JPH0689219A - Data base system - Google Patents

Abstract

PURPOSE:To provide a data base system which speedily makes access to the data of a complicated structure where the information structure of the real world is directly reflected by means of an index. CONSTITUTION:A data base 1 where reference can be executed between data is provided. A file management part 2 adds the identifiers 6a-6n of data required for the access operation of a file at respective data and attributes 7a-7b as indexes 8a-8n, and holds the identifier 6i of data of the file and the attribute 7n of data in the file, correlating them with each other, as a new index 10 for directly retrieving the identifier 6i of data in a different file 4i where the value of the attribute 7n of data in the file 4n is set to be a retrieval key 9. A data base management part 3 operates the new index 10 so as to make access to the file when an inquiry for requesting the identifier of different data for the value of the attribute of data is given.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a database system for accessing data stored in a database, and more particularly to a database system for efficiently accessing data having a more complicated structure that directly reflects the information structure of the real world. Regarding

[0002]

2. Description of the Related Art A database system is an information management system that accesses data stored in a database and is essentially processed by a computer. Also, modern database systems access not only relatively simple and regular data, but also more complex data structures that directly reflect real world information structures. In order to efficiently access such a complicated data structure, file management that determines the page on the secondary storage to be actually accessed is important. In particular, in file management, the indexing technique of adding auxiliary access information to data has a great influence.

FIG. 11 is a block diagram of a conventional database system of this type. In FIG. 11, the database 1 is provided with a plurality of files (not shown), and each file stores data having a complicated structure.

The inquiry Q from the user for this data is first converted into a number of commands by the inquiry processor 11, and further the database management unit 3
Is converted into a file operation by. File management unit 2
Has information on the storage structure of the database 1 on the secondary storage, and determines the page to be accessed on the secondary storage required for the file operation based on this information.

The file management unit 2 uses an index as auxiliary access information for accessing the data corresponding to the inquiry Q in the data stored in the database 1 in order to speed up the data retrieval process. have.

When data in a file is searched by using this index, a method such as when using a book index is used. That is, instead of scanning every page of a book, you can see which page you should read by looking only at the word you are looking for in the index.

An index is a collection of a set (x, p), and a pointer to a record having a key x in a file is p. By looking at this index, it is possible to know where in the file the record having a certain key x is located, without having to access the entire file.

Further, in the conventional database system,
An index is generated for each file, and in the case of relatively simple and regular data, improvement in data processing speed can be expected.

[0009]

However, a query for data having a more complicated structure that directly reflects the information structure of the real world, which is often used recently, that is, data that cannot be searched without operating a plurality of indexes, is made. If there is, an index is provided for each file, so that there are many references between files. In other words, since there are many indexes to be operated, it takes a considerable amount of time to search for data having a complicated structure, and the data processing speed is reduced.

The present invention has been made in view of the above points, and an object thereof is to quickly access data having a complicated structure directly reflecting the information structure of the real world by an index. To provide a database system.

[0011]

In order to solve the above problems and achieve the object, the present invention has the following constitution. FIG. 1 shows the principle of the present invention. Data 5a to 5n having a plurality of files 4a to 4n and data in each of the files
A file management unit 2 having a database 1 that can be referred to between n and adding data identifiers 6a to 6n and attributes 7a to 7n necessary for file access operations as indexes 8a to 8n for each data of the database 1. And a database management unit 3 for accessing the files 4a to 4n by operating the indexes 8a to 8n added for each data by the file management unit 2 when a data inquiry is made. .

The file management unit 2 has a file 4
The attribute 6n of the data of another file is used as a new index 10 for directly searching the identifier 6i of the data of another file 4i using the value of the attribute 7n of the data of n as the search key 9. 7n are held in association with each other.

The database management unit 3 is configured to access the file by operating the new index 10 when the user makes an inquiry for the identifier of another data with respect to the value of the attribute of a certain data. To do.

More preferably, the following is preferable. That is, the file management unit 2 is configured to generate a new index 10 by using the files 4a to 4n when the inquiry frequency exceeds a predetermined frequency.

Further, the search key 9 may be a value of the attribute 5n of the last data 7n to be referred to, and the identifier of the other data may be the identifier 6a of the first data 5a to be referred to.

Further, the database 1 is an object-oriented database, a plurality of classes corresponding to the files are prepared, each class has an object identifier, each class has a plurality of attributes, and the attributes of the upper class are As a new index for constructing a composite object that refers to a subordinate class and using the value of an attribute of a class as a search key to directly search the object identifier of another class, the identifier of another class and the attribute of a class are used. To hold.

