JP2507399B2 - Database equipment - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] According to the present invention, it takes a long time to create a database because the number of I / O accesses for creating a random record in a database apparatus is large. In order to solve the problem, an initialization space management means for storing the maximum value of the page numbers stored in the random record, a page number for storing a newly designated random code, and the initialization space management means By providing address number comparison means for comparing the size with the stored page number, it is possible to continuously initialize without accessing the page contents for sequential address access (leadless method). Enable the initialization processing function to reduce the number of I / O accesses required to initialize the page contents. It is obtained by realizing a greatly shortened creation time of the random record in the database.

[Industrial applications]

The present invention relates to a database device, and more particularly, to a database device capable of speeding up creation by reducing the number of initialization processes required when creating a random record.

[Conventional technology]

In the creation utility that creates the database, the page that stores the data that is the unit of input and output is first initialized by the initialization process, and then the process of writing records to the initialized page is performed. We will create a database. Next, a conventional technique of this creation utility will be described.

When the records to be stored are normal records that are stored in the order of the record numbers of the records, the database is created by sequential processing. Therefore, a database management system that manages the database (hereinafter referred to as DBMS)
That is, the database creation utility will be executed by only remembering how many page numbers have been initialized and by initializing the next page and accepting data input. That is, the initialization process is executed by the so-called leadless method without reading the page contents to the CPU side.

However, when these records are so-called random records that are stored in the page of the page number which is the address determined by the address conversion (randomizing process) of the record number, the storage address It is the user who uses the database, not the DBMS, to specify the page number of, so that the user may specify an arbitrary page number each time, so the CPU first checks all pages via I / O. Initialize the page. Then, when the page number that is the storage address is specified by the randomizing process, the contents of the page are read out via I / O,
By writing the record that is the stored data via I / O, the creation utility will be executed and the database will be created.

In addition, as another conventional technology of the random record creation utility, there is also a mode in which the CPU side is provided with a table having information on the presence or absence of initialization processing corresponding to all pages of the database. In this mode, when the storage page is specified by the randomizing process, the CPU uses this table to determine whether the page has been initialized, that is, whether the record after initialization has been stored. . If it indicates that it has been initialized, the CPU stores the record as it is via I / O, and if it indicates that the record has already been stored and is not initialized, then the I The database is created by initializing the page again via / O and then writing a record that is new storage data.

[Problems to be solved by the invention]

A conventional random record that creates a random record by checking the existence of a record by reading the contents of the page specified by the randomizing process each time via I / O, and determining the necessity of initialization. In the creation mode of, at least three I / O accesses, that is, for initialization of all pages executed first
I / O access.

I / O access to read the contents of the page specified by the randomizing process.

I / O access for writing to store records.

I needed at least 3 I / O accesses. The time required for this I / O access becomes considerable as the database capacity increases. Therefore, requiring at least three I / O accesses means that a considerable amount of absolute time is required to create the database, which reduces the efficiency of database operation. Had brought a point.

On the other hand, in the conventional random record creation method in which the CPU side is provided with a table having information on the presence / absence of initialization processing corresponding to all pages of the database, the presence / absence of initialization processing can be known using the table. Therefore, I / O access is performed at least twice, that is, for initialization of all pages executed first.
I / O access.

I / O access for writing to store records.

Although it can be reduced once compared to the previous conventional technology, the CPU for all pages must be provided, so the CPU
It has an essential drawback that it is not realistic because the storage capacity on the side is large and it is consumed.

The present invention has been made in view of such circumstances,
In creating the database, the minimum number of I / O access required when creating a random record in the database is one I / O access for writing to store the record. By making it possible to greatly reduce the time required and to make the storage capacity of the table required to realize this processing extremely small, a practical mode of creating random records can be realized. It is intended to be provided.

[Means for solving problems]

FIG. 1 is a principle configurational diagram showing a database device according to the present invention.

In the figure, 1 is randomizing processing means, 2 is address number comparison means, 3 is initialization space management means, 4 is storage area continuous initialization means, 5 is data storage means, 6 is storage area, 10 is auxiliary storage means Reference numeral 20 is an auxiliary storage means management means, and 30 is a user. Here, the auxiliary storage means management means 20 corresponds to a DBMS.

The storage area 6 is a storage area for storing random records designated by the user 30, and is configured on the auxiliary storage means 10 to form a database. The data storage means 5 is designated by the randomizing processing means 1 by the user 30. A random record is created by storing a record also designated by the user 30 in the storage area 6 of the address number obtained by performing the address conversion on the record number of the record to be stored.

