JPH09223159A - Database access system for composite object set - Google Patents

Abstract

PROBLEM TO BE SOLVED: To effectively utilize a main storage area and to speed up inquiring processing for a composite object set. SOLUTION: An window area 108 is an optional area in a main storage area for acquiring composite data necessary for the judgement of an inquiring processing condition to a database 106. A composite data acquiring part 102 constructs composite data in the window area 108, and at the time of construction, preferentially accesses not data existing only in a secondary storage area but data existing also in the main storage area based upon the inspection of an object identifier(OID). At the time of tracing objects, the transition of the object is executed in each page by the use of page identifier(PID) obtained from the OID of the object. A condition judging processing part 105 judges the condition of the composite data in the window area 108 based upon an instruction from the acquiring part 102.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a query processing method of an object-oriented database management system,
To effectively access the data existing in the database on the secondary storage area by effectively using the main storage area when acquiring the composite data necessary for determining the query condition for the complex object set that is complicatedly associated The present invention relates to a database access method (database access method for a complex object set) for the purpose of speeding up the query processing of.

[0002]

2. Description of the Related Art A conventional database access method for a complex object set is "Proceedi".
ngs of ACM SIGMOD'90 pp. 3
00-311 "," A Perform "
ance Evaluation of Pointe
Pointer-ba of "r-based Joins"
sed Join method and “Proceedings
of ACM SIGMOD'91 pp. 148-1
57 ”,“ EfficientAsse ”
An assembly method of mbly of Complex Objects ”has been proposed.

Pointer-based Hyb, which is one of the Pointer-based Join methods
In the rid Hash Join method, when accessing an object set having a part-of relationship from an object to be searched, the object identifier (O
ID. Page ID (PID. Page IDentifier) obtained from Object IDentifier
er) hash value is acquired, all OIDs are expanded on the main storage area (secondary storage area if overflowed), and the OIDs are accessed for each divided partition (bucket). Optimization has been done.

In addition, Pointer-based Jo
Pointer-base which is another method of the in method
In the d Sort-Merge Join method, all PIDs obtained from the OIDs of the transition destination objects are acquired, and the OIDs are sorted by sorting them, so that physical pages in the database can be accessed in PID order when the objects are acquired. Becomes

On the other hand, in the Assembly method, an area called a window area for expanding composite data required for inquiry processing is secured in the main storage area, and the number of composite data that can be expanded in the window area is physical disk. It is assembled while access is scheduled. This makes it possible to optimize disk access.

[0006]

The above-mentioned conventional database access method for a complex object set has the following problems.

The above-mentioned Pointer-based Jo
In the in method, the objects having the reference relationship are traced to all the objects one by one, so when the complex objects are arranged in the vicinity by clustering, the effect that the method originally aims to achieve (access efficiency There is a drawback that the effect of improvement) cannot be obtained.

Further, in the above-mentioned Assembly method,
Since the performance is greatly affected by the disk access scheduling method, physical disk access scheduling has to be performed, so that there is a drawback that it cannot be implemented on a file system of a general-purpose operating system.

Further, both of the above-mentioned formulas have a drawback that only the data existing in the secondary storage area is taken into consideration and the data existing in the main storage area is not taken into consideration.

In view of the above points, an object of the present invention is to integrate both of the above-mentioned methods in order to alleviate the above-mentioned drawbacks of the conventional method (the Point-based method based on the Assembly method).
By using the database access method in the ter-based join method), the main storage area can be effectively used, and the database access method that can speed up the query processing of the complex object set (complex object set) Database access method).

[0011]

A database access method for a complex object set according to the present invention is an object-oriented database management system for performing a query process when performing a database access process for a complex complex object set that is complicatedly associated. A window area, which is an arbitrary area on the main storage area for acquiring composite data required for condition determination, and composite data are assembled in the window area, and secondary storage is performed based on an OID inspection in the assembly. Data which exists not only in the area but also in the main storage area is preferentially accessed, and when the object is traced, the transition of the object for each page can be performed using the PID obtained from the OID of the object. The composite data acquisition unit Based on the instruction from the data acquisition unit,
And a condition determination processing unit that determines the condition of the composite data in the window area.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of a database access method for a complex object set according to the present invention. The database access method for the complex object set of this embodiment is database 1
Data page access unit 101 for accessing page by page from 06, composite data acquisition unit 102 for acquiring composite data required for executing query processing from within those pages (pages accessed from database 106), and transition of objects. Object identifier (O
Hash processing unit 103 that performs hashing of the page number (PID) obtained from ID), sort processing unit 104 that sorts OIDs, and condition determination processing unit that performs query condition determination while referring to attribute information of composite data. And 105.

