JP3292201B2 - Access control device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to relates to access control apparatus for preventing data loss files that occurs when a system crash of the computer system, in particular, to be able to prevent data loss without acquiring journal data And an access control device .

In a computer system, data that needs to be continued beyond system stoppage is stored in a direct organization file in a nonvolatile storage device. When updating the existing data in the direct organization file, a method of reading data from the main storage unit in physical block units, changing the contents thereof, and writing the data in physical block units is adopted. However, if a system crash occurs during data writing, the data of the physical block to which the data is being written may be incomplete. From now on, it is necessary to provide means for restoring it, but this means must be realized without imposing a large load.

[0003]

2. Description of the Related Art For example, in a storage device such as a system storage device, a CPU controls data transfer without passing through a channel device.
When U stops, data will only be transferred halfway. In a storage device such as an external storage device that passes through a channel device, such inconvenience does not occur due to the interposition of the channel device. However, in a hot standby system, in order to speed up system switching, the current system is used. When a crash occurs, an I / O reset is issued to the channel device. From now on, this I /
The non-transferred data is zero-padded by the O reset, and data different from the original data is written.

[0004] In the conventional data processing apparatus, in preparation for data loss occurring at the time of such a system crash,
For data that needs to be continued beyond the system stop, a configuration is adopted in which journal data is written to a separately prepared journal file. The method of taking measures by reading the data was adopted.

[0005] Even when data to be updated in synchronization with the progress of data processing is expanded to a direct organization file in a nonvolatile storage device, abnormal update data is stored in the direct organization file due to a system crash. Before the data in the organization file is updated, the updated data is written to the journal file in preparation for writing. The approach was to take measures by reading out the data before the occurrence.

[0006]

According to such a conventional method, when the direct organization file is a user file for storing the processing data of the user, the data that needs to be stored for a long period of time. Although this does not cause a major problem, in the case of a checkpoint file that stores data that is updated directly in synchronization with the progress of data processing, save temporary data. Therefore, there is a problem that the following adverse effects occur. That is, the overhead for acquiring journal data is large.

[0007] A journal file for acquiring journal data is required, and backup and other operations are difficult. When a system crash occurs, a utility program is required to input journal data and directly recover the organization file.

The present invention was made in view of such circumstances, without obtaining the journal data, provides a new access control system that allows preventing data loss files that occur when the system crashes It is intended for.

[0009]

FIG. 1 shows the principle configuration of the present invention. In the figure, 1 is a data processing device equipped with the present invention, 2 is a direct organization file constructed in a directly accessible non-volatile storage device, 3 is an access processing unit that executes access processing to the direct organization file 2, 4 is This is a higher-level function that issues an access request to the access processing unit 3.

The direct organization file 2 of the present invention develops two physical blocks 5-1 and 2 that are developed in a continuous or discontinuous manner. These two physical blocks 5-1 and 2 are treated as one logical block 6 for the upper function 4 according to the function of the access processing unit 3.

The access processing unit 3 of the present invention controls the two physical blocks 5-1 and 2 developed on the direct organization file 2 so as to treat the higher-level function 4 as one logical block 6. When there is a write request to the logical block 6 from the unit 8 and the upper function 4 to the two-block control unit 8, the physical blocks 5-1 and 2 are alternately selected, and the selected physical block 5--2 is selected. For example, a write control unit 9 that writes data to the first and second parts and writes generation information 7 to the first and last parts of the data, When there is a read request, the physical block 5 that records the latest valid data according to the generation information 7 managed by the physical blocks 5-1 and 2
-1,2, and a read control unit 10 for reading data recorded in the specified physical blocks 5-1 and 2 are adopted.

[0012]

According to the present invention, when there is a write request of data to be updated in synchronization with the progress of data processing from the upper function 4 to the two-block control unit 8, the write control unit 9 executes the previous writing. The physical block 5-1 or 2 which is not the physical block 5-1 or 2 is selected, data is written to the selected physical block 5-1 or 2, and the data is added to the beginning and end of the data. The generation information 7 is written. In this way, if no system crash has occurred, 2
The data of the latest generation is written into one of the physical blocks 5-1 and 2 while the data of the previous generation is written into the other.

On the other hand, when there is a request for reading the latest generation data written in the physical blocks 5-1 and 2 to the two-block control unit 8, the read control unit 10 It is detected whether or not the generation information 7 written in the head part and the end part of the 5-1 and 2 match, and when the generation information 7 of one of the physical blocks 5-1 and 2 does not match, Reads out the data of the matching physical block 5-1, 2 and notifies the upper-level function 4 of the data.
When the generation information 7 is the same as the generation information 7, the data of the physical blocks 5-1 and 2 having the latest generation information 7 is read and notified to the upper function 4.

According to the processing of the access processing unit 3, when the latest generation of data is being written to the physical blocks 5-1 and 2, a system crash occurs, and the latter half of the latest generation of data has a zero value. However, at the time of recovery processing after the occurrence of a system crash, the read control unit 10 determines that the data is invalid data, and records 1 in the other physical block 5-1 or 2. The data before the generation is read out and notified to the upper function 4. As a result, the upper-level function 4 can resume data processing using the data of the previous generation.

[0015] Thus, according to the present invention, without obtaining the journal data is to become a data loss of files, such as direct file 2 can be prevented which occurs if the system crashes.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail with reference to embodiments. FIG. 2 illustrates an embodiment of the present invention. In the figure, FIG.
The same components as those described in (1) are indicated by the same symbols, 11 is a user program, 12 is a checkpoint acquisition processing unit, and 13 is a nonvolatile storage device.

