JPH04165545A - Data fault recovering system - Google Patents

Abstract

PURPOSE:To improve the processing performance by duplicating and storing data to be updated and updated data in an updating object file, and executing the fault recovery processing by only utilizing selectively duplicated data. CONSTITUTION:Data to be updated and updated data are duplicated and stored on a secondary storage device 13a, and a storage position of those data to be updated and updated data is controlled by an old physical block number Q(x) and the present physical block number C(x). Therefore, in the case of executing a fault recovery processing, write of data to a different journal file is not executed, and by only utilizing selectively one of duplicated block data, the consistency of updating data is secured. In such a way, the fault recovery processing can be executed in a short time, and the processing performance can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION [Objective of the Invention (Field of Industrial Application) This invention relates to a data failure recovery method in a data processing device.

(Prior art) In general, in failure recovery (data recovery) processing in a data processing device using a parallel processing method, a data update program collects the block contents before update of the file to be updated into a journal file, and then updates the block contents in a journal file. The contents of the target block are rewritten using a predetermined work area to complete the update process.

After executing a series of update processes, the program either authenticates the updates as correct by executing a commitment instruction, or disables the updates and updates the block contents at the start of the series of update processes using role pack processing. Recover.

That is, if the update is certified as correct, the updated file is updated with the new block content updated in the work area under commitment processing, whereas if the update is recognized as invalid or If some kind of system failure occurs during update processing, the data collected in the journal file is written back to the update target file by roll pack processing.

In a data processing device that performs parallel processing, it is possible to guarantee the consistency of updated data by performing failure recovery using such commitment processing and roll pack processing.

However, in conventional data processing devices, in order to perform such commitment processing and roll pack processing, a journal file is prepared separately from the update target file.
Since it is necessary to write the block to be updated into the journal file for each update process, there is a drawback that sufficient processing performance cannot be obtained under a program where update processes occur frequently.

(Problem to be Solved by the Invention) Conventionally, in order to perform failure recovery processing such as commitment processing or roll pack processing, a journal file is prepared separately from the update target file, and the update target block is linked to the update target file. Processing such as reading data from the journal file and writing it to a journal file must be executed for each update process, which has the disadvantage that sufficient processing performance cannot be obtained in programs where update processes frequently occur.

This invention was made in view of these points. By duplicating and storing pre-update data and post-update data in an update target file, failure recovery can be achieved only by selectively using the duplicated data. enable processing to be performed,
The purpose of this invention is to provide a data failure recovery method that can improve processing performance.

[Structure of the Invention] (Means and Effects for Solving the Problems) A data failure recovery method according to the present invention is provided in a data processing device equipped with a storage device for storing a file composed of a plurality of blocks. 2 blocks on the storage device
a block management table that holds, for each logical block number indicating each block, first and second physical block numbers corresponding to physical storage positions before and after update of the block, respectively, so as to be stored in a duplicated manner; When updating the block, refer to the block management table, change the second physical block number to a value other than the corresponding first physical block number, and update the changed second physical block. and means for writing updated data to a physical storage location indicated by a number, and is characterized in that one of the blocks stored in duplicate on the storage device is used to perform data recovery processing. .

In this data failure recovery method, pre-update data and post-update data are stored in duplicate on the storage device, and the storage locations of the pre-update data and post-update data are determined by the first physical block number and the second physical block number. managed by. Therefore, when performing disaster recovery processing such as commitment processing or roll pack processing, one of the duplicated data is simply used selectively without writing data to another journal file. This makes it possible to guarantee the consistency of updated data. Therefore, since failure recovery processing can be executed in a short time, it is possible to obtain sufficient data processing performance even under a program in which update processing frequently occurs.

(Example) Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 shows the configuration of a data processing device for realizing a data failure recovery method according to an embodiment of the present invention.

This data processing device is a so-called parallel processing type computer system, and its failure recovery processing function section includes a data recovery program execution module 11. It is composed of a data update program execution module 12 and a file system 13.

The data recovery program execution module 11 performs data recovery by managing logical block numbers and physical block numbers in a management table 132 provided in the file system. The logical block number is assigned to each block constituting the file to be updated, and the physical block number indicates the actual storage location on the secondary storage device 13a of the block specified by the logical block number. .