[0017]

With the above configuration, the file management unit 2 allows the identifier 6 of the data of another file 4i using the value of the attribute 7n of the data of a certain file 4n as the search key 9.
As the new index 10 for directly searching i, the identifier 6i of the data of another file and the attribute 7n of the data of a certain file are held in association with each other, and the database management unit 3 sets the value of the attribute 7n of the certain data. When there is an inquiry for another data identifier 6i, the attribute 7n of the data of the file of the new index 10
The identifier 6i can be immediately obtained from

Therefore, the number of file access operations can be reduced and the data processing speed can be improved. Further, since a new index is generated when the frequency of inquiries exceeds a predetermined frequency, data processing can be performed according to the actual situation.

Further, since the identifier 6a of the first data 5a to be referred to is immediately obtained from the value of the attribute 7n of the last data 5n to be referred to in the new index, the number of file accesses can be greatly reduced.

When the database is an object-oriented database, it operates as follows. Figure 2
Is a diagram for explaining the indexing of the object-oriented database according to the present invention, FIG. 3 is a diagram showing an index newly generated according to the present invention, and FIG. 4 is a diagram showing the contents of each index for each file by the conventional indexing. The object-oriented database uses predicates of classes, instances, and attributes corresponding to files, records, and fields.

In FIG. 2, for example, three class 2
1, 22 and 23 each class has three attributes 21a to 21
c, 22a to 22c, 23a to 23c, and each instance has an object identifier OID31 to 33.

In this case, "obtain the OID 31 of the compound object in which the value of the attribute 23c of the class 23 is j"
The system of the present invention is compared with the conventional system when there is an inquiry for "a".

First, in the present invention, as shown in FIG. 3, the attribute 23c and the index 41 of the OID 31 are prepared. Only the attribute 23c of the class 23 of the compound object and the OID 31 of the class 21 are held in one index 41. To answer the inquiry, only the single index 41 is operated and the value j of the attribute 23c is directly changed to O.
The ID a can be accessed.

On the other hand, in the conventional indexing shown in FIG. 4, the attribute 21c of the class 21 is different from the attribute 21c and OI.
The index 51 of D31 is prepared, the attribute 22c is prepared for the attribute 22c of the class 22, and the index 52 of the OID32 is prepared, and the attribute 23c is prepared for the attribute 23c of the class 23.
An index 53 of c and OID 33 is prepared.

To answer the inquiry, first, the index 53 is used to obtain the g of the OID 33 from the value j of the attribute 23c,
Next, at the index 52, from the value g of the attribute 22c to the OID3
Then, d of 2 is obtained, and finally, a of the OID 31 is obtained from the value d of the attribute 21c by the index 51.

As described above, the conventional system needs to operate the indexes of the classes constituting the composite object in response to the inquiry, but the system of the present invention, in the case of an inquiry for a complicated data structure, that is, between files. Even if there are many references, it is not necessary to operate the index for each reference-related file, and high performance can be expected by only operating a single index.

[0027]

EXAMPLES Specific examples of the present invention will be described below.
FIG. 5 is a configuration block diagram of an embodiment of the database system according to the present invention, and FIG. 6 is a flow chart showing an index generation procedure of the embodiment.

In FIG. 5, the database 1 has a plurality of files 4a to 4n and has data in each of the files, and has a hierarchical structure in which the data 5a to 5n can be referred to.

The file management unit 2 indexes the data identifiers 6a to 6n and the attributes 7a to 7n necessary for the file access operation for each data of the database 1 as an index 8a.
~ 8n is added.

The inquiry processor 11 converts an inquiry from the user about the data into several commands, and outputs these commands to the database management unit 3.

When a data inquiry is made, the database management unit 3 operates the files 8a to 8n by operating the indexes 8a to 8n added by the file management unit 2 for each data.

Further, the file management unit 2 uses the value of the attribute 7n of the data of a certain file 4n as a search key 9 and sets a different index 10 as a new index 10 for directly searching the data identifier 6i of another file 4i. The file data identifier 6i and the file data attribute 7n are held in association with each other, and a new index 10 is generated when the frequency of inquiries exceeds a predetermined frequency. .

The database management unit 3 operates the new index 10 to access the file when the user makes an inquiry for the identifier of another data with respect to the value of the attribute of a certain data. Has become.

The database system in the embodiment configured as described above is an object-oriented database, and this object-oriented database includes classes corresponding to files, records, and fields, respectively.
Predicates of instance and attribute are used. <Structure of Object-Oriented Database> FIG. 7 is a diagram showing the contents of each class in the embodiment, FIG. 8 is a diagram showing a newly generated displacement / vehicle index, and FIG. 9 is a contents of each conventional index for each class. FIG.

