JPH06119219A - File data precedence read-in system - Google Patents

Abstract

PURPOSE:To obtain the file data precedence read-in system whose economical effect is also high without increasing the capacity of a main storage device by reading in data from a file in advance by preceding a processing, at the time or reading in successively the data in ascending order or in descending order. CONSTITUTION:A main control part 102 checks whether a CI which is read in before the CI in which a record which is being processed at present is stored in a CI buffer 109, and a data processing is finished already exists in the buffer 109 or not. As a result, when it exists, the CI of the next ascending order or descending order of the CI which is being processed at present is read in an area of the CI buffer 109 whose processing is finished by the number of CIs whose processing is finished. In this case, read-in of the CI is executed from one of a master file 111 or a slave file 112, an access is selected preferentially from the file whose access frequency is lower or whose channel is busy, and a load distribution is attempted. In such a way, the processing time of a data processing of ascending order or descending order can be shortened.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to reading data from a file in an auxiliary storage device, and more particularly, when reading data sequentially in ascending or descending order, file data is read from the file prior to processing. Regarding the read-ahead system.

[0002]

2. Description of the Related Art A conventional method of reading data from an auxiliary storage device is to sequentially read CIs containing records from a file in the storage region of the auxiliary storage device into a CI buffer in the storage region of the main storage device, and record the corresponding CIs. Had taken out. When the corresponding record exists in the CI buffer, the input / output operation for reading the record is not performed with respect to the auxiliary storage device. When the corresponding record does not exist in the CI buffer, the corresponding record is read into the storage area of the main storage device, so an input / output operation for reading the record from the auxiliary storage device occurs, and the record read request source is an auxiliary device. The start of processing is kept waiting until the input operation to the storage device is completed.

Referring to FIG. 6 showing a conventional file data reading method, in the file data reading system,
File open part 6 to specify the file reading conditions
02, a data sequential extraction unit 603 that sequentially reads CIs from the file 608 into the CI buffer 606, a reading unit 604 that sequentially fetches records from the CI buffer 606 to the data buffer 607, and a file closing unit that instructs the end of data reading. And 605. For example,
To sequentially process the data in the file 608, the file open unit 602 reads the data in the ascending or descending order and declares the sequential data processing. Put in. When reading the second and subsequent data, if the record of the corresponding data exists in the CI buffer 606, the CI input operation from the file 608 to the CI buffer 606 does not occur. The CI input operation is performed by the CI buffer 60.
It occurs when there is no data in 6, and the reading unit 604
Reads the CI in which the record of the corresponding data exists from the file 608 into the CI buffer 606 sequentially. The data sequential extraction unit 603 records the record of the corresponding data in the CI.
Transfer from the buffer 606 to the data buffer 607.
Therefore, the record of the corresponding data is the CI buffer 60.
6 does not exist, the data requesting source has to wait until the CI is input from the file 608 to the CI buffer 606, so that the data processing time is extended by the amount of data acquisition.

[0004]

In the above-mentioned conventional file data reading system, since the reading of file data and the processing of the request source operate in synchronization with each other, the record of the requested data is stored in the storage area of the main storage device. CI
When not in the buffer, the requester is kept waiting until the operation of obtaining the record from the auxiliary storage device is completed.

Further, if the record of the file in the storage area of the auxiliary storage device is made resident in the storage area of the main storage device, it can be processed quickly without waiting time, but it is necessary when the data amount of the file becomes large. Therefore, it becomes physically difficult to make a file resident in the main storage device.

It is an object of the present invention to, when the records of the file in the auxiliary storage device are sequentially read in ascending or descending order, the records to be read are asynchronous to the operation of the request source and the running level is low in the priority of processing execution. Thus, by reading in advance in advance of the processing of the request source, when the request source requests the reading of the request record, the request record can be quickly taken out from the record already read in the main memory, Further, since the record is read just before it is needed for data processing, it is not necessary to make all the records resident in the main storage device. In other words, it is possible to provide a file data read-ahead system with a high economical effect without increasing the capacity of the main storage device.