The data update program execution module 12 issues logical block update requests, commitment requests, and role pack requests.

The file system 13 forms the core for realizing the data failure recovery process according to the present invention, and the file system 13 is provided with a management table 132 in addition to the secondary storage device 13a.

The secondary storage device 13a of the file system 13 is a physical realization area of files, and this secondary storage device 13m
, a set of physical blocks P (x) in which the block should exist is provided for a logical block number X indicating each logical block constituting the file. Each physical block P (x) is composed of two physical blocks.

One block in the physical block P (x) is used as a storage area for a pre-update data block, and the other block is used as a storage area for a post-update data block.

The management table 132 is a correspondence table between the logical block number It shows the physical storage area of the block on the secondary storage device 13a, and the current physical block number C(x) shows the physical storage area of the updated data block. Therefore, before the block update is executed, the pre-update data block and the post-update data block have the same content, so the old physical block number Q (x) and the current physical block number C (x) have the same value, In other words, Q (
x) and C(x) are both 2 of physical block P(x)
indicates one of the blocks.

In addition, when updating, the current physical block number C(x
) is changed to a value indicating the other of the two blocks of the physical block P (x).

Therefore, the respective values indicated by the old physical block number Q (x) and the current physical block number C (x) always belong to two blocks of the corresponding physical block P (x).

Next, details of the management table 132 will be explained with reference to FIG.

FIG. 2(A) shows the first table that forms part of the management table 132.
This first block correspondence table defines a logical block number X and two physical block numbers in the physical block P (X) corresponding to it.

That is, for logical block number #1, two physical blocks with physical block numbers #1 and #2 are prepared on the secondary storage device 13a, and similarly, for logical block number #2, two physical blocks with physical block numbers #1 and #2 are prepared on the secondary storage device 13a. For logical block number #3, there are two physical blocks with physical block numbers #5 and #6, and for logical block number #4, there are two physical blocks with physical block numbers #7 and #8.
For logical block number #5, two physical blocks with physical block numbers #9 and #10 are provided on the secondary storage device 13a.

FIG. 2(B) shows a second block correspondence table that forms another part of the management table 132, and this second block correspondence table includes logical block numbers X and their corresponding old physical blocks. A block number Q (x) and a current physical block number C(x) are defined. Here, as described above, the old physical block number Q (x) and the current physical block number C (x) have values that belong to two physical blocks within the corresponding physical block p (x). There is.

In this second block correspondence table, logical block number #1
, #2, #4 are before the update, and the blocks specified by the logical block numbers #3, #5 are after the update or are in the process of updating, so the logical block numbers For #1, #2, and #4, the corresponding old physical block number Q (x) and current physical block number C (
x) are the same, and for logical block numbers #3 and #5, the corresponding old physical block numbers Q(x) and current physical block numbers C(x) have different values.

FIG. 3 shows the contents of the secondary storage device 13a when the second block correspondence table is set as shown in FIG. 2(A).

Physical block #6 indicated by diagonal lines in FIG.
, #10 indicates a physical block that is being updated or has been updated.

Next, referring to the flowcharts in FIGS. 4 to 6,
A logical block update process, a commitment process, and a roll pack process in the data processing apparatus shown in FIG. 1 will be explained.

The update process for logical block X is performed in accordance with the procedure shown in the flowchart of FIG.

That is, first, the data update program execution module 12 issues an update request to the logical block to be updated, and the data recovery program execution module 11 refers to the management table 132 and updates the physical block number corresponding to the logical block X. Execute. In this case, first,
Current physical block number C(x) and old physical block number Q
The values of (x) are compared and it is determined whether they are the same (step Al).

If the values of the current physical block number C(x) and the old physical block number Q(x) corresponding to the logical block number (x) is changed to a different value, that is, a value indicating a block different from the block indicated by the old physical block number Q (x) among the two physical blocks of physical block P (x) (Step A2) . For example, in Figure 2 (A
), (B), assuming that the values of the old physical block number Q (x) and the current physical block number C (x) corresponding to logical block #3 are both #5,
The two physical blocks assigned to logical block #3 are #5 and #6, so the current physical block number C(x)
The value of is updated from #5 to #6.

