JPS62293448A - Space managing system for file - Google Patents

Abstract

PURPOSE:To improve the use efficiency of a file space and the performance at the time of a file access, by providing a matrix showing a dead block by a block number train, and a matrix showing a logical coupling of a busy block. CONSTITUTION:The allocation of a file using a block number matrix is executed by defining a data set name, and moving the block number of a necessary portion to a user block number matrix UBM from a free block number matrix FBM. At the time of using a dead block from the head of the FBM, in case of saving the contents of the used block, its block number is put in the end of the FBM. Write of a data is executed by a block unit in order from the head block of the UBM corresponding to the data set name. At the time of allocation, whenever write to the block is generated, the block number is moved newly into the UBM from the FBM, and write is continued by securing the block to which write is executed in the next time.

Description

[Detailed Description of the Invention] 3. Detailed Description of the Invention [Field of Industrial Application] The present invention relates to reading, writing, searching, adding, deleting, etc. of data files on a computer, etc. The present invention relates to a file space management method suitable for improving usage efficiency and performance during file access.

[Conventional technology]

The conventional file space management method is JP-A-57-15.
As described in Japanese Patent No. 3334, there are a pre-region setting method and a bitmap method.

In the bitmap method, a volume such as a magnetic disk is divided into a data area and a bitmap area, and the data area is further divided into blocks, which are the smallest access units. This one block corresponds to one control bit,
A bitmap area is created. The bit is usually It I
11 indicates that the block is in use (1 indicates that the block is in use) (011 indicates that the block is free).
Logical data continuity is guaranteed by the succeeding block pointer NP and the preceding block pointer BP.

Bitmap file allocation.

Define the dataset name and write ``0'''' on the bitmap.
This is done by searching for the block corresponding to it and setting it as the first block. At this time, the corresponding bit on the bitmap is set to tt l 71. The second and subsequent blocks are secured in the same way, and the NP of the first block is secured.

Rewrite the BP of the second block and connect.

Data reading is performed by taking out blocks one after another using pointer NP.

Using the bitmap method in this way allows data to be written and added as long as there is free space in the volume, depending on the amount of data generated, which can improve usage efficiency when a single volume is shared by multiple files. can.

Also, inserting data.

Deletion can be easily performed by rewriting the pointer.

[Problem that the invention seeks to solve]

In the above-mentioned conventional technology, using JP-A-57-153334, even if the empty state of a block is quickly known from the bitmap on the additional memory, the logical continuity between blocks can be determined by reading out the block and taking out the pointer NP. I didn't know until I tried it.

For example, to know the contents of the 100th block,
The block must be read out into memory 100 times, starting with the block.

In this way, in the conventional technology, since a block has pointers such as NP and BP that indicate a logical connection relationship, the number of accesses to the file increases as the block is read to follow the pointer, resulting in performance problems. The problem was that it got worse.

Also, bitmaps can only represent binary information, such as vacancy or not. Therefore, even if the property is vacant, it may be necessary to further distinguish it and display complex information such as whether it has been used in the past or how it was used in the past. Another problem was that the data structure became complicated.

The purpose of the present invention is to reduce the number of accesses to a file and improve file read/write performance by managing the logical connection status between blocks separately from the blocks. By managing the bond state in a concise representation.

The objective is to provide a method for effectively reusing used blocks.

[Means for solving problems]

The above objective can be achieved by replacing the bitmap BM with a free block number matrix FBM, separating the pointers NP and BP within the block from the block, and replacing them with the user block number matrix UBM, as shown in FIG. .

A feature of the present invention is that logical continuity between blocks is expressed by the order in which the numbers assigned to the blocks are arranged.

First, all blocks in the file are uniformly sized, and all blocks are numbered.

The bitmap is replaced by the block number sequence FBM of the vacant block. If blocks are used in order from the block number at the beginning of the FBM, priority is given to empty spaces depending on the position of the block number in the column.

The logical connection state between blocks in use can be seen at a glance by reading the UBM. Since the UBM is also simply a sequence of block numbers, by moving the numbers, data areas can be secured, added, deleted, released, etc. without accessing the actual blocks.

[For production]

File allocation using the block number matrix method is performed by defining a data set name and moving the necessary block numbers from the FBM to the UBM.

Empty blocks are used from the beginning of the FBM, but if you want to preserve the contents of used blocks as much as possible, you can put the block number at the end of the FBM.

Further, by setting a special indicator in the FBM block matrix, the block number matrix following the indicator can have a different meaning. In other words, it is possible to divide the matrix with indicators and use them with multiple meanings.

