JPH04190434A - Device and method for data base control - Google Patents

Abstract

PURPOSE:To avoid long-time locking by locking and updating data in a data base by a data base control program when a premise data specification and an update data specification are inputted by an application program. CONSTITUTION:The application program 1 inputs the premise data specification 2 and updated data specification 3 and sends an updated request to the data base control program 4. The program 4 locks the data in the data base 8 by a lock means 5 according to the specifications 2 and 3. Then a comparing means 6 compares the specification 2 with the value of a specified data item in a file to decide whether data updated is allowed or not, and then updates the data base 8. Consequently, the data in the data base can be updated without long-time locking.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a database management device and a database management method, and in particular, to update processing of a database shared by a plurality of parallel processes. It relates to database management programs.

[Prior Art] Generally, when the same database is used simultaneously by multiple processes, parallel processing control is required to maintain consistency of the database.

Therefore, conventionally, this control has been performed using a two-phase locking protocol as shown in FIG. 7, for example.

This is Henry F., Korth, Abraham
am 5ilberschatz rDATA BAS
E CONCEPTSJ Page 375 This is a flowchart of the method described in McGraw-Hill Book Company, published in 1986. In the figure, (26) to C29) are cooperative operations between the application program and the database management program. It represents the program steps of the process performed.

In step (26), all data items are locked and read from the database, in step (27), the changed values of the data items are set, and in step (28), the data in the database is updated, and then in step (29) Then, release the lock made in step (26).

In this way, the database shared by multiple processes being processed in parallel is updated, so the consistency of the database is maintained.

By the way, in interactive processing in which data input from a database is displayed on a terminal and the terminal operator refers to the value and determines the value to update the data, this terminal processing should be performed in step (27). become.

In general, in the case of conversational processing, the time required for terminal operation depends on the terminal operator's method of operation; if it is fast, it can be completed in a few seconds, but if it is slow, it can take several hours.

Since it is impossible to predict this time in advance, it is desirable to create an application program assuming the case where it will take the longest time.

In this case, the data locked in step (26) will remain locked for several hours during terminal processing in the worst case.

If it was determined that it was inappropriate for the system's performance to keep data locked for a long time, a program was created to perform the process shown in Figure 2.

In the figure, (30) to (38) represent program steps of processing performed in cooperation with the application program and the database management program.

In step (30), data items are input from the database without locking them. In step (31), data that is a prerequisite for the data update performed in step (35) is saved in the work area. In step (32), data items are input without locking. , perform terminal processing, and step (3)
In step 3), after locking all data items from the database, read the locked data, and in step (34) compare the data saved in the work area in step (31) with the value entered in step (33). If the values are all the same, step (3)
Update the database in step 5), and then proceed to step (33).
The lock made in step (36) is released.

Here, in updating the database in step (35), since normal application programs are written in a high-level language, update commands provided in the high-level language are used. For example, if you use the REWRITE instruction,
The REWRITE instruction usually first reads the data,
Since this step (35) is made up of a routine that updates and rewrites the relevant items, input processing is also performed outside the application program's knowledge.

Now, if the comparison in step (34) is different, in step (37) as an exception handling,
After notifying that the update was not performed as requested and releasing the lock made in step (33) in step (38), the process is re-executed from step (31).

This procedure shortens the locking time and updates the database.

[Problems to be Solved by the Invention] In conventional database updating using conversational processing where terminal processing time cannot be predicted, data must be kept locked for a long time as described above, or the same data must be input more than once. It was necessary to do it.

However, it is possible to distinguish whether the lock has not been released because the terminal processing took time, whether the lock has not been released because the application program has terminated abnormally, or whether the lock has not been released because there is an error in the application program. Because this is difficult, database management programs usually limit the amount of time a lock can be held, and then release the lock after that period of time, eliminating subsequent update processing.

Therefore, in the method of locking data and then performing terminal processing, there is a problem in that the time available for terminal processing is limited by the time that the lock can be continued.

Also, in the method of inputting the same data twice,
There is a problem in that unnecessary input processing that places a large load on the computer must be performed more than once.

Furthermore, in either case, the order in which data is locked depends on the coding of the application program, so there is a problem that the possibility of deadlock occurring cannot be avoided.

