JPS63189934A - Sub-entry processing system for data base - Google Patents

Abstract

PURPOSE:To improve processing efficiency based on sub-entry and maintenance efficiency by determining the storing position of an index concerned in an index storing file by key value randomizing processing. CONSTITUTION:An address (block number) to store an index 15 in the index storing file 14 is determined by a key value based on operation by the randomizing processing part 11. Thereby, block access to the file 14 is executed usually only once in the retrieving or updating processing of an index. Since key values are randomized, the number of indexes in each block is averaged. Thus, the processing efficiency of sub-entry and the maintenance efficiency can be improved.

Description

[Detailed Description of the Invention] (Summary) In a database side entry processing method that searches a database for a corresponding record using the contents of data items included in the record as a key.

When storing and searching an index that combines key values and record storage addresses to realize secondary entries,
The structure is such that the storage position of the corresponding index in the index storage file is determined by key value randomization processing, and the processing efficiency and maintenance efficiency in database management are improved.

[Industrial application field]

The present invention is a database side entry processing method.

In particular, 1. For example, a database side entry processing method for a network type database for online transactions that requires high processing power.In online transaction processing, ensuring high processing power is an essential issue.

For this reason, when selecting various database entry methods, it is necessary to adopt a main entry with high processing performance, which results in a lot of effort being spent on database design.

If a so-called secondary entry method can be realized with processing power close to that of the main entry method, it will be beneficial for users.
It becomes possible to significantly expand the scope of application of the database. Therefore, there is a need for a secondary entry processing method that has higher processing performance than conventional secondary entry processing methods.

[Conventional technology] FIG. 3 shows an example of the conventional method. In fig.

Reference numeral 16 represents a network type database, 20 represents an index storage file, and R represents a record.

There is a so-called secondary entry method as a method for directly obtaining records in one database using the contents of data items as keys. In the secondary entry method.

It manages an index based on a pair of a key value and storage location information of a record having that key value, and uses the index to immediately access a desired record.

In the conventional method, the index is sorted in ascending order of key values (alphabetically), and 1
For example, as shown in FIG. 3, management was performed using a tree-structured index.

Note that in database management systems that do not use secondary entries, a method is sometimes used in which the storage location of the record itself in the database is determined by a predetermined hashing process. It is not possible.

[Problem that the invention seeks to solve]

In order to search an index realized by a tree-structured index as shown in FIG. 3, it is necessary to access blocks corresponding to nine levels in the index storage file 20 from the upper level to the lower level. Furthermore, when a record R in the database 16 corresponding to the index is added or deleted, or when the contents of a data item are changed, it becomes necessary to maintain the index, and the tree structure changes dynamically. . In particular, there is a problem in that a large processing cost is required when the hierarchy of a tree-structured index is changed.

The present invention aims to solve the above-mentioned problems, and aims to improve processing efficiency and maintenance efficiency using side entries without changing the structure of the existing database main body.

[Means for solving problems]

FIG. 1 shows an example of the basic configuration of the present invention.

In FIG. 1, 10 is a processing device consisting of a CPU and a memory, 11 is a randomization processing unit that randomizes key values using a predetermined function, 12 is an index storage processing unit that performs processing to store an index based on the key value, and 13 1 is an index search processing unit that performs processing to search an index based on a key value;
4 is an index storage file in which an index is stored, 15 is an index that is a pair of a key value and a record storage address, 16 is a database, Bl, B2 . ... Bn represents a block serving as an access unit in the index storage file 14, and R represents a record.

The index storage file 14 has n blocks 81 to Bn, and each of these blocks B1 to Bn includes a control unit,
An index 15 corresponding to each block is stored. Index 1
5 consists of a key value and a storage address of the record having that key value.

For the given key value, the randomization processing unit 11
A predetermined function is applied to randomize the key value within a range corresponding to the number of blocks in the index storage file 14.

When a record is added or deleted in the database 16, or when there is a change in the contents of a data item that is the subject of a secondary entry, the index storage processing unit 12 stores the key value that is the contents of the item. Specify and call the randomization processing unit 11.