In FIG. 7, three classes of a car class 21, a drive system class 22 and an engine class 23 are prepared, and these classes 21 to 23 form a hierarchical structure to form one compound object. The automobile class 21 has, as attribute values, a drive 21a such as drive777, an average weight 21b such as 1.5t, and a manufacturer 21c such as N company. The drive system class 22 has an engine 22 such as eng111 as an attribute value.
It has a transmission 22b such as a and a manual. Engine class 23, displacement 23 such as 4000cc
It has cylinders 23b such as a and 8 cylinders.

Each instance has an OID, the automobile class 21 has an automobile OID 31 such as autoO000, the drive system class 22 has a drive system OID 32 such as drive000, and the engine class 23 has an engine OID 33 such as eng111.

The drive 21a of the automobile class 21 refers to the drive system class 22 and refers to the engine 22 of the drive system class 22.
The a refers to the engine class 23.
In addition, with respect to the reference classes having a relatively well-researched complex class structure among the vehicle class 21, the drive system class 22, and the engine class 23, the reference class is the vehicle OID31 at the beginning and the last reference class. Displacement 2
Only 3a and 3a are retained as the displacement / vehicle index 41 (shown in FIG. 8).

The indexing method in the embodiment can be implemented by using a commonly used indexing method such as B + tree. The non-leaf node describes the value of the attribute to be indexed and the pointer to the lower-level non-leaf node of the B + tree.

The leaf node stores the value of the indexed attribute, the number of OIDs corresponding to it, and its OID group. However, unlike the normal B + tree, the value of the attribute described in the non-leaf node and the O stored in the leaf node
ID belongs to another class. <Index Generation Procedure> Next, a procedure for generating the displacement / vehicle index 41 in the embodiment will be described with reference to FIG. First, in step 101, the relationship between the retrieved attribute A and the retrieved class C is obtained.
For example, the order is automobile, drive, engine, displacement.

Next, in step 102, it is judged whether or not there is a record that has not been examined in class C. If there is a record, i is set to 1 in step 103. In step 104, it is determined whether i is n. If i is n, in step 105, a record having the In value as the OID is searched. For example, the engine value eng888 is O
Search for the record [eng888,1500cc, 4] that you have as an ID.

Then, in step 106, the value of the attribute A and the OID of the record of the class C are held as the index 41. For example, the attribute displacement of 1500cc and class car O
Holds IDauto999 as an index. further,
After checking the next record in step 107, the process returns to step 102, and the processes after step 102 are repeated.

On the other hand, in step 104, when i is not n, a record having the value of Ii as OID is searched. For example, a record with the drive value drive777 as the OID [dr
ive777, eng888, automatic]. Step 10
In step 9, i is incremented by 1 and the process returns to step 104 to perform the processing of step 104.

On the other hand, if there is no unexamined record in class C in step 102, the held index is output in step 110 and the process ends. <Data Processing Using Index 41 of Embodiment> Next, the conventional index method and the index method of the embodiment will be described using a simple example.

Here, it is assumed that an inquiry for auto000 is made, for example, "obtain the vehicle OID31 of the compound object in which the displacement 23a of the engine class 23 is 1500cc."

For this inquiry, the indexing in the embodiment and the conventional indexing are compared. (1) First, in the indexing method according to the embodiment, as shown in FIG. 8, an exhaust volume 23a and an automobile OID 31 are prepared as an exhaust volume / vehicle index 41. Here, only the displacement 23a of the engine class 23 of the composite object and the automobile OID31 of the automobile class 21 are held as one index 41. To answer the inquiry, the value of the displacement 23a at index 41 1500cc
From this, it is possible to immediately obtain the auto000 of the automobile OID 31 simply by operating the single index 41. (2) On the other hand, in the conventional indexing method, as shown in FIG. 9, the drive 21a and the automobile OID31 are prepared as the index 51 for the attributes of the automobile class 21, and the engine 22a is provided as the index 52 for the attributes of the drive system class 22. , Drive system OID32 is prepared, and displacement 2 is set as index 53 for the attribute of engine class 23.
3a, engine OID33 is prepared.

In order to answer the inquiry, first, at index 53, from the value of displacement 23a 1500cc to engine OID3
Find eng888 of 3, then engine 2 at index 52
The drive777 of the drive system OID32 is obtained from the value eng888 of 2a, and finally the value drive777 of the drive 21a is driven by the index 51.
Find auto000 of car OID31 from.

In the conventional index method, inquiries for a complex object that are frequently searched require the operation of indexes for the number of classes forming the complex object. According to this embodiment, a single index is used. Data with complicated structure can be accessed only by operating.

Therefore, the number of file access operations can be reduced and the data processing speed can be improved. In the embodiment, the number of classes and attributes has been described as 3 classes and attributes 2 or 3, but other numbers may be used.

Furthermore, the index 41 may be generated when the number of inquiries exceeds a predetermined frequency. In this case, for example, the same inquiry may be counted by using a counter or the like, and the predetermined frequency may be set when a predetermined count is exceeded. By doing so, the frequently used data can be immediately searched, and the effect is great.