Then, the contents of physical block #6 indicated by the current physical block number C(x) are updated (step A3).

The commitment process is executed according to the procedure shown in the flowchart of FIG.

That is, first, the data update program execution module 12 makes a commitment request, and the data recovery program execution module 11 refers to the management table 132 and updates the corresponding physical block number. In this case, first, a search process is performed on the management table 132 (step A11), and a logical block number X whose old physical block number Q (x) and current physical block number C (x) are different in value is detected. . Then, for each detected logical block number X, the value of the old physical block number Q (x) is changed to the value of the corresponding current physical block number C(x) (step A12).

For example, the second correspondence table of the management table 132 is shown in FIG.
If a commitment request is made in the state shown in B), as a result of the commitment process, the value of the old physical block number Q (x) corresponding to logical number #3 or from #5 Along with being changed to #6,
Old physical block number Q (x) corresponding to logical number #5
The value of is changed from #9 to #10.

The roll pack process is executed according to the procedure shown in the flowchart of FIG.

That is, first, the data update program execution module 12 makes a roll pack request, and the data recovery program execution module 11 refers to the management table 132 and updates the corresponding physical block number. In this case, first, the management table 132 is searched (
In step A21), a logical block number X in which the values of the old physical block number Q (x) and the current physical block number C(x) are different is detected. Then, for each detected logical block number X, the value of the current physical block number C(x') is changed to the value of the corresponding old physical block number Q(x,) (step A22).

For example, the second correspondence table of the management table 132 is shown in FIG.
If a roll pack request is made in the state shown in B), as a result of roll pack processing, the current physical block number C corresponding to logical number #3 is
The value of (x) is changed from #6 to #5, and the value of the current physical block number C(x) corresponding to logical number #5 is changed to #5.
Changed from 10 to #9.

As described above, in this embodiment, the pre-update data and post-update data are stored in duplicate on the secondary storage device 13a, and the storage locations of the pre-update data and post-update data are the old physical block numbers Q. (x) and the current physical block number C(x). Therefore, when performing disaster recovery processing such as commitment processing or roll pack processing, you can simply selectively write one of the duplicated block data without writing data to another journal file. By using this method, it is possible to guarantee the consistency of updated data. Therefore, since it is no longer necessary to write data to a separate journal file, failure recovery processing can be executed in a shorter time, so sufficient data processing performance can be maintained even under programs that frequently update. It is possible to obtain it.

Note that the management table 132 only needs to be able to identify the old physical number and the current physical block number from the logical-physical number, so the management table 132 may have a configuration with two correspondence tables as shown in FIGS. It is not limited.

[Effects of the Invention] As described above, according to the present invention, by duplicating and storing pre-update data and post-update data in an update target file, failures can be prevented by only selectively using the duplicated data. Recovery processing can now be executed, and processing performance can be improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a data processing device for realizing a failure recovery method according to an embodiment of the present invention;
2 is a diagram showing an example of a specific configuration of a management table provided in the data processing device shown in FIG. 1, and FIG. 3 is a diagram showing the state of data in the secondary storage device provided in the data processing device shown in FIG. 1. FIGS. 4 to 6 are flowcharts illustrating the operation of the data processing apparatus shown in FIG. 1. II...Data recovery program execution module,
12...Data update program execution module, 13
...File system, 13a...Secondary storage device,
132... Management table. Applicant's agent Patent attorney Takehiko Suzue Figure 1, IF5, tJ5, Figure 6

Claims (1)

[Scope of Claims] A data processing device including a storage device for storing a file composed of a plurality of blocks, wherein each of the blocks is stored in duplicate on the storage device. a block management table that holds, for each logical block number indicating a block, first and second physical block numbers corresponding to physical storage locations before and after the block is updated, respectively; After referring to the management table, changing the second physical block number to a value other than the corresponding first physical block number, and updating the physical storage location indicated by the changed second physical block number. 1. A data failure recovery method, comprising: means for writing data, and executing data recovery processing using one of blocks stored in duplicate on a storage device.