JPH02222044A - Data processor - Google Patents

Abstract

PURPOSE:To process data at a high speed by decreasing the I/O frequency for restorage making the most of the clustering characteristics with a clustered structure type data base. CONSTITUTION:The records are successively read out of an old data base 32, and a step-based record sorting table 21 is produced by reference to a data dictionary file 31. The record contents of the table 21 are successively simulated via a new data base 33 to decide the places for storage of these contents. In this process, a pointer conversion table 22 is produced. The pointer part of the record contents of the table 21 is converted by reference to the table 22. Then the contents of the table 21 are rearranged for production of a new data base store record table 23. The record contents of the table 23 are successively written into the base 33.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Field of Industrial Application) The present invention relates to a data processing device characterized by a restorage processing function of a clustered structural database.

(Prior art) In structural databases, especially databases with a network structure, there is a parent-child relationship, and child records are placed near parent records (clustering) to speed up searches. Can be done.

Parents and children are managed on a record-by-record basis.

The place where these records are stored is called a file.

This file is realized on a medium such as a magnetic disk that has a random access mechanism and the information remains permanently.

The relationship between a parent and a child can be realized on a file using information called a pointer that indicates a position within the file.

Normally, files have a limited capacity, and as the number of records increases, the capacity becomes insufficient.

At this time, it is necessary to increase the file capacity and store the records again (re-storage processing).

In this re-storage, the position of the child record to be clustered is determined after the position of the parent record is determined.

Conventionally, in a configuration in which a secondary storage device such as a magnetic disk is connected to a data processing device, a process is performed in which a parent record is read from an old database and written to a new database. Therefore, 110 times of reading to the old database is required by multiplying the total number of I10 units (blocks) of the database for each record type that makes up the data structure, and 110 times of writing is required, which is approximately the number of records. was needed.

For example, in the data structure shown in FIG. 6(a), if the old database is as shown in FIG. One record type (RA.

RB, RC, RD, RE) multiplied by the number of blocks in the old database, which is 110 times (number of writes) in storing to the new database shown in Figure 6 (C).
is approximately the same as the total number of records Rm stored in the old database shown in FIG.

(Problem to be Solved by the Invention) As described above, in the past, the restoring process of a clustered structural database required a large number of 110 times, which was a big problem in trying to speed up the process. Ta.

SUMMARY OF THE INVENTION An object of the present invention is to provide a data processing device that speeds up processing in a clustered structural database by reducing the number of times of restorage by 110 times by taking advantage of clustering characteristics.

[Structure of the Invention] (Means and Effects for Solving the Problems) The present invention provides a data dictionary that defines a data structure and a data dictionary that defines a data structure and an old database to be processed, in a process of restoring a clustered structural database. means for sequentially reading records, classifying the same records by level of data structure with reference to the data dictionary, and creating a stage-based record classification table; means for simulating the storage block number of the new database; means for creating a conversion table that corresponds old pointers and new pointers for each record; means for converting pointer portions of record contents; means for rearranging the staged record classification table in the order in which they are stored in the new database to create a new database storage record table; The system is configured to have a means for sequentially writing to the new database, and a means for writing to the old database.
By keeping the 0 times (number of reads) to the number of blocks and the 110 times (number of writes) to the new database to the maximum number of blocks in the new database, the 110 times can be significantly reduced, and processing can be speeded up accordingly. .

(Example) An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an apparatus in an embodiment of the present invention.

In FIG. 1, ■ is a data processing unit that controls the system, and here, the mass storage device 2 and magnetic disk device 3, which will be described later, are accessed and rasterized as shown in the flowchart of FIG. Execute the re-storing process for the structured database. Reference numeral 2 denotes a large-capacity storage device with a sort/merge function, which performs data input/output processing including sort/merge processing under the control of the data processing section 1. A stage-based record classification table 21, which is classified according to the stage to be stored in the new database, as shown in FIG. A pointer conversion table 22 having a map structure corresponding to new and old pointers for rewriting pointers on the same table, a new database storage record table 23 as shown in FIG. 5, etc. are realized. 3 is a magnetic disk device accessed under the control of the data processing section 1, and on this magnetic disk device 3 there is a data dictionary file 31 having a data structure as shown in FIG.
, old and new databases 32, 33, etc. are realized.

FIG. 2 is a flowchart showing a procedure for restoring a clustered structured database in an embodiment of the present invention, which is executed under the control of the data processing section 1. In the figure, Sl is a step for creating a staged record classification table 21, in which records are read in order from the old database 32, and while referring to the data dictionary file 31, the staged record classification table 21 is created as shown in FIG. Create 21. S2 is a step of executing a new database storage simulation, in which it is simulated in which location on the new database 33 the record contents of the staged record classification table 21 are stored in order, and in this process, the pointer conversion table 22 is Create. S3 is a step of converting an intra-record pointer, and converts the pointer part of the record contents existing in the stage-based record classification table 21 to the pointer conversion table 22.
Convert while referring to. S4 is a step of creating a new database storage record table 23 as shown in FIG. 5 from the stage-based record classification table 21 shown in FIG. Sort by stored position. S5 is a step of writing to the new database 33, and the new database storage record table 2
The record contents in 3 are sequentially written to the new database 33.

FIG. 3 is a diagram showing the structure of the staged record classification table 2I implemented on the mass storage device 2. As shown in FIG.

