JPS62262154A - Data file device - Google Patents

Abstract

PURPOSE:To increase a processing speed by executing an access to a record by the use of a record number which signifies a physical record position. CONSTITUTION:A constitution H in each record of an indexed order organization file E has a KEY being an index use record name, and records corresponding to each KEY of a relative organization file F and G having various data actually. In case the constitution of the files F, G has been varied, matching of the record No. corresponding to the KEY is always taken together in an application function side, and a correct record No. data is always held. In case of generating a time series data, the record of the file G is fetched successively, and the latest data is read out of the file F by using the record No. concerned on the file F, which has been stored. Subsequently, the time series data into which the latest value has been integrated is stored in the original position of the file G.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Field of Industrial Application) The present invention relates to a data file device with increased data retrieval/writing speed.

(Prior Art) As one of the basic software of a computer system, there is a data management mechanism usually called a file management system. This file management system organizes and manages data based on certain rules.

Retrieving data from the AppCage 1 software
Upon receiving a write (hereinafter referred to as data access) request, the data is retrieved from the storage location and the application cage 1
We provide a service for providing and writing to software.

There are usually several types of data organization methods allowed in file management systems, and the following are typical ones:

ω Sequential File A file is generally made up of regular repetitions of a group of data called a record, which is a unit of input/output in a program, as shown in FIG.

A sequential file can be accessed sequentially in the forward or reverse direction of the record number (hereinafter referred to as record number) & after specifying the access start position it (given by record number) as an access method to this structure. Refers to the file type.

(2) Relative file The structure of the file is the same as the sequential file shown in FIG. 7, but the access method is a file type in which records can be accessed at arbitrary positions by randomly specifying records.

■ Indexed sequential file (indexed sequential file) Like the files in ■ and ■ above, records are not accessed by physical position specification (record list specification), but by logical record name (hereinafter referred to as KE).
This is a file type that allows random access to records using the filename (Y).However, in order to input/output a record, it is necessary to convert the logical record name KEY to a record movement of the physical position. This process is performed by a file management system. Various conversion methods have been considered, but a typical example is B-tre, which is often used recently.
A method with a hierarchical structure called the e method will be explained. 8th
Each layer except the data section of the lowest layer shown in rA is collectively called the index section, and the index block C of each layer is expanded into a tree structure from the root index block B, which always exists in the top layer. Ru.

The depth of the layer is managed within an appropriate range for each file management system, and usually depends on the maximum number of records n. All index blocks, except for the layer directly above the data section, have information about where and how many keys each index block in the next layer manages, and when a certain key is given, It is possible to determine which block in the primary layer is managing the data.

Also, unlike other index blocks, the index block immediately above the data section is
We have a record for EY.

When this layer is reached, the record island of the data section for the corresponding KEY is known. Therefore, when an access request by KEY occurs, the search starts from the root index block, and the search is performed sequentially.

After going through each stage, the data section is finally accessed by record Nα.

There is also a management method in which the lowest layer of the index block is used as the data section, but the idea is not much different.

The feature of this indexed sequential file is that the application program does not need to perform access by specifying a physical position (specifying record selection) as in the file organization methods described in ① and ②.A logical key can be used, so the application program can be completed. After that, even if the records in the file change for some reason, if the KEY of the record required by the application program remains the same, there is no need to make any modifications on the application program side, which is good maintainability. .

However, this indexed sequential file is
Since it requires the work of converting Y into a physical record storm, it has the disadvantage of slow access speed. The disadvantage of this is that if speed is required, especially if the file has a record of the order Nobuko, and
If all the records had to be processed within a short time, it would be difficult to complete the processing within the specified time, and despite the ease of maintenance, the system was often not used.

(Problem to be solved by the invention) In other words, the problem to be solved by the invention is that a file structure with good maintainability slows down access speed, and a file structure with fast access speed reduces maintainability. It is to be.

Therefore, an object of the present invention is to provide a data file device that has high access speed and is easy to maintain.

[Structure of the invention]

(Means for Solving the Problems) The data file device of the present invention has a plurality of records, each of which is a group of data serving as an input/output unit in a program, and accesses to these records are controlled by the physical location of the record. The first and second files are accessed using the record numbers shown in the table, and the third file has a plurality of the records and accesses the records using a logical record name set for each record. It is equipped with

In order to exchange data between corresponding records of the first and second files, each record of the third file is configured to exchange data of the first and second files. Record numbers corresponding to each other are stored.

Furthermore, each record in the first file stores the record name and data of the corresponding record in the third file that stores the record number, and the record number of the corresponding record in the second file. . Furthermore, the second
Each record in the file stores the record name and data of the corresponding record in the third file that stores the record number.