[0007]

According to the present invention, there is provided a file data read-ahead system for reading data from a file stored in an auxiliary storage device into a main storage device. A file opening means for instructing the data reading condition and a record as a logical processing unit are obtained from the file in ascending or descending order. Therefore, a control interval (hereinafter C
(Referred to as I) from the file, a data sequential extraction means, a CI buffer provided in the main memory for temporarily storing the CI read from the auxiliary memory, and an ascending or descending order of the records from the CI buffer. To the data buffer, an asynchronous read means for reading the CI from the file into the CI buffer to prefetch the record prior to the start of processing, and an asynchronous read for activating the asynchronous read means. It comprises a starting means and a file closing means for instructing the end of processing of the file.

[0008]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 showing the configuration of the first embodiment of the present invention.
File data read-ahead system 10
For 0, the file 111 in the storage area of the auxiliary storage device (not shown) and the CI including the request record are obtained from the file 111, and the CIs of the main storage device (not shown) are sequentially acquired in ascending or descending order. When the CI is fetched in the buffer 109, the record is fetched from the CI for data processing, the record is moved to the data buffer 110, and when the requested record does not exist in the data buffer 110, the file 11
The main control unit 101 that reads the CI including the new record from 1 and further, when the record of the request is already present in the data buffer 110, is started by the main control unit 101 to make a request for the file 111. And a main control unit 102 for prefetching records.

Further, the main control unit 101 reads the records in ascending or descending order, performs sequential data processing on the read records, and marks a sequential data processing flag in a common area (not shown) that can be referred to by each processing unit. And a CI that reads from the file 111.
Data to be stored in the CI buffer 109 for temporary storage, and whether or not records sequentially exist is checked. When no sequential records exist, the CI in which the requested record is stored from the file 111 is read into the CI buffer 109. A sequential fetching unit 104, a reading unit 105 for fetching and transferring request records from the CI buffer to the data buffer 110,
It has an asynchronous read activation unit 106 that is activated by the reading unit 105 and activates the main control unit 102, and a file closing unit 107 that closes the file 111 when a record to be processed does not exist in the file 111 and the CI buffer.

Next, the main control unit 102 causes the CI buffer 1
From 09, the CI that was read before the CI in which the record currently being processed is stored and whose data has already been processed is the CI.
It is checked whether or not it exists in the buffer 109. CIs corresponding to the number of processed CIs when processed CIs exist
To the CI buffer area. When the last CI of the file 111 has already been read, or when the file is closed by the file closing unit 107, the CI
The operation of the main control 102 is completed without reading the CI and reading the CI.

Further, when the record currently being processed is read before the stored CI and there is no processed CI, a new CI is not read and the main controller 10 is not executed.
2 completes the operation. The main control unit 102 is the main control unit 101.
It operates at a running level where the priority of processing execution in the electronic computer is lower, and operates during a quiet period of online processing. Then, an online processing unit (not shown) including the main control unit 101 preferentially operates at a traveling level having a higher priority of processing execution order than the main control unit 102.

Referring to FIG. 2 showing the configuration of the CI buffer 109 of this embodiment, the detailed configuration of the CI buffer 109 will be described in terms of the area in which the system overhead length is added to the CI length in the storage area of the main memory unit. CI consisting of area 202 of 1 to area 204 of CI buffer n
It is composed of a buffer 109. Then, the chain is managed from the latest area read recently by the LRU method to the old area.

When the CI in which the record of the file 111 is stored is read, the CI buffer 109 is read in advance, and the requested record is deblocked from the CI area and transferred to the data buffer 110. .