The initialization space management means 3 has a function of storing the largest address number among the address numbers of the storage area 6 obtained by the address conversion executed by the randomizing processing means 1 up to then, and the address number comparison means 2 is A function of comparing the address number stored in the initialization space management means 3 with the address number in the storage area 6 of the next record newly obtained by the randomizing processing means 1 is provided.

The storage area continuous initialization means 4 is configured on the auxiliary storage means management means 20 which manages the auxiliary storage means 10 and is initialized by a leadless method without reading the contents of the storage area 6 for sequential address access. It has the function of executing processing.

[Action]

The address number of the storage area 6 for a new record is designated by the user 30 via the randomizing processing means 1,
When the address number comparison means 2 indicates that the new address number is larger than the address number stored by the initialization space management means 3, the storage area continuous initialization means 4
The storage areas 6 of all the address numbers from the address numbers stored by the initialization space management means 3 to the newly designated address numbers are continuously initialized without reading the contents of those storage areas 6, and The data storage means 5 stores the user 30 in the storage area 6 of the new address number.
Write the record specified by. Then, the maximum address number that has been initialized is described in the initialization space management means 3.

Further, the new address number is the initialization space management means 3
When it indicates that the new address number is smaller than the stored address number, the position of the new address number has already been initialized and writing is performed.

In this way, by providing the initialization space management means 3 and the address number comparison means 2, it is possible to initialize without reading the contents of the storage area 6 of the storage area continuous initialization means 4 of the auxiliary storage means management means 20. The initialization function can be used, and thus the number of I / O accesses required for creating a random record can be reduced, so that the creation time of the random record in the database can be greatly shortened. .

〔Example〕

An embodiment of the database device according to the present invention is shown in FIG. 2 and FIG. Here, FIG. 2 is an explanatory diagram for explaining the management content of the initialization space management means, which is one of the characteristic parts of the present invention, and FIG. 3 is executed by the present invention to create a random record. It is a flowchart to do.

In FIG. 2, 10 is an auxiliary storage device, which corresponds to the auxiliary storage means 10 in FIG. Reference numeral 6 is a page for storing data, which is a unit of input / output, and corresponds to the storage area 6 in FIG. A range 7 is composed of a plurality of pages 6 and divides a storage space on the auxiliary storage device 10.
In this explanatory diagram, the space on the auxiliary storage device 10 is shown to be composed of three ranges 7. Reference numeral 3 denotes an initialization space management means (table), which corresponds to the initialization space management means in FIG.

As shown in FIG. 2, the initialization space management table 3 is configured to exist for each range 7, and each initialization space management table 3 is a page included in each range that is initialized in the order of page numbers. By storing the largest initialized page number, the presence or absence of initialization of each page in each range is managed.

Next, the creation mode of the random record according to the present invention will be described with reference to the flowchart shown in FIG.

In step 20 following step 10, each record is input to the creation utility program one by one in order to create a random record. Subsequently, at step 30, the page number, which is the address of the page where the record is stored, is determined by performing randomizing processing, that is, address conversion on the input record number.
FIG. 5 shows the contents of this address conversion with a concrete example. As shown in FIG. 5, for example, record number 100
The page number determined by the address translation of 10 and the page number determined by the address translation of record number 200 are 5, and the page number that is the page where the record is stored is independent of the ascending order of the record numbers. The characteristics of random records are where they are decided. The processing in steps 20 and 30 corresponds to the processing executed by the randomizing processing means 1 in FIG.

When the page number is determined in step 30 (this page number is n ₂ ), this page number is determined in step 40.
n ₂ and the page number stored in the initialization space management table 3 corresponding to the range being processed (this page number and n ₁₎
Compare the size with. As described above, the page number n ₁ stored in the initialization space management table 3 is the largest page number that has already been initialized among the pages that are initialized in the order of page numbers. By comparing the sizes, it is possible to determine whether or not the page with the page number n ₂ determined in step 30 has already been initialized. The process in step 40 corresponds to the process executed by the address number comparing means 2 in FIG.

According to the magnitude comparison in step 40, when n ₂ > n ₁ ,
That is, when it is found that the page newly determined as the storage destination of the record has not been initialized yet, the process proceeds to step 50, from the page number stored in the initialization space management table 3 to the newly determined page number. the n ₁ + 1 to n ₂ is a page for every page numbers continuously initialized. The initialization processing in step 50 corresponds to the processing executed by the storage area continuous initialization means 4 in FIG. The continuous initialization process executed in step 50 can be executed by sequential access according to the order determined in the page number order, as in the normal record initialization process. Therefore, the leadless method, that is, the contents of the record is used. Can be executed on the DBMS side without reading the
It will be possible to initialize without O access.