Further, as these control functions, means for sequentially extracting each record of the first file according to its record number;
Based on the record number stored in the retrieved record, a means for retrieving the corresponding record from the second file is used to determine whether or not the record name stored in each of the retrieved records matches, and if they match, these records are If the means for exchanging data between records and the record name stored in each retrieved record do not match, the third file is accessed using the record name stored in one record. means for extracting the corresponding record and storing the record number of the second file stored in the record in the record of the first file as the record number of the other party.

(Function) In the present invention, access for normal data exchange is performed by record number, keeping the access speed fast, and adding and changing items.

When maintenance is required, such as when records are misaligned due to deletion, etc., a file indexed by logical record names is used to maintain record numbers, thereby facilitating maintenance.

(Example) An example of the present invention will be described below with reference to the drawings. Third
In the figure, the central processing unit (CPU) of the computer is connected to the main memory and auxiliary external memory through paths. Application programs operate on the main memory, and the data necessary for processing is input/output from data files usually stored in auxiliary external memory.

Also, functionally, as shown in Figure 4, data files and
A file management system exists as a part of the operating system or as a support function between the operating system and the application programs, and provides services for the various file organization methods described above.

FIG. 1 shows an example of the file structure of the present invention.

The internal structure H of each record of the indexed sequential file E is the record name KEY for indexing and the relative organization files F and G that actually have various data.

I have record rods that correspond to each of the above keys.

If there is a change in the configuration of file F or G, the application function always matches the records corresponding to KEY to always maintain correct record data. Here, the above relative organization file G is the first file, F is the second file, and the indexed sequential organization file E is the third file. ”.

Each record component of the second file F in the relative organization has the KEY of the third file E and data regarding the KEY (in this example, the current process value indicated by the KEY). . Hereinafter, it will be referred to as the "current value file."

The structure J in each record of the second file G with relative organization is as follows:
KEY of the third file E and data related to the KEY (in this example, 1 of the process value indicated by the KEY)
minute-by-minute time series data). Hereinafter, it will be referred to as "r-minute unit time series file".

The maximum number of records n, number, and α of each file are different. In other words, there are cases where data that exists in file F does not exist in the file. Furthermore, the order of records in each file does not match each other.

The file configuration example shown in FIG. 5 is an example for realizing the following functions.

In a system that monitors processes and records the data in time series, time series data is created every minute as follows. First, every minute, each record of the minute time series file G, which is the first file, is sequentially retrieved, and the corresponding record in the current value file F, which is the second file, is read to obtain the current value, and the time series After importing ffi as the latest value, it is stored again in the original position of the minute-by-minute time series file. The operation based on this file configuration example for realizing this function is shown in the flowchart of FIG.

The normal operation is as shown by the thick line route.

Steps to sequentially retrieve records of minute file G■
), reads the latest data from the current value file F using the corresponding record island on the stored current value file F (
Step ■■), check whether this value is far from the mutual key
After confirming by cross-checking (step ■■), store the time series data incorporating the latest value to the original position of the minute-by-minute time series file (step ■), and repeat this for all records. (Step ■), what should be noted.

In this case, for the minute-by-minute time series file G and the current value file F. All file accesses are only accesses to relatively organized files using the recorder, which greatly improves processing speed.

There are cases in which it is necessary to access the KEY-record correspondence file (third file) E, which is an indexed sequential file, and these are the following two cases.

■ Newly added items to minute time series file G (
record), and record Na on the corresponding current value file F has not yet been stored.

■ The record order of the current value file F may be affected by some reason (
(Adding, changing, deleting items, etc.), the stored record numbers become obsolete.

For (2), use the record island storage presence/fA check (step (2)) on the 2131st flowchart.

Also, for ■, it is determined by the same <, KEY verification check (step ■). In either case, read the KEY-record demon correspondence file E and automatically correct the correct record blood in step 0 @ ), the corresponding record on the current value file F is read using the corrected record (step Oo).

In either case, the KEY-record island corresponding file E, which is an indexed JIIIiIg file, is read only once, and from the first time onwards, normal route travel with only record access is possible.

For comparison, the flowchart in Figure 6 shows the operation when both the current value file and the minute-by-minute time series file in the above process are configured as indexed sequential files.
In this case, the processing method is simple, but all file accesses are to indexed sequential files using KEY, and the processing speed is slow.

As mentioned above, if both the current value file and the minute-by-minute time series file are configured as indexed sequential files (a
1) The present invention separates the indexing order and organization file function section as a KEY-record corresponding file E, and creates a current value file F.

The improvement in processing speed when both minute-by-minute time-series files G are relative organization files (referred to as b) is calculated based on several assumptions.As in this embodiment, the processing speed is almost negligible in filing processing. It is determined by the number of input/outputs (this is called transfer) between the main memory and the auxiliary storage device, and the processing time of other processing units is almost negligible compared to the time required for transfer.