That is, taking the data structure shown in FIG. 6(a) as an example, the stage-based record classification table 21 has RA at level 1.
, R'B, RC at level 2, RD at level 3, RE, ・
It is a table classified by level, such as...
One entry contains a level number LN, a parent pointer (
(Null value if there is no parent) PI3. Pointer PA indicating the location on the old database, record identifier RID, and data content DA on the old database
Has TA.

FIG. 4 is a diagram showing the record format of the data section (DATA) stored in the staged record classification table 21 shown in FIG. 3, and a pointer section is provided along with the user data section.

FIG. 5 is a diagram showing the structure of the new database storage record table 23 which is also implemented on the mass storage device 2. As shown in FIG. This new database storage record table 23 is
The data are sorted/merged in the order of the block numbers of the new database 33, and data within the same block is arranged in the order of the level numbers shown in FIG.

Here, the operation in one embodiment of the present invention will be explained according to the flowchart shown in FIG.

In step S1 in FIG. 2, the data processing unit 1 reads the record contents of the old database 32 block by block from the magnetic disk device 3, and constructs the data structure by referring to the data dictionary file 31 for all the records in the block. The records are classified by level and stored in the staged record classification table 21 shown in FIG. 3 in the mass storage device 2.

In this case, level l does not have a parent and the storage location is determined by the key value in the data (in Figure 6 (a), RA,
RB). Level 2 has a parent and can be traced at most twice from a record classified as level 1, that is, it can be traced directly from level 1 (RC in FIG. 6(a)). Level 3 has a parent and can be traced at most three times from a record classified as level 1 (in Figure 6(a), R
D, RE). It is classified as follows.

This classification process is completed by reading all blocks of the old database 32, thereby completing the staged record classification table 2 as shown in FIG.

In step S2 of FIG. 2, the storage block number of the new database 33 is simulated by tracing the staged record classification table 21 in the order of level 1 entries. At this time, the old position (pointer) and new position (pointer)
A pointer conversion table 22 is created that corresponds to the pointer conversion table 22. This pointer conversion table 22 is arranged in ascending order of pointer values.

In the case of level 1, the simulation is performed using the key value. For records that are clustered near the parent at level 2 or below, a new pointer is obtained from the pointer conversion table 22 using the old parent pointer, and that point is set as the storage location.

For records that do not require clustering, simulation is performed using key values or location information on the old database 32.

As a result of the simulation, the location of the staged record classification table 21 on the old database 32 as shown in FIG. 3 is converted to the location on the new database 33.

In step S3 in FIG. 2, as a result of the process in step S2, the pointers on the old database 32 indicating parent and child records in the data content part of the updated staged record classification table 21 are converted to pointers on the new database 33. do.

In step S4 of FIG. 2, the staged record classification table 2I having the configuration shown in FIG.
The new database storage record table 23 as shown in FIG. 5 is completed. This new database storage record table 23 has data arranged in block order and level order.

In step S5 in FIG. 2, the record contents of the new database storage record table 23 are written into the new database 33 in the order of their arrangement. The 110 times (number of writes) at this time is the number of block numbers in the table 23, and is the maximum number of blocks in the new database 33.

By using the above-mentioned restorage processing means, it is possible to make the 110 times (number of reads) to the old database equal to the number of blocks, and at the same time make the 110 times (number of writes) to the new database equal to the number of blocks in the new database. As a result, the input time and output time can be shortened, and the structural database restoring process can be speeded up.

In addition, by storing the new database storage table on external storage such as magnetic tape, it is possible to save/save the database.
Can be migrated.

Note that the restorage processing means according to the present invention can also be applied to the reconfiguration and reconstitution functions of a structural database.

[Effects of the Invention] As detailed above, according to the present invention, in the restoring process of a clustered structural database, records are sequentially retrieved from the data dictionary that defines the data structure and the old database to be processed. means for reading and classifying the same records by level of data structure with reference to the data dictionary to create a stage-based record classification table; means for simulating the storage block number of the database; means for creating a conversion table that corresponds old pointers and new pointers for each record; means for converting the pointer portion; means for rearranging the staged record classification table in the order in which they are stored in the new database to create a new database storage record table; With this configuration, the number of times the old database is read as the number of blocks, and the number of times that the new database is written as the number of blocks of the new database can be significantly reduced by 110 times. , Accordingly, the restorage process can be sped up.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention, FIG. 2 is a flowchart showing the restorage processing means in the above embodiment, and FIG. 3 is a staged record classification table in the above embodiment. 4 is a diagram showing the formation of records in the above table. FIG. 5 is a diagram showing the structure of the new database storage record table in the above embodiment.
The figure is a diagram for explaining a data structure targeted by the present invention and a restorage processing means in a conventional database. ■...Data processing unit, 2...Mass storage device, 3.
...Magnetic disk device, 21...Stage record classification table, 22...Pointer conversion table, 23...
・New database storage record table, 31...Data dictionary file, 32...Old database, 33.
...New database.

Claims (1)

[Claims] In the process of restoring a clustered structured database, records are read sequentially from the data dictionary indicating the data structure and the old database to be processed, and the records are classified by level of data structure with reference to the data dictionary. and a means for creating a staged record classification table;
Means for simulating the storage block number of the new database by tracing each record of the staged record classification table in the order of level entries, and means for creating a conversion table that corresponds the old pointer and new pointer for each record; Means for converting the pointer part of the record contents of the stage-based record classification table by referring to the conversion table, and creating a new database storage record table by rearranging the stage-based record classification table in the order in which it will be stored on the new database. and the means to
A data processing device comprising means for sequentially writing records in the new database storage record table to the new database.