The present invention was made to solve these problems, and it avoids locking data items for a long time, reduces the number of data input processes, and also avoids deadlocks and updates the database. [Means for Solving the Problems] A database management device according to the present invention has the following elements.

(a) a database that stores data; (b) means for operating an application program, job, task, or process that issues the following information to request updates to the data stored in the database; (b1) data; (b2) Update data specification that specifies the data after the data to be updated; (c) Database management process (database management) that has the following elements: (c1) Locking means for locking the data in the database by inputting the prerequisite data specification and update data specification issued by the application means (application program) (c2) Inputting the prerequisite data specification and the value of the data in the database and (c3) an updating means for inputting update data designation and updating data in the database based on the comparison result of the comparing means.

Further, the database management method according to the present invention includes the following steps.

(a) An application process that inputs data from a database and, after predetermined processing, issues a data update request by issuing update prerequisite values and update data; (b) A database management process (b1) that includes the following steps:
) A locking process that inputs a data update request issued by the application process, locks and inputs the data in the database, (b2) inputs the update prerequisite value and the value of the data in the database, and compares these values. comparison step; (b3) update step of inputting update data and updating data in the database based on the comparison result of the comparison means;

(b4) An unlocking step for unlocking the data locked in the locking step.

[Operation] In the present invention, the application program performs the first input and issues an update request to the database management program. The database management means (database management program) locks the data based on the prerequisite data specification and update data specification issued by the application program in the locking means and performs a second input, and in the comparison means, the data is inputted for the second time based on the prerequisite data specification and the update data specification issued by the application program. The value of the data item in the file is compared with the value in the file to determine whether the data can be updated, and the updating means updates the data based on the update data specification and the result determined by the comparison means, and updates the database. Let's do it.

In this manner, locking is performed in the database management program only from the second data entry until the data is updated, thereby avoiding locking for a long time.

Furthermore, since the database management program is written in a low-level language to improve efficiency, it can be written in a manner that avoids unnecessary input during updating.

Furthermore, since there is only one database management program in the system and multiple application programs executed in parallel use the locking means of this database management program, deadlocks can be avoided.

[Embodiment] Figure 1 shows an embodiment of a database management device according to the present invention. In this figure, (1) indicates an application program, and (2) indicates a precondition for updating data. (3) is an update data specification that specifies the updated data of the data to be updated; (4) is a database management program that serves as a database management means; (5) is a locking means; (6) ) is a comparison means, (7) is an update means, (8) is a database, (10) is an update data record, and (11) is an end code.

FIG. 2 is a diagram showing an example of an update request command issued by the application program (1) to the database management program (4).
An example of the TE command and the prerequisite data designation (2) and update data designation (3) issued with this command are shown.

Prerequisite data specification (2) in Figure 2 requests that if the value of item number 5 of RECORD number 4 of FILE number 1 is "3", it is updated to the value specified in update data specification (3). are doing.

That is, if the database is as specified in the above prerequisite data specification (2), it is requested that the value of item number 5 of RECORD number 4 of FILE number l be set to "4".

In this way, the application program (1) requests the database management program (4) to update data using, for example, the UPDATE command as shown in FIG. As shown in Figure 2, it is issued from the application program (11) to the database management program (4) and consists of multiple data item specifications and their values.

Also, as shown in FIG. 2, the update data designation (3) is issued from the application program (1) to the database management program (4), and is a locking means (3) consisting of multiple data item designations and their values. 5) is
A part of the database management program (4) locks data based on the prerequisite data specification (2) and the update data specification (3).

Comparison means (6) is a database management program (4)
In this part, it is determined whether data can be updated based on the update data record (io) and the prerequisite data specification (2).

The data update means (7) is a part of the database management program (4), and updates the update data record (10) from the update data designation (3) based on the determination by the comparison means.

The update data record (10) is a data record in the database (8).

The exit code (11) is the database management program (
4) to the application program (1).

In a system having such a configuration, the database is updated based on the flowcharts shown in FIGS. 3 and 4. 3 and 4 are flowcharts showing an embodiment of the database management method according to the present invention, FIG. 3 is a flowchart showing the processing of the application program (1) which is the application step, and FIG. It is a flowchart showing the processing of the locking means (5), the comparing means (6), and the updating means (7) of the database management program (4), which are management steps.