FIG. 10 shows the result of a simple comparison between the indexing method of the embodiment and the conventional indexing in the example of the composite object shown in FIG. In addition,
Here, each class consists of 600,000 instances, the OID length is 8B (bytes), and the indexed attribute length is 8B. In addition, the page size is 4KB
(Kilobytes) and the fanout of B + tree was set to 218, and the following three queries Q1 to Q3 were evaluated. (Q1) Compound object OID1 whose attribute value is j
Ask for (Q2) Attribute values are j1, j2, j3, j4, j5, j
Find the OID1 of a compound object that is 6, j7, j8, j9, j10. (Q3) Attribute values are j1, j2, j3, j4, j5, j
6, j7, j8, j9, j10, j11, j12, j1
3, j14, j15, j16, j17, j18, j1
OID1 of compound object that is either 9 or j20
Ask for

Inquiries Q1 to Q3 are shown in FIG.
Indicates the page access count of the index for. Since the page access count of the secondary storage is proportional to the access cost of the secondary storage, which is the largest element of the processing cost,
It can be seen that performance improvement can be obtained by using the indexing method of the embodiment.

[0052]

According to the present invention, the file management unit 2
Search key 9 for the value of attribute 7n of data in a certain file 4n
As a new index 10 for directly searching the data identifier 6i of another file 4i, the data management unit 3 holds the data identifier 6i of another file and the attribute 7n of the data of a certain file in association with each other.
When there is an inquiry for the identifier 6i of another data for the value of the attribute 7n of a certain data, can immediately obtain the identifier 6i from the attribute 7n of the data of the file of the new index 10.

Therefore, the number of file access operations can be reduced and the data processing speed can be improved.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention.

FIG. 2 is a diagram for explaining indexing of an object-oriented database according to the present invention.

FIG. 3 is a diagram showing an index newly generated in the present invention.

FIG. 4 is a diagram showing the contents of each index for each file by conventional indexing.

FIG. 5 is a configuration block diagram of a database system according to an embodiment.

FIG. 6 is a flowchart showing an index generation procedure according to the embodiment.

FIG. 7 is a diagram showing the contents of each class in the embodiment.

FIG. 8 is a diagram showing a newly generated displacement / vehicle index.

FIG. 9 is a diagram showing the contents of each index for each conventional class.

FIG. 10 is a diagram showing a comparison between the indexing of the embodiment and the conventional indexing.

FIG. 11 is a configuration block diagram of a conventional database system.

[Explanation of symbols]

 1 ... database 2 ... file management section 3 ... database management section 4a-4n ... file 5a-5n ... data 6a-6n ... identifier 7a-7n ... attribute 8a-8n ... index 9 ... search Key 10 ... New index 11 ... Inquiry processor 21 ... Automotive class 22 ... Drive class 23 ... Engine class 31 ... Automotive OID 32 ... Drive system OID 33 ... Engine OID 41 ... Displacement ... Car index

Claims (4)

[Claims] 1. A plurality of files (4a to 4n) and data (5a to 5n) each of which has data.
n), a database (1) that can be referred to, and data identifiers (6a to 6n) and attributes (7a to 7n) of data necessary for file access operation are indexed for each data of this database (1). Each file is operated by operating the file management unit (2) added as 8a to 8n) and the index (8a to 8n) added to each data by the file management unit (2) when a data inquiry is made. (4a-4
n), a database management unit (3) for accessing and operating the file management unit (2), wherein the file management unit (2) uses another value of the attribute (7n) of the data of a certain file (4n) as a search key (9). As a new index (10) for directly searching the data identifier (6i) of (4i), the data identifier (6i) of another file and the data attribute (7n) of a certain file are held in association with each other. The database management unit (3), when the user makes an inquiry for the identifier of another data for the value of the attribute of a certain data, the new index (10).
A database system characterized by accessing and manipulating files by operating. 2. The file management unit (2) uses the files (4a to 4n) to create a new index (1) when the frequency of the inquiry exceeds a predetermined frequency.
0) is generated, and the database system according to claim 1. 3. The search key (9) is a value of an attribute (7n) of the last data (5n) to be referenced, and the identifier of the other data is an identifier of the first data (5a) to be referenced. (6a), The database system according to claim 1 or 2. 4. The database (1) is an object-oriented database, wherein a plurality of classes corresponding to the file are prepared, each class has an object identifier, each class has a plurality of attributes, and is a superclass. As a new index for directly searching an object identifier of another class whose attribute constitutes a composite object in which an attribute refers to a subordinate class and the value of the attribute of one class is used as a search key, the identifier of another class and that of a certain class The database system according to claim 1, wherein the database system holds an attribute and.