Referring to FIG. 3 showing the structure of the file 111 of this embodiment, the relationship between the file 111 and CI which is a data transfer unit at the time of input / output is 30 when CI is CI1.
2, stored in the file 111 in the order of 305 of CI2 to 307 of CIn. The record, which is the unit of processing, is stored in the order of record 1 303 and record 2 304, and the end of the file is recorded at record n 308. When sequentially read, CIs are sequentially read, and records from the CI are record 1 303, record 2 304
Finally, the records n are taken out in the order of 308 and passed to the request source. Records may cross CIs. 306 of the record k indicates a state where the CI is crossed, and the next CI
Is combined with the rest of the record parts included in the above to form one record. In this example, the record is C
It can be processed even when it exists across I.

Next, referring to the flow chart 4 showing the operation of this embodiment, the main control unit 101 is started in response to a request to obtain process data from a request source (not shown). First, the file open unit 103 instructs the control unit (not shown) to open the read file and specifies the processing condition of the record. When a sequential read process having a read-ahead function is designated in the file open, the file data read-ahead system of the present invention operates (step 401 in FIG. 4).

Next, the data sequential taking-out unit 104 is activated, and the data sequential taking-out unit 104 determines whether or not the record reading is completed. When the record reading is completed, the file 11
Because of the closing process No. 1, control is passed to the file closing unit 107 (step 404). When continuing the record reading process, a record reading request is made (step 403). When reading a record, the data sequential extraction unit 104 checks whether or not the requested record exists in the CI buffer 109 (step 405). If the request record does not exist, the CI storing the request record is read from the file 111 into the CI buffer 109 (step 406).

Further, the data sequential extraction unit 104 is
As in the case where the requested record exists in the buffer 109, it is checked whether or not there is a CI older than the CI of the record currently being processed in the CI buffer 109 (step 40).
7). When the CI that has already been processed is present in the CI buffer 109, the data sequential extraction unit 104 activates the asynchronous reading unit 106, and the asynchronous reading activation unit 106 advances the CI buffer 109 with the CI to be read next. In order to read in, the main controller 102 is activated (step 408). After activating the main control unit 102, the main control unit 101 does not wait for the completion of the operation of the main control unit 102, but asynchronously performs the process of the next main control unit 101. The reading unit 105 transfers the requested record from the CI buffer 109 to the data buffer 110 (step 409) and performs data processing of the record. Then, to determine whether or not to repeat the data processing, the process returns to the data read end determination (step 411).

The main control unit 102 is started by the main control unit 101, and the CI to be processed next is transferred from the file 111 to the C
The I buffer 109 is read into the area of the old CI which has already been processed. Next, when the CI to be processed is read from the file 111, in the case of sequential reading in ascending order, the CI in the ascending order after the unread CI is moved backward, and in the case of sequential reading in the descending order, the CI unread after the unread CI is read. Read the descending CIs forward. When there are a plurality of old CI areas that have already been processed, the above-described read processing is repeated until the CI buffer becomes full. Then, the plurality of CIs to be processed are read into the area of the CI buffer 109 in the storage area of the main storage device (step 412). In ascending sequential reading, the last CI to be processed has already been read into the CI buffer 109, or in descending sequential reading, the first CI has already been read into the CI buffer 109. Completion (step 413).

FIG. 5 showing the configuration of the second embodiment of the present invention.
Referring to FIG. 3, a file data read-ahead system in which files are duplicated by a master file 511 and a slave file 512 is shown.

This file data read-ahead system performs CI processing when records stored in files 111 and 112 are sequentially processed in ascending or descending order.
When the corresponding record does not exist in the buffer 109, the main control unit 101 which inputs the CI including the record from the file 111 or 112 in synchronization with the online processing, and the pre-reading of the CI including the record are performed asynchronously with the main control unit 101. It comprises a main control unit 102 and a data buffer 110 for temporarily storing records to be processed.