In step 60 following step 50, since the page number n _{2 has} been initialized in step 50, the page number stored in the initialization space management table 3 is rewritten from n ₁ to n ₂ . By this processing, the initialized space management table 3 can store the largest initialized page number of the pages initialized in the order of page numbers. After the process of step 60, the process proceeds to step 70 which is a process for storing records.

On the other hand, when n ₂ ≦ n ₁ is determined by the size comparison in step 40, that is, when the page newly determined as the storage destination of the record has already been initialized, step 5
The process directly proceeds to step 70 which is a process for storing a record without going through 0 and 60.

In order to understand the processing of steps 40, 50, 60 explained above, how the example of record input concretely shown in FIGS. 4 and 5 is initialized in steps 40, 50, 60. Figure 6 shows a summary of how to proceed. As shown in the figure, the process is performed while managing the maximum address number that is being initialized.

Next, the step for executing the process for storing the record on the page of page number n ₂ determined in step 30
The functions of 70 will be described. This step 70 corresponds to the data storage means 5 in FIG. Step 40
When it is determined that n ₂ > n ₁ in step 70, step 70
In step 50, since the page of page number n ₂ has already been initialized, the process of immediately writing the record, which is the stored data, to the page of page number n ₂ via I / O is performed. On the other hand, when it is determined in step 40 that n ₂ ≦ n ₁ , step 70 reads the contents of the page with page number n ₂ to the CPU side via I / O, and stores the stored data via I / O. The process of writing the record is performed.

When the storage of records is completed in step 70, step 20
Then, the process for writing the next input record to the page of the specified page number is started. In this way, by storing all the input records in the page of page number n ₂ determined in step 30, a random record in the database will be created.

Among the processes executed by step 70, the process executed when n ₂ > n ₁ is one of the characteristic processes of the present invention because the number of times through I / O access is one, and n ₂ ≦ n _The processing performed in the case of ₁ requires at least two I / O accesses and is not a substitute for the conventional procedure for storing data. However, even if n ₂ ≤ n _1, the continuous initialization process is executed in step 50 without intervening I / O access, so it was always necessary for conventional random record creation. Since the process of initializing all pages via I / O first is not necessary, the number of I / O accesses can be reduced by one.

〔The invention's effect〕

According to the present invention, when creating a random record in a database, it is possible to use an initialization processing function of a database management system that can sequentially initialize without accessing the contents of a page for sequential address access. Therefore, 1 is the least
Random records can be stored in the page of the desired address only by performing I / O access once. Therefore, the number of I / O accesses can be reduced from the number of I / O accesses required when creating a conventional random record, and the creation time of a random record can be greatly shortened.

Moreover, the memory capacity of the initialization space management table 3 required for executing this creation mode is sufficient if the maximum address of the page number stored in each range can be specified. It is possible to realize with a very practical configuration without being taken for this reason.

[Brief description of drawings]

FIG. 1 is a principle configuration diagram for explaining the principle of the present invention, FIG. 2 is an explanatory diagram for explaining the management contents of the initialization space management means, FIG. 3 is a flow chart executed by the present invention, and FIG. An example of input data, FIG. 5 is a concrete example of contents of address conversion, and FIG. 6 is a diagram showing a concrete example of contents of initialization processing. In FIG. 1, 1 is randomizing processing means, 2 is address number comparison means, 3 is initialization space management means, 4 is storage area continuous initialization means, 5 is data storage means, 6 is storage area, and 10 is auxiliary. Storage means, 20 is auxiliary storage means management means, and 30 is a user.

Claims (1)

(57) [Claims] 1. Randomizing processing means (1) for specifying the data storage area (6) and the address numbers in the data storage area (6) for storing data in no particular order.
And a data storage means (5) for storing data in an area of an address number designated by the randomizing processing means (1), further comprising an address of an initialized area of the storage area (6) Initialization space management means for storing the largest number (3)
Of the address number stored in the initialization space management means (3) and the address number specified by the randomizing processing means (1), and the randomizing processing means (1). When the address number comparison means (2) indicates that the designated address number is larger than the address number stored by the initialization space management means (3), randomizing is performed from the address number stored by the initialization space management means (3). A database device comprising: a storage area continuous initialization means (4) for continuously initializing an area up to an address number designated by the processing means (1).