The main storage area used in the database access method for the composite object set according to the present embodiment includes an object cache area 107 for caching physical pages in the database 106 on the main storage area, and composite data. A window area 108 for temporary storage and an operation area 109 for hash processing and sort processing (hash table and sort area)
And an attribute information area 110 for registering attribute information of composite data.

FIG. 2 is a diagram showing an example of a class configuration in the database 106. This class configuration is an example of the class configuration in the company information database when the database 106 is the company information database.

FIG. 3 shows a query (O) in the database access method for the complex object set of this embodiment.
QL (Object Query Language)
FIG. 4 is a diagram showing an example of an inquiry described in (1).

FIG. 4 is a diagram showing attribute information of the inquiry shown in FIG.

FIG. 5 is a flow chart showing the processing of the database access method for the complex object set of this embodiment. This processing is performed by the window area vacancy presence / absence determination step 501 and the composite data remaining / absence determination step 502
Complex data root element acquisition step 503, and OI
D check step 504, object cache area existence determination step 505, window area data acquisition step 506, pointer conversion presence / absence determination step 507, sort area OID storage step 508, hash table OID storage step 509, combined Data remaining determination step 510, composite data condition determination step 511, condition satisfaction determination step 512, result data output / composite data deletion step 513, condition determined completion check step 514, and hash table O
An ID presence / absence determination step 515, a hash table base data acquisition step 516, a sort area OID sort step 517, a sort area base data acquisition step 518, and a composite data remaining determination step 519.

6 and 7 are diagrams for explaining the operation of the database access method for the complex object set of this embodiment.

Next, the operation of the database access method for the complex object set according to the present embodiment having the above configuration will be described. Here, the company information database having the class structure shown in FIG. 2 is the database 1
When it is 06, the operation when the inquiry as shown in FIG. 3 is made to the database 106 will be described as an example.

The database 106 (company information database) shown in FIG. 2 has the following class configurations.

As classes related to the department class, there are an employee class and a base class which are subclasses of the human class.

There are a laboratory class and a business unit class as subclasses of the department class.

The laboratory class is related to the research achievement class.

The business unit class is related to the project class.

Project classes are related to product classes.

Database 10 having such a class structure
Consider inquiring 6 about "Which research institute is associated with the business unit that has the project developing the OODBMS product?" This inquiry is described in OQL as shown in FIG. Here, “business division” represents an extent set of a business division.

The attribute information of the above inquiry is as shown in FIG. This attribute information represents the structure of the composite data required to execute the query process, and is the "attribute name".
, "Type", and "flag" of an attribute necessary for condition determination.

In the attribute information shown in FIG. 4, the attribute information of the project class and the attribute information of the product class are obtained from the attribute information of the business unit class, which is the root of the composite data, in order to obtain the product name included in the query condition expression. The flags of the business unit class, the project class, and the product class are all set to 1. Further, in order to obtain the name of the laboratory as a result (inquiry result), a link is provided from the attribute information of the business unit class to the attribute information of the laboratory class. However, since the laboratory does not need data for condition determination, the laboratory class flag is 0.

The operation of the database access method of this embodiment will be described with reference to FIGS. 5 to 7 based on the above database 106 and the example of the inquiry.

First, the composite data acquisition unit 102 determines whether or not the window area 108 in the main storage area has a free space (step 501).

The composite data acquisition unit 102 executes step 50.
When it is determined in 1 that "there is a free space in the window area", it is determined whether or not there is a remainder in the extent set of the business unit class which is the root of the composite data to be expanded (step 502). The extent set of the class that is the root of the composite data to be expanded (here, the extent set of the business unit class) has been read in advance in the object cache area 107.

The composite data acquisition unit 102 executes step 50.
When it is determined in 2 that "there is a rest in the extent set of the business unit class (route of the composite data to be expanded)",
One of the elements of the extent set (the root element of the compound data to be expanded) is set as the object cache area 10
7 into the window area 108 (step 5)
03).

Next, the composite data acquisition unit 102 checks the OID traceable from the data expanded in the window area 108 in step 503 (step 50).
4). Here, the OID of the project instance and the laboratory instance that can be traced from the business unit instance is inspected, and the OID of the product instance that can be traced from the project instance is further inspected.

Then, in the composite data acquisition unit 102, the data to be traced (data traceable from the data expanded in the window region 108 in step 503) exists in the object cache region 107 based on the inspection in step 504. It is determined whether or not (step 50)
5).