When performing the sequential data processing, the file open unit 103 declares the sequential data processing in the ascending order or the descending order, whereby the read-ahead function of the record operates. By the declaration of the sequential data processing of the file open unit 103, the sequential data processing flag is marked in a common area (not shown) that can be referred to by each processing unit of the main control unit 101. The data sequential extraction unit 104 uses the CI buffer 1
09, the request record in the CI buffer 109 is extracted to the data buffer 110 if it exists, and if it does not exist, the request record is stored from the master file 511. The existing CI is read into the CI buffer 109, and the record of the request is fetched. The reading unit 105 extracts the request record from the CI buffer 109 into the data buffer 110,
The control is passed to the asynchronous reading activation unit 106. The asynchronous read activation unit 106 activates the main control unit 102, and the main control unit 102
The processing of the main control unit 101 is continued without waiting for the completion of the operation. When there is no record to be processed, the main control unit 101 activates the file close unit 107 to end the sequential data processing, or when there is a record to be processed, returns control to the data sequential extraction unit 104 to repeat the processing. The main control unit 102 reads the CI that has been read before the CI in which the record currently being processed is stored in the CI buffer 109 and has already processed the data.
9 is present. When the processed CI is present in the CI buffer 109, the CIs in the ascending or descending order next to the currently processed CI are read into the area of the processed CI buffer 109 by the number of processed CIs. At this time, the CI is read from either the master file 111 or the slave file 112, and the access is preferentially selected from the file having a low access frequency or the file that is not in the channel busy state to distribute the load. When the last CI of file 111 has already been read,
The operation of the main control unit 102 is completed.

Further, when the record currently being processed is read before the CI in which it is stored and there is no CI buffer 109 that has already been data processed, the CI is not read and the main control unit 102 completes the operation. To do. Main controller 10
2 operates at a running level having a lower priority in processing execution order than the main control unit 101, performs processing during a quiet period of online processing, and preferentially executes the main control unit 101 included in the online processing.

Further, in the first and second embodiments,
The functions described above can be achieved by replacing the auxiliary storage device with a mass storage device such as a disk storage device or an extended storage device.

[0025]

As described above, according to the present invention,
When sequentially processing file data in ascending or descending order, the execution unit that processes the data and the reading unit that reads the file data in advance have separate processing execution functions, and the preceding reading unit is the processing execution unit. In contrast, by operating at a low priority level of asynchronous and processing execution order,
Data to be processed that precedes the data being processed by the execution unit that processes the data can be read from the file. Then, the processing time of the ascending or descending data processing can be shortened.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a CI buffer of FIG.

FIG. 3 is a diagram illustrating a mutual relationship between a CI and a record of the file of FIG.

FIG. 4 is a flowchart showing an operation of a file data prefetching process in the embodiment of FIG.

FIG. 5 is a diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 6 is a diagram showing a conventional file data reading system.

[Explanation of symbols]

 100 file data read-ahead system 101 main control unit 1 102 main control unit 2 103 file open unit 104 sequential data extraction unit 105 read unit 106 asynchronous read activation unit 107 file close unit 108 asynchronous read unit 109 CI buffer 110 data buffer 111 file 202 CI buffer 1 203 CI buffer 2 204 CI buffer n 302 CI1 303 record 1 304 record 2 305 CI2 306 CIl (record across CI) 307 CIn 308 record n 511 master file 512 slave file

Claims (2)

[Claims] 1. A file data read-ahead system for reading data from a file stored in an auxiliary storage device to a main storage device and instructing a reading condition of the data from the file stored in the auxiliary storage device. A file opening means and a data sequence for reading a control interval (hereinafter referred to as CI), which is a physical file data transfer unit, from the file in order to obtain a logical processing unit record from the file in ascending or descending order. Fetching means, CI buffer provided in the main memory for temporarily storing the CI read from the auxiliary memory, reading means for transferring the records from the CI buffer to a data buffer in ascending or descending order, the record To preempt the start of the processing,
An asynchronous read means for reading the CI from the file into the CI buffer; an asynchronous read start means for activating the asynchronous read means; and a file close means for instructing the end of processing of the file. File data read-ahead system characterized by: 2. The file data read-ahead system according to claim 1, wherein each of the files is stored in two auxiliary storage devices and is a duplicated file.