To write data, write tJ corresponding to the dataset name.
This is performed block by block sequentially starting from the first block of BM. When allocating a file, if only the first block is reserved, each time a write to a block occurs, the block number is moved from the FBM to the UBM to secure the next block to be written. As I continue writing.

Furthermore, if a predetermined number of blocks are prepared at the time of file allocation, writing proceeds in the order of block numbers arranged in the UBM.

Data can be read from any block within the UBM with a single input/output. If sequential data is included in a record, the specified data can be searched for using the binary search method in units of block numbers, thereby greatly reducing the number of accesses to the file.

〔Example〕

An example in which the present invention is applied to a read/write control system for a message recording file will be described below by combining FIGS. 1 and 2.

Message recording information must be collected separately for each user. Therefore, a profile matrix PM is also required that indicates whether the user has a UBM.

When building the system, all message recording files are divided into blocks of uniform size and the blocks are numbered. Block numbers 1 to 4 are for FB M, and 5 to 4 are for FB M.
No. 6 is used for PM, and No. 7 onwards are used as UBM or message block MB for actually storing message information at any time.

For example, when user A logs on, he first configures 1 for UBM.
Extract the block number from the first position of block unit M. Further, user A moves the block number string corresponding to the specified message recording data page from the FBM to the UBM. At this time, the PM memorizes the user A's user name and UBM number.

When one block of user A's message data is accumulated, it is written to the block at the beginning of the UBM. In PM,
Among the user data, a block number FωBN containing the oldest data and a block number FωBN containing the newest data are inserted.

The data consists of one line of message output to the VDT terminal, the time the message was output, and the page on the terminal as a code. 6 Specify a specific time or number of pages to record the message record data. When redisplaying, if you search for data at a specified time or page in the order in which the block numbers are arranged, in the worst case, you will have to access the disk N times for all blocks. At the same time, the time or page number increases sequentially, so if you perform a binary search on the block number string, read out the block, and search for the specified data, the number of accesses to the file will be JZyzN times, and the amount of data will increase. However, the number of accesses to the file did not increase much.

A constant search time can be guaranteed.

Furthermore, when user A logs off, the block on the UBM that user A was using is moved to the FBM again and reused as a free block.

As shown in FIG. 3, the FBM is oooo.

It is separated by three special indicators: 0001 and FFFF. After logoff, the block number is rearranged in order of oldest data and moves between 0001 and FFFF.

When IPL is performed, 0001 and FF are displayed as shown in Figure 4.
The block number string between FF is oooo and 000 as is.
1.

The data of the blocks in these special indicators can also be printed by command if the blocks are not reused.

Furthermore, when the block before the indicator oooo is completely exhausted, it is used again as a block for a new user.

〔Effect of the invention〕

According to the present invention, since the joint management between blocks is separated from the blocks, reading and writing of a file can be performed at one time by reading the UBM, thereby improving the performance of reading and writing.

Addition or deletion of blocks can be performed simply by moving numbers between block matrices, so performance is much better than before, especially when making large additions or deletions.

Furthermore, if the data includes sequential elements, binary search can be performed on a block-by-block basis when searching for data. Compared to the case of searching in order, the number of accesses to a file can be reduced by (N-%zN) times, where N is the number of blocks. This difference increases as the value of N increases, and the search performance also improves.

Furthermore, you can easily use it in a variety of ways by assigning priorities or dividing the matrix by inserting special indicators depending on its position on the matrix.

[Brief explanation of drawings]

Figure 1 shows a conventional file access control system. An explanatory diagram of an embodiment in which the bitmap method of the present invention corresponds to the block number matrix method of the present invention, FIG. 2 shows a profile matrix table in which matrix numbers used by each user are registered, and FIG. 3 shows a profile matrix table in which a user logs off. FIG. 4 is an explanatory diagram showing an example in which the block number is later returned to the FBM, and FIG. 4 is an explanatory diagram showing an example in which the block number in the FBM is further moved after IPL. BM...Block map, B...In use, ■...
Sky. FBM...Free block number matrix, BN...
・Block number, IND...Special indicator. DB...Data block, BP...Preceding block pointer, NP...Subsequent block pointer, UBM...
・User block number matrix, BN...Block number r DR1~DR4...Record, PM...
Profile matrix, USNAME...User name. UBMNO: UBM block number.

Claims (1)

[Claims] 1. In a file read/write control method consisting of a data area consisting of a plurality of blocks and a control area of the block,
Each block is assigned an individual number, and a control area is provided with a matrix in which free blocks are represented by their block number sequences.
A file space management method characterized by providing a matrix that shows logical combinations of blocks in use by block number sequences.