The composite data acquisition unit 102 executes step 50.
5, “the relevant data is the object cache area 107
Exists in the window area 108 from the object cache area 107 (step 506).

On the other hand, if the composite data acquisition unit 102 determines in step 505 that "the relevant data does not exist in the object cache area 107", then step 5
Based on the inspection of 04, it is determined whether or not the OID of the data is pointer-converted (OID in the database 106 is converted to OID in the main storage area) (step 507).

The composite data acquisition unit 102 proceeds to step 50.
When it is determined in 7 that "the OID has been pointer-converted", the OID is stored in the sort area (the sort area in the calculation area 107) (step 508).

The composite data acquisition unit 102 executes step 50.
If it is determined in 7 that “the OID has not been pointer-converted”, the OID is passed to the hash processing unit 103. The hash processing unit 103 hashes the OID with the PID, and stores the OID in a predetermined bucket (bucket obtained as a result of the hash) of the hash table (hash table in the calculation area 107) (step 509). .

The composite data acquisition unit 102 performs the above-described series of processing until the window area 108 becomes full (when there is no space in the window area 108, when the elements of the extent set of the business unit class disappear Up to) (see steps 501 and 502).

The data expansion state shown in FIG. 6 is a state in which such a repetitive process is performed and the window area 108 becomes full (the state in the middle of the process shown in FIG. 5).
FIG. In this figure, the data (instance) surrounded by a solid line and shaded represents acquired data (developed in the window area 108), and the data surrounded by a broken line represents unacquired data.

The composite data acquisition unit 102 proceeds to step 50.
When it is determined in 1 that "the window area 108 has no free space" (for example, when the data is expanded as shown in FIG. 6), it is composite data in which even one set of necessary data (instance) is prepared. It is determined whether there is composite data that has not been subjected to the condition determination and remains in the window area 108 (whether there is remaining composite data) (step 5).
10).

The composite data acquisition unit 102 executes step 51.
When it is determined that 0 is "remaining compound data",
One of the composite data (composite data in which even one set of necessary data is prepared) is passed to the condition determination processing unit 105. The condition determination processing unit 105 determines the condition of the composite data (step 511). That is, the condition of the composite data for which the product instance data is acquired is determined. Then, the condition determination processing unit 105 determines whether the composite data satisfies the condition (step 512).

Here, the condition judgment processing unit 105 judges the condition even for partially assembled composite data (composite data in which at least one set of necessary data is prepared but not all data). To do this, the attribute information in the attribute information area 110 (the attribute information of the composite data necessary for the inquiry process) is referred to.

The idea that the condition can be determined even when the composite data is partially assembled is not an essential idea in the invention according to claim 1 or claim 2 in the present application. The invention according to claim 1 or 2 can also be realized by performing the condition determination only on the composite data for which all the assembling is completed. In this case, the determination in step 510 (step 519 described later)
(The same applies to the determination of), "remaining composite data" means "completely assembled remaining composite data", and the composite data to be determined in step 511 is "composite data in which even one set of required data is available." It becomes "completely assembled composite data" instead of "data".

The condition determination processing section 105 executes step 512.
If it is determined that "the composite data satisfies the condition", the composite data acquisition unit 102 is notified to that effect. The composite data acquisition unit 102 outputs the composite data as the resulting data and deletes the composite data from the window area 108 (step 513). For example, in the example shown in FIG. 6, the product instance marked with a circle and the business unit instance and the project instance of the composite data related to the product instance are deleted.

On the other hand, if the condition determination processing unit 105 determines in step 512 that "the composite data does not satisfy the condition", it checks the composite data for the condition determination (step 514). The cross mark in FIG. 6 indicates the check for which the condition determination has been completed.

These processes (the processes and determinations in steps 511 to 514) are repeated for all the composite data in the window area 108 (see step 510).

The composite data acquisition unit 102 executes step 51.
If it is determined that 0 is "there is no remaining composite data",
It is determined whether or not there is an OID in the hash table in the calculation area 109 (step 515).

The composite data acquisition unit 102 executes step 51.
When it is determined in 5 that “the OID exists in the hash table”, the OID in an arbitrary bucket in the hash table is acquired, and the OID and control are passed to the data page access unit 101. Data page access unit 10
1 accesses the area in the database 106 designated by the pointer related to the OID and reads the data identified by the OID into the object cache area 107. The composite data acquisition unit 102 moves from the object cache area 107 to the window area 108.
The data is obtained (step 516).