In general file management systems, this transfer is performed in units of several records called blocks.

We make the following assumptions.

(i) The indexed sequential file is in B tree format, and the index section has a three-layer structure.

(...) For indexed sequential files and relative files, all blocks have a length of 256 bytes, and the record storage rate in each block is, in principle, 100%.

(■) The length of each record is as follows.

1st, 211m index section... 1st (item 3rd)
. . . 12) Once the index block of the indexed sequential file is brought to the main memory, it remains in the main memory until the end of processing.

, (Ma) Hit rate of data part of indexed sequential file (
The rate at which the record to be accessed this time is included in the block transferred to the main memory last time is -50%.

In other words, block transfer occurs once every two times in the data section.

(vi) The hit rate is also 50 when reading the current value file as a relative organization in the data arrangement of the minute-by-minute time series file.
%.

Based on the above assumptions, when the total number of data points is 1024 (all files are 1024 records), the number of blocks in each part of the indexed sequential file is as shown in Figure 5, and this value is This is a reasonable value.

In this case, the number of transfers using the file configuration method in case a is as follows.

Root index block transfer lX2 = 2 times
(For both files) 2nd layer index block transfer 6 x 2 = 1211 (for both files) 3rd layer index block transfer 86 x 2 = 172 times (for both files) Data section transfer r minute unit time series file
256 times (current value file section x 172 times total)
614 times On the other hand, the number of transfers in case b is as follows.

Current value file 86X2=l) 2 times total
As a result of this 428 times, both of the above become as follows.

As is clear from the above equation, the number of transfers can be significantly reduced, that is, the processing speed can be improved.

In addition, in configuration b, both the current value file and the fractional time series file can be logically accessed by KEY via the KEY-record correspondence file, which maintains maintainability. The processing speed has been improved without changing the settings.

In this way, the index section by KEY is made into an independent indexed sequential file as one file, the data section is configured as a relative organization file containing KEY in the record, and the record h of the relative organization file to be referenced is Store as file data. If you have a file structure like this.

Logical access using KEY allows access to a relative organization file containing desired data via an indexed sequential organization file as a KEY-Record Island compatible file, and processing speed is required while maintaining maintainability. In terms of functionality, processing speed can be ensured only by accessing record islands, and the benefits of indexed sequential files can also be applied to functional sections that traditionally sacrificed maintainability due to slow processing speed. becomes possible.

〔Effect of the invention〕

As described above, according to the present invention, processing speed is increased by accessing records using the record number, which is the physical record position, and even if maintenance is required, data can be accessed using the logical name of the data. Since the files are used independently, this does not impair maintainability. In other words, it is possible to improve maintainability and processing speed at the same time, which was difficult in the past.

[Brief explanation of drawings]

FIG. 1 is a file configuration diagram showing one embodiment of a data file device according to the present invention, FIG. 2 is a flowchart explaining the operation of the device shown in FIG. 1, and FIGS. 3 and 4 are diagrams for carrying out the present invention. Block diagram showing the outline of the device configuration, No. 5
The figure is a diagram for explaining the relationship between the number of blocks in an indexed sequential file. Fig. 6 is a flowchart of a conventional operation example shown for comparison with the device of the present invention, Fig. 7 is a diagram showing general sequential organization and relative organization file structures, and Fig. 8 is a general indexed sequential organization. FIG. 3 is a diagram showing the structure of a file. E... indexed sequential file, F, G... relative file, H, I, J... record content of each file. Agent Patent Attorney Nori Ken Yudo Hirofumi MitsumataFigure 1Figure 2Figure 4Figure 5Figure 6Figure 7Figure 8

Claims (1)

[Claims] It has a plurality of records that are a group of data that are input/output units in a program, and access to these records is as follows:
It has first and second files that are accessed using record numbers that represent the physical locations of records, and a plurality of records, and access to these records is performed using logical record names set for each record. In order to exchange data between records corresponding to each other in the first and second files, each record of the third file includes a third file containing the first and second files. Record numbers corresponding to each other for sending and receiving data of the files are stored, and each record of the first file contains the record name and data of the corresponding record of the third file that stores the record number, The record number of the corresponding record of the second file is stored, and each record of the second file stores the record name and data of the corresponding record of the third file that stores the record number. , these control functions include means for sequentially retrieving each record of the first file according to its record number, and means for retrieving the corresponding record of the second file according to the record number stored in the retrieved record. a means for determining whether record names stored in each of the retrieved records match and, if they match, transmitting and receiving data between these records; and a record name stored in each of the retrieved records. If they do not match, the third file is accessed using the record name stored in one record, the corresponding record is retrieved, and the record number of the second file stored in that record is assigned to the first file. A data file device comprising means for storing a record number of the other party in a record of a file.