This embodiment discloses an embodiment of the present invention in a checkpoint acquisition process in a transaction process. The user program 11 first issues a transaction start instruction, and then issues a database process. Is executed, and when a transaction end instruction is issued last, the checkpoint acquisition processing unit 12
Synchronously registers the state data of “active state”, “in-doubt state”, “commit state”, and “invalid state” in the logical block 6 via the access processing unit 3 in synchronization with the processing progress. By doing so, during recovery processing after a system crash, it is possible to know which processing should be restarted.

FIG. 3 shows an example of a processing procedure that the access processing unit 3 executes upon receiving the update request of the checkpoint acquisition processing unit 12. Where
One of the two physical blocks 5-1 and 2 is represented by "P1", and the other is represented by "P2". As described with reference to FIG. 1, the access processing unit 3 synchronizes with the progress of the data processing from the checkpoint acquisition processing unit 12, which is the higher-level function 4, and updates the state data registered in the logical block 6. If there is, the physical block 5-1 or 2 which is not the physical block 5-1 or 2 written previously is selected, the corresponding state data is written to the selected physical block 5-1 or 2, and the state is written. Generation information 7 consisting of a serial number is written at the beginning and end of the data. In this way, when a system crash has not occurred, the state data of the latest generation is written into one of the two physical blocks 5-1 and 2 and the other is one generation earlier. State data will be written.

Thus, as shown in FIG.
The state data of the latest generation (generation information 7 “N”) is written to the physical block P1, and 1 is written to the physical block P2.
When the state data of the previous generation (generation information 7 of “N−1”) is being written, the checkpoint acquisition processing unit 1
When there is an update request from the second state data to the next state data, the access processing unit 3 writes the state data in the physical block P2, and generates the generation “N + 1” at the beginning and end of the state data. Information 7 is written. If a system crash does not occur during the writing process, as shown in the flow on the left side of the drawing, the state data of the previous generation (generation information 7 of “N”) is written into the physical block P1 and The state data of the latest generation (generation information 7 of “N + 1”) is written to the physical block P2. On the other hand, if a system crash occurs during this writing process, as shown in the flow on the right side of the drawing, the state data of the previous generation (generation information 7 of “N”) is written to the physical block P1. At the same time, abnormal state data in which the latter half indicates a zero value is written to the physical block P2.

On the other hand, as described with reference to FIG. 1, the access processing unit 3 performs recovery processing after a system crash.
When there is a request to read the latest state data written in the logical block 6 from the user program 11 which is the higher-level function 4, first, the data is written into the head part and the last part of each of the physical blocks 5-1 and 2. It is detected whether or not the generation information 7 of the physical blocks 5-1 and 7 match, and if the generation information 7 of one of the physical blocks 5-1 and 2 does not match, Read data and user program 11
If the generation information 7 is the same for both physical blocks 5-1 and 2, the data of the physical blocks 5-1 and 2 having the latest generation information 7 is read and notified to the user program 11. To go.

In this way, as shown in FIG.
When passing through the flow on the left side in the figure, the access processing unit 3 reads out the state data of the latest generation (generation information 7 of “N + 1”) written in the physical block P2 and notifies the user program 11 of the state data. When the data passes through the flow on the right side in the figure, the state data one generation before (generation information 7 "N") written to the physical block P1 is read out and notified to the user program 11.

According to the processing of the access processing unit 3, even if a system crash occurs during the update of the checkpoint, the checkpoint one generation before is valid, and the transaction processing can be resumed from the checkpoint. It becomes.

[0023]

As described above, according to the present invention,
Without obtaining the journal data is to become the de <br/> over data loss files such organized directly files can be prevented which occurs if the system crashes.

[Brief description of the drawings]

FIG. 1 is a principle configuration diagram of the present invention.

FIG. 2 is an embodiment of the present invention.

FIG. 3 is an explanatory diagram of a processing procedure executed by an access processing unit.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Data processing apparatus 2 Directly organized file 3 Access processing part 4 Upper function 5 Physical block 6 Logical block 7 Generation information 8 2 Block control part 9 Write control part 10 Read control part

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-2-35538 (JP, A) JP-A-3-27428 (JP, A) JP-A-2-186453 (JP, A) JP-A-62-162 256054 (JP, A) JP-A-63-228348 (JP, A) JP-A-62-239248 (JP, A) JP-A-2-143343 (JP, A) (58) Fields investigated (Int. ⁷ , DB name) G06F 12/00 531

Claims (2)

(57) [Claims] The method according to claim 1 Two physical blocks that are deployed in a continuous or discontinuous manner on a Tsunofu Airu, and 2 block control unit for controlling so treated as one logical block to the upper function, from the upper function When there is a write request to the logical block to the two-block control unit, the physical block is alternately selected, data is written to the selected physical block, and the data is stored in the physical block. A write control unit that writes generation information in association with the above, and when a higher-level function requests the two-block control unit to read from the logical block, records valid and latest data according to the generation information. A read control unit that specifies the physical block and reads data recorded by the specified physical block. Access control device. 2. A continuous or discontinuous state on one file
Two physical blocks deployed in the same way
Block that is controlled to be treated as one logical block
Block control unit and the logical block from the higher-level function to the two-block control unit.
When there is a write request to the lock, the above physical block is
Alternately select the data in this selected physical block.
Data at the beginning and end of the data
A write control unit for writing generation information to the two-block control unit from a higher-level function.
When there is a read request from the lock, the
When the generation information matches the last generation information,
Determined that the physical block records valid data
While valid and up-to-date data
Identify the physical block that records the
Read control for reading data recorded in the physical block
Further comprising a section, the access control device according to claim.