The composite data acquisition unit 102 proceeds to step 51.
When it is determined in step 5 that “the OID does not exist in the hash table”, control is passed to the sort processing unit 104. The sort processing unit 104 sets O in the sort area in the calculation area 109.
The IDs are sorted (step 517).

The composite data acquisition unit 102 proceeds to step 51.
The OIDs are acquired in the order sorted by 7, and the object cache area 107 is acquired by the pointers associated with each OID.
Is accessed to acquire the data identified by the OID from the object cache area 107 into the window area 108 (step 518).

The composite data acquisition unit 102 executes step 51.
When the process of 6 is completed or the process of step 518 is completed, the process returns to the determination of step 501 again.

The data expansion state shown in FIG. 7 represents that new composite data is expanded by deleting the composite data satisfying the conditions from the data expansion state shown in FIG. In this figure, the data (instances) surrounded by solid lines and shaded represent newly acquired (expanded) data, and the data surrounded by solid lines and not shaded are in the state shown in Fig. 6. The data surrounded by the broken line represents the data that has not been acquired yet.

If the composite data acquisition unit 102 determines in step 502 that "there is no rest in the business unit class extent set (route of the composite data to be expanded)", the remaining composite data in the window area 108 is displayed. It is determined whether or not there is data (determination similar to step 510) (step 519), and if there is no remaining composite data, the processing ends (if there is remaining composite data, the above-mentioned step is performed). Processing and determination after 511 are performed).

[0056]

As described above, according to the present invention,
Both methods (Pointer-based) in the prior art
By integrating the Join method and the Assembly method), the following effects (1) to (4) are produced.

The physical scheduling of disk access becomes unnecessary, and the effect of clustering becomes prominent in improving the processing efficiency of query processing.

Even if the acquired composite data (composite data expanded in the window area) is partially configured, the processing can be performed if the condition determination in the query processing is possible, so that the processing is as redundant as possible. It is possible to reduce unnecessary page access. That is, at the time of page access of the database, it is determined whether or not the page exists in the main storage tilt area, so that page access to the redundant secondary storage area can be suppressed.

By considering pointer conversion,
It is possible to effectively use the object cache area.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a database access method for a complex object set according to the present invention.

FIG. 2 is a diagram showing an example of a class configuration in a database (company information database) in FIG.

FIG. 3 is a diagram showing an example of an inquiry (inquiry described in OQL) in the database access method shown in FIG.

FIG. 4 is a diagram showing attribute information of the inquiry shown in FIG.

5 is a flowchart showing a processing procedure of a database access method for the complex object set shown in FIG. 1. FIG.

FIG. 6 is a diagram for explaining a state in the middle of processing of the database access method for the complex object set shown in FIG. 1.

FIG. 7 is a diagram for explaining a state in the middle of processing of the database access method for the complex object set shown in FIG.

[Explanation of symbols]

 101 data page access unit 102 composite data acquisition unit 103 hash processing unit 104 sort processing unit 105 condition determination processing unit 106 database 107 object cache region 108 window region 109 calculation region 110 attribute information region

Claims (3)

[Claims] 1. In an object-oriented database management system, a main storage area for acquiring composite data necessary for determining a condition of a query process when performing a database access process for a large-scale complex object set that is complicatedly associated. A window area, which is an arbitrary area above, and composite data is assembled into the window area, and in the assembly, not only data existing only in the secondary storage area based on the inspection of the OID but also existing in the main storage area. Data to be accessed preferentially, and when the object is traced, a composite data acquisition unit that can perform transition of the object for each page using a PID obtained from the OID of the object, and based on an instruction from the composite data acquisition unit , The condition judgment of the composite data in the window area A database access method for a complex object set, comprising: a condition determination processing unit for performing. 2. Based on the control of the composite data acquisition unit,
If the target data does not exist in the main memory area, P the OID
A hash processing unit that makes it possible to read a page from the database into the main storage area by considering the access order of physical pages by hashing with an ID, and the OID is pointer-converted based on the control of the composite data acquisition unit. And a sort processing unit that makes it possible to obtain data in an address order from the object area in the main storage area to the window area by sorting the OID. Database access method for complex object sets. 3. When the composite data is partially assembled by referring to the attribute information area in the main storage area having the attribute information of the composite data necessary for the inquiry processing and the attribute information in the attribute information area However, the database access method for the complex object set according to claim 1 or 2, further comprising: the condition determination processing unit capable of performing a condition determination of the composite data.