First, the application program (1) executes processing based on the flow shown in FIG.

In step (12), data is input without locking the data. In the example of Figure 2.

FILE number I RECORD number 4 Value of item number 5 “3” Enter.

In step (13), among the data input in step (12), when updating in the subsequent step (15),
Save what is used as prerequisite data specification (2) in the work area. Save the data input in step (12) above, FILE number l° RECORD number 4, value “3” of item number 5, in the work area.

In step (14), terminal processing is performed, and step (1)
The data entered in step 2) is displayed on the terminal, and the terminal operator is prompted to input the updated data value. In the example of FIG. 2, enter the value "4''' for FILE number I RECORD number 4 and item number 5.

In step (15), the data saved in step (13) is designated as prerequisite data (2), and the data specified by the terminal operator in step (15) is designated as update data (3) in the database management program (4). A data update request is made to the server.

In step (16), the result of the data update request in step (15) is determined from the completion code (11), and if the update has not been performed as requested, the terminal operator It notifies that the update was not performed as requested, and re-executes from step (12).

Step (1) of the above application program (1)
The data update request made in step 5) is processed by the database management program (4) according to the processing procedure shown in FIG.

Steps (181, (19)) are locking means (first half) (5a) which is the first half of locking means (5).

This locks and inputs data (locking process).

In step (18), the application program (1
), all update data records (10) including the data items specified by the prerequisite data specification (2) and the update data specification (3) are locked.

This locking order is determined, for example, as follows.

If the data records are on different files,
Data records are processed in the order of uniquely added file numbers within the database, and in the case of data records in the same file, data records are processed in the order of uniquely added record numbers within the file.

Therefore, even if this step (18) is called and executed by multiple processes or multiple application programs that are executed in parallel, deadlock will not occur.

Next, in step (19), all update data records (10) including data items specified by prerequisite data specification (2) and update data specification (3) are sent to the database (8).
The locks and data record inputs performed here are contained in the same data record, even if they are different data items or with different specifications. If so, only one lock and one input are performed on that data record. For example, if you want to update item 3 and item 5 of RECORD number 1, RECORD number 1
It is only necessary to lock RECORD once and input it once, and there is no need to lock each item and input each item individually.

Next, the comparison means (6) consisting of steps (20) and (21) determines whether or not the update is possible (comparison step).
The value of the data item specified in prerequisite data specification (3),
The value of that item in the update data record (10) input in step (19) is compared in step (20).

If all of these are the same, then this update is considered valid and control is transferred to the data update means.

If it is different, this update is invalidated, and in step (21), an invalidation termination code "11" is set, and the application program (1) is notified that the update is invalid.

Next, an update process for updating data based on the result of the above-mentioned update availability determination will be described.

Steps (22) to (24) are update means (7), which is specified by the update data designation (3) among the update data records (10) input into the data buffer on the memory in step (22). After changing the specified data item to the specified value, the data record is output from the data buffer to the database (8) in step (23).

As for the output here, since the data record has already been input in step (19), there is no need to input it again, and it is sufficient to simply output it.

This is achieved because the database management program is written in a low-level language, which allows for minimal and flexible counders, and this also improves processing speed by eliminating unnecessary input. .

Then, in step (24), the update OK exit code is "0".
0" to notify the application program (1) that the update has been completed.

Finally, the locking means (second half) (5bl), which is the second half of the locking means (5) consisting of step (25), unlocks the data (release step).

The above operation updates the database.

In the above embodiment, the data item specified in the premise data specification (2) and the data item specified in the update data specification (3) are the same, but as shown in FIG. The data item specified by data specification (2) and the data item specified by update data specification (3) may be different.

In FIG. 5, (9) is a comparison data record, and the comparison data record (9) is a data record in the database (8), and the application program (
It is compared with the data item specified by the prerequisite data specification (2) issued from 1). For example, if the value of FILE number I RECORD number 4 item number 5 is ``3'' and you want to change the value of FILE number I RECORD number OR number l to ``8'', the comparison data record (9) is
Number I RECORD number 4, and the update data record (lO) is between FILE number I RECORD number OR, the data item specified by prerequisite data specification (2) and the data specified by update data specification (3) This is a case where the items are different.

