JPS63253431A - Retrieving system for data base of inverted structure - Google Patents

Abstract

PURPOSE:To improve the processing speed with the titled retrieving system by providing plural blocks into a block pool in a main memory. CONSTITUTION:The record numbers are extracted out of the index records read out after retrieving an index part of a data base 1 and converted into the block numbers within a block pool 2 and the bit numbers in the blocks. When the same blocks are included in the pool 2, the bit shown by the corresponding bit numbers is turned on and the next record number is processed. If the same blocks are not included in the pool 2, the corresponding bit is turned on as long as an idle block exists in the pool 2 and the next record number is processed. If no idle block is available, one block is selected out of the pool 2 and written to the corresponding block in the file 3. At the same time, it is checked whether the file 3 includes the same blocks or not. If so, these relevant blocks are read out. The bit of the read block is turned on an the next record number is processed. In such a way, the number of times of input/output is decreased, and the processing speed is improved with a data base retrieving system.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a database search method with an inverted structure, and in particular, an inverted structure that improves processing speed by reducing the number of data inputs and outputs between main memory and external memory. Concerning database search methods.

[Conventional technology]

This type of processing method is used when searching the database 11 having an inverted structure. As shown in Figure 3, an inverted structure is a structure that has an index part that consists of an index that identifies data records and the record number of the extracted data. By searching this index section instead of searching the data section consisting of data, the database 11
It becomes possible to localize access to.

Based on FIG. 2, the operation when searching for a data record whose index has a value of In or more will be described.

First, the index section of the database 11 is searched to find In, and the index record is read. The record number is extracted from the read index record and converted into a block number and a bit number within the block that correspond to the block of the file 13 (stored in the external memory). (This conversion is performed according to a predetermined method, but for example, of a record number with the upper N digits and the lower value M digits, the upper N digits are the block number in the file, and the lower M digits are the block number in the block. 1 number)) Then, the bit indicated by the bit number of the current block 12 stored in the main memory in the central processing unit is set on. This is done for all record numbers in the read index records. At this time, if a record with a different block number is detected, the contents of the current block 12 up to that point are written to the position of the file 13 indicated by the block number and then set.

The above process is performed for all index records whose index has a value of In or more. However, for the second and subsequent index records, when a record number with a different block number is detected, the current block 12 is written out and It is checked whether the block has already been written to the file 13, and if it has been written, the setting process is performed after reading the block.

Finally, the current block 12 is written to the file 13 to complete the search process, and the searched data records are stored as a set of bits on the file. To obtain the actual data record from this bit set, the process is reversed. All you have to do is read the block from the file, find the on bit, convert the number indicating the position of the bit and the block number into a record number, and access the data section shown in FIG. 3.

Thus, each time the record number corresponds to a different block within the file, the contents of the current block are transferred to the file in external memory.

[Problem that the invention seeks to solve]

The problem with the conventional inverted structure database search method described above is that each time the record number of the read index record corresponds to a different block in the file, the contents of the current block in main memory are transferred to the external memory. Since the data is transferred to a file, the processing time is long, that is, the processing speed is slow.

SUMMARY OF THE INVENTION An object of the present invention is to provide a database search method with an inverted structure that solves the above-mentioned drawbacks.

[Means for solving problems]

The database search method with an inverted structure of the present invention includes a database with an inverted structure, and a plurality of first index records with undefined block numbers stored in a main memory device that retrieves index records from the database and stores the record numbers. a block boolean having a block, and a file having a plurality of second blocks each having a predetermined block number stored in an external memory according to the record number of the index record, the record number of the called index record; A record number having the same block number is stored in one of the first blocks, and a record number having the same block number as the index record is stored in one of the first blocks. If the record number is stored in the first empty block, and there is no first block storing a record number having the same block number as the record number of the index record, and the first block is empty. If there is no available block, write one of the first blocks to the corresponding second block in the file, read the contents stored in the second block that should store the record number, and write the first block to the first free block. and writing the record number.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

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

First, an outline of this embodiment will be described. The structural difference between this embodiment and the conventional example is that the number of blocks in which record numbers are converted and bits are set is increased from one to a plurality of blocks. The difference in operation due to this is that in the conventional case, data input/output with the file is required every time a record number different from the block number of the current block is detected.
In the case of the present invention, data input/output to/from a file is not performed until a record number different from a plurality of block numbers in a block pool stored in a main memory in a central processing unit appears. In this example, database 1
, a block pool 2, and a file 3.

Based on FIG. 1, the operation of searching for a data record whose index has a value of In or more will be described in the same way as in the [prior art].

First, the index section of database 1 is searched to find In, and the index record is read. The record number is extracted from the read index record and converted into a block number within the block pool 2 and a bit number within the block (this method is similar to the conventional technique). Then, it is checked whether the same block exists in the block pool. If found, the bit indicated by the corresponding bit number of that block is turned on (O is set to 1) and the next record number is processed. If it turns out to be strange,
Check whether there is an empty block in the pool, and if it does, turn on the bit indicated by the corresponding bit number of this block and process the next record number. If it does not exist (all blocks in block pool 2 are in use), carry any one block from block pool 2, write it to the corresponding block on file 3, and write the same block to file 3. Check whether the block is stored, and if so, read that block and return it to a free block in block pool 2. Then, the bit of this block is turned on to process the next record number.

After the above is performed for all index records whose indexes have a value of In or more, the blocks used in the block pool 2 are written to the corresponding blocks of the file 3, and the search process is completed. As a result, the retrieved data record will be obtained as a set of bits on file 3.

For example, if the record numbers of index In are 1 and 1 million, I n
Even in the extreme case where +1 is also the same, assuming that the conventional current block can accommodate record numbers in units of 100,000, conventional technology requires data input and output to the file 6 times before the search process is completed. , one embodiment of the present invention requires only two inputs and outputs of data to the file.

In addition, by providing multiple blocks in block Boolean 2, the number of data inputs and outputs for writing and reading blocks between block pool 2 and file 3 until the search is completed is reduced compared to conventional technology. and decrease. If a block in block Boolean 2 exists corresponding to a block in file 3, data input/output only needs to be performed once.

〔Effect of the invention〕

As explained above, by providing a plurality of blocks in the block pool in the main memory, the present invention allows data to be written and read between the block pool and the file in the external memory until the search is completed. This has the effect of reducing the number of input/output operations and improving processing speed.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, FIG. 2 is a block diagram showing an example of the configuration according to the conventional technology,
FIG. 3 is an explanatory diagram showing the inverted structure of the database. 1...Database, 2...Block pool, 3.
··File.

Claims (1)

[Scope of Claims] A block having an inverted structure database and a plurality of first blocks with undetermined block numbers stored in a main memory device that reads an index record from the database and stores the record number. a file having a plurality of second blocks each having a fixed block number stored in an external memory according to the record number of the index record, the block number being the same as the record number of the called index record; is stored in one of the first blocks storing a record number having the same block number as the record number of the index record, and is empty when there is no first block storing a record number having the same block number as the record number of the index record. If there is no first block storing a record number with the same block number as the record number of the index record and there is no empty first block, , write one of the first blocks to the corresponding second block in the file, read the contents stored in the second block that should store the record number, and write it to the empty first block. A database search method with an inverted structure, characterized in that the record number is written together with the record number.