A block in the index storage file 14 in which the index 15 is stored is determined.

The index search processing unit 13 specifies the key value of the record to be searched, calls the randomization processing unit 11, and obtains the record storage address of the record to be searched from the block of the index storage file 14 corresponding to the randomization result.

Note that the contents of multiple data items are used as key values.

Randomization is also possible.

[Effect]

According to the present invention, the address (block number) at which the index 15 is to be stored in the index storage file 14 is determined from the key value by the calculation performed by the randomization processing unit 11 . Therefore, in index search and index update processing, the block access to the index storage file 14 usually only needs to be made once. Since the key values are randomized, the number of indexes in each block is averaged.

By simply changing the structure of the index storage file 14 and the parts related to access to the index 15, the existing database 16 itself can be used as is, and various application processing units that use the database 16 can be used. etc., an index structure using a conventionally used tree-structured index and an index structure according to the present invention, which do not require any changes.

If the user is allowed to make a selection depending on the characteristics of the key data item, further flexibility will be achieved.

〔Example〕

FIG. 2 shows an embodiment of the invention.

The symbols in FIG. 2 correspond to those shown in FIG.

B50 represents the 50th block in the index storage file 14.

Each block in the index storage file 14.

The control section consists of a control section and an array of indexes 15.

In this example, it has the number of indexes included in the block, the start address of free space in the block, overflow and other control information.

Block B50 has two indexes whose key value is "Tokyo", and these and record storage addresses.

They are addresses 1000 and 1005, respectively.

In the case of searching or updating the index 15 with the key value = "Tokyo", the randomization processing unit 11 is called by specifying the key value - "Tokyo".

The randomization processing unit 11 uses a predetermined hashing function to execute an operation using "Tokyo" as a parameter, and notifies the request source of the randomization result BN.

The randomization result BN corresponds to the block number in the index storage file 14, and if BN is 50, the 50th block B5Q in the index storage file 14 is set as the index storage location.

For example, when adding a new record, block B
The head address of the free area is obtained from the control unit 50, and "Tokyo" is stored there as a key value, as well as the storage address of the new record. Also, the number information in the control unit.

Update free space information, etc. If there is no free space in the block 50 and a new index cannot be stored, the new index is stored in the primary block or a block prepared for overflow, and overflow control information is set in the control unit as necessary. .

For finding records by specifying a key value.

The index 15 stored in the block B50 of the block number notified from the randomization processing unit 11 is sequentially read out, and the corresponding record storage address is extracted. In this example, records at addresses 1000 and 1005 in the database 16 are the desired records.

〔Effect of the invention〕

As explained above, according to the present invention, most of the processing costs related to index searches and updates are absorbed by key value randomization processing, making it possible to significantly reduce overall processing costs.

It can demonstrate high processing capacity. At the same time, there is no need to maintain and manage traditional tree-structured indexes.
Processing is simplified, and application to existing databases can be realized relatively easily.

[Brief explanation of the drawing]

FIG. 1 shows an example of the basic configuration of the present invention, FIG. 2 shows an embodiment of the invention, and FIG. 3 shows an example of a conventional system. In the figure, 10 is a processing device, 11 is a randomization processing section, 1
2 represents an index storage processing section, 1'3 represents an index search processing section, 14 represents an index storage file, 15 represents an index, and 16 represents a database.

Claims (1)

[Claims] In a database sub-entry processing method in which a corresponding record is searched from a database (16) using the content of a data item included in a record as a key, a key value and a storage position of a record having the key value are provided. An index storage file (14) in which an index (15) consisting of information that corresponds to information is stored together for each block serving as an access unit; a randomization processing means (11) for randomizing within a range according to the number of blocks of the index storage file (14); and a randomization processing means (11) for specifying a key value of a record to be stored in the database (16). , and specify the index storage processing means (12) that stores the index of the record in the block of the index storage file (14) corresponding to the randomization result and the key value of the record to be searched, and perform the randomization process. Means (11)
and an index search processing means (13) for obtaining storage position information of a record to be searched from a block of the index storage file (14) corresponding to the randomization result. method.