 or the like corresponding to the arguments and a record storage means 6 inputs/outputs a disk corresponding to the disk identifier to/ from the file concerned. Since the file I/O request can be uniformly distributed to plural disks, a response can be easily shortened and secured in a large scale on-line system concentrating I/O requests into a specific file and attaching importance to a response.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention is a disk input/output method in an online system that employs a disk input/output method, particularly a general file organization method other than a database organization, and is equipped with a plurality of disks. Regarding.

[Conventional technology]

In large-scale online systems where response time is important, it is desirable to consciously and effectively distribute input/output requests to each disk.

In the case of database organization, such a request has traditionally been satisfied by the range function, but in the case of a general organization file that is not database organization, it is necessary to look at a part of the key and go to the data section for input/output. Therefore, the key part and the data part must be in the same file, and there is no function to divide input/output requests to disks (physical files).

[Problem to be solved by the invention]

The conventional disk input/output method described above does not have a function to divide input/output requests into physical files, so input/output requests are not leveled for each disk, so they are concentrated on a specific disk, increasing response time. There is a drawback.

In addition, database organization requires specialists compared to general organization files in terms of business system design and construction.
Furthermore, when changing the data structure, it takes a long time to rebuild and lacks flexibility, so it is not necessarily easy to use for designers and operators.

[Means to solve the problem]

The method of the present invention is a disk input/output method for an online system that adopts a general file organization method and is equipped with a plurality of disks, in which a disk identification is held in advance for each key so that files are uniformly distributed among the disks. a file distribution table to be stored, and a file distribution table access means for accessing the file distribution table according to an argument accompanying the disk input/output request and obtaining the corresponding disk identification when a disk input/output request is made from a business program; and record storage means for accessing a disc corresponding to the disc identification.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 shows a block diagram of one embodiment of the invention.

In the figure, DSKI to DSK5 are disks, 2 is a file,
Reference numeral 5 represents an input/output request from a business program that issues an input/output request, 6 represents a record storage means, 7 represents a file distribution table access means, and 8 represents a file distribution table.

According to this diagram, the inventory mask consists of five disks DSK1 to DSK5.
, the special mask is on three disks DSK1, DSK3 and DSK5, and the slip file is on two disks DSKI.
The two-digit number in the diagram indicates the key of the file.

FIG. 2 shows the contents of the file distribution table 8 necessary to perform file distribution as shown in FIG. From these drawings, it can be seen that the inventory mask, customer mask, and slip mask are distributed across multiple disks. The file distribution table 8 is created in accordance with the file distribution table creation rules according to instructions from the operator.

Furthermore, the record storage means 6 and the file distribution table access means 7 are configured as subroutines of the business program.

Now, when an input/output request is issued from a business program, the key value and file identification of the file distribution table 8 are passed to the file distribution table access means 7 as arguments. The file distribution table access means 7 receives the disk identification (DSK1, etc.) corresponding to this argument, and the record storage means 6 performs input/output to the corresponding file of the disk corresponding to the disk identification.

For example, from a business program to an inventory mask, key =
Assume that an input/output request of 01 occurs. The argument is passed to the file distribution table access means 7, and as a result of the table search, the disk identification EFNI is obtained, so the record storage means 6 inputs and outputs the inventory mask of the disk DSKI.

〔Effect of the invention〕

According to the present invention, file input/output requests can be balanced and distributed to multiple disks, so it is possible to ensure a short response time in a large-scale online system where input/output requests for a specific file are concentrated and where response is important. It becomes easier.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a diagram showing a specific example of a file distribution table in this embodiment. DSKI to DSK5...disk, 2...file, 5...input/output request, 6...record storage means, 7
...File distribution table access means, ...File distribution table.

Claims (1)

[Claims] In a disk input/output method for an online system that adopts a general file organization method and is equipped with a plurality of disks, a disk identification is held in advance for each key so that files are uniformly distributed among the disks. a file distribution table to be stored, and a file distribution table access means for accessing the file distribution table according to an argument accompanying the disk input/output request and obtaining the corresponding disk identification when a disk input/output request is made from a business program; 1. A disk input/output method comprising: record storage means for accessing a disk corresponding to a disk identification.