Even in such a case, in step (19), the comparison data record (9) is locked and input in the same way as the update data record (1a), and in step (20), the comparison data record (9) is entered instead of the update data record (10). If the data record (9) is compared with the data specified in the prerequisite data specification (2), the same effect as in the above embodiment can be obtained. This shows an example. Step (26) is step (19)
This is the step of determining whether the input processing is successful. At the time of input processing in step (19), the update data record (10) or comparison data record (9) has already been created.
may have been deleted due to other requests. In this case, the input process in step (19) results in an error, making it impossible to execute subsequent processes. Therefore, processing for setting the end code "01" for input errors has been added. In this example, the determination is made after the input processing in step (19), but the determination may be made immediately after the lock processing in step (18) if an error can be detected in the lock processing in step (18).

Further, in the above embodiment, an application program that involves terminal processing has been described, but it goes without saying that the present invention can also be used for application programs that perform other processing.

Further, the name does not necessarily have to be an application program, and it may be called a process, job, task, or the like.

Further, in the above embodiment, the unit of locking is a data record, but this unit may be a larger file unit or a block unit, or may be a part of a data record, or a single data item unit.

Also, although the unit of file input/output is a database, it may be a file unit or a data block unit as long as it is the same or smaller than a lock unit, or it can be a part of a data record or a single unit. It may be a data item unit.

Furthermore, in the above embodiment, locking is performed in the order of unique numbers added to each data unit.
This order may be any other method as long as the order of locking is uniquely determined. For example, the order of files to be locked may be the order of file names, and the order of records may be the order of physical positions on the file.

Further, in the above embodiment, the case where the application program is executed is explained, but the database management device according to the present invention only needs to be equipped with a means for executing the application program, and the database management device according to the present invention only needs to be equipped with a means for executing the application program. There's no need.

Further, in the above embodiment, many parts have been explained as programs, but some or many parts may be realized by hardware or firmware.

[Effects of the Invention] As explained above, according to the present invention, the database management means (process) includes a locking means (process) for locking data, a comparison means (process) for determining whether or not data can be updated, and a data management means (process) for locking data. Since we have provided an update means (process) to update the data items in the file, there is no need to lock data items on the file for a long time, there is no unnecessary input from the database, and the possibility of deadlock is eliminated. Thus, it is possible to obtain a database management device and a database management method that can update data in a database.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of an embodiment of a database management device according to the present invention, FIG. 2 is a diagram of an embodiment of update command, prerequisite data specification, and update data specification according to the present invention, and FIG. FIG. 4 is a flowchart showing the operation of the application process that performs the data update process in this invention; FIG. An overall configuration diagram of another embodiment; FIG. 6 is a flowchart showing other operations during update processing of the database management process in the present invention; FIGS. 7 and 8 show operations of conventional database processing. It is a flowchart figure. (1) is the application program, (2) is the prerequisite data specification, (3) is the update data specification, (4) is the database management program, (5) is the locking means, (6) is the comparison means, (power is the update means) , (8) is the database, (10) is the update data record, and (11) is the end code.

Claims (2)

[Claims] (1) Database management device (a) having the following elements:
a database that stores data; (b) means for operating an application program that issues the following information in order to request updates to the data stored in the database; (b1) data that is a prerequisite for updating the data; (b2) Update data specification that specifies the updated data of the data to be updated; (c) Database management means (c1) that has the following elements:
) A locking means that takes as input the prerequisite data specification and update data specification issued by the application program and locks the data in the database; (c2) Takes as input the prerequisite data specification and the value of data in the database and compares these values. Comparing means, (c3) Updating means that receives update data designation as input and updates data in the database based on the comparison result of the comparing means. (2) Database management method having the following steps (a)
An application process that issues data update requests by inputting data from a database and issuing update prerequisite values and update data after predetermined processing; (b) a database management process (b1) that includes the following steps;
) A locking process that inputs a data update request issued by the application process, locks and inputs the data in the database, (b2) inputs the update prerequisite value and the value of the data in the database, and compares these values. comparison step; (b3) update step of inputting update data and updating data in the database based on the comparison result of the comparison means; (b4) An unlocking step for unlocking the data locked in the locking step.