JPH0259837A - Data recovery processing system - Google Patents

Abstract

PURPOSE:To restore the data to the latest one at a high speed without deteriorating the data processing performance of a system by saving the data of a state set right before the occurrence of the system abnormality into a nonvolatile memory as the data for recovery only when the system abnormality occurs. CONSTITUTION:A system abnormality detecting part 5 monitors a system and detects the system abnormality if occurs to inform this to a saving mechanism part 3. Then the part 5 gives an instruction for saving the restored data. The part 3 saves the data on a shared buffer 21 and a control information part 22 into a memory dump file 7 out of those data stored in a volatile memory 2 as the data for recovery. A restoring mechanism part 4 uses the recovery data stored in the file 7 and restores data into a file 8 via the buffer 21. Thus it is possible to obtain a data recovery processing system where the data can be restored to the latest one at a high speed without deteriorating the data processing performance of the system.

Description

[Detailed Description of the Invention] [Summary] A data recovery processing method for recovering data in the event of system abnormality in a data processing system enables high-speed data recovery of the latest data without degrading the data processing performance of the system. The purpose is to.

In a data processing system including a data processing unit, a first nonvolatile storage area, and a volatile memory, an evacuation mechanism unit that saves data, a recovery mechanism unit that recovers data, and detects a system abnormality. The system abnormality detection section includes a system abnormality detection section and a second non-volatile storage area, and the system abnormality detection section detects a system abnormality and instructs the evacuation mechanism section to save data, and the evacuation mechanism section stores data in the volatile memory. The recovery mechanism section saves the data to the second non-volatile storage area, and the recovery mechanism section saves the data to the second non-volatile storage area.
To recover data to non-volatile storage area, use dilute acid.

[Industrial application field]

The present invention relates to a data recovery processing method, and more particularly, to a data recovery processing method for recovering data when a system abnormality occurs in a data processing system.

With the increase in the amount of data handled by data processing systems such as computer systems in recent years, there is a need to shorten the processing time as well as the time required for data recovery processing in the event of a system abnormality.

[Conventional technology]

Conventional data recovery processing was performed as follows.

During data processing such as when updating data, recovery data for the data is logged in a nonvolatile storage device such as a magnetic disk device. Then, when a system abnormality occurs, data is restored using logging data (recovery data).

That is, the data is returned to the state before the update, for example.

[Problem to be solved by the invention]

According to the above-mentioned conventional technology, in order to recover data, it is necessary to save the original data to a magnetic disk device or the like as recovery data each time data is updated. Saving this data takes a lot of time.

Therefore, there is a problem in that saving data during data processing degrades the data processing performance of the system. This problem is particularly noticeable in small-scale systems with low data processing performance. For this reason, it is virtually impossible to perform data recovery processing as described above in a small-scale system, and even if it were adopted, data processing performance would have to be sacrificed.

An object of the present invention is to provide a data recovery processing method that enables high-speed data recovery of the latest data without reducing the data processing performance of the system.

[Means for Solving the Problems] FIG. 1 is a diagram showing the basic structure of the present invention, and shows a data processing system according to the present invention.

In FIG. 1, ■ is a data processing section including a central processing unit (CPU) and memory, 2 is a volatile memory as the memory of the data processing section l, 3 is an evacuation mechanism section, 4 is a recovery mechanism section, and 5 is a data processing section including a central processing unit (CPU) and memory. System abnormality detection unit.

6 is a non-volatile memory such as a magnetic disk device, 7 is a memory dump file, 8 is a file, 21 is a shared buffer,
22 is a management information section.

The volatile memory 2 consists of a semiconductor storage device, for example,
It is a volatile storage area. The shared buffer 21 is an area for storing data used by the data processing unit 1, and is an area for temporarily storing data input/output to the file 8. The management information section 22 is an area that stores various management information for managing the data of the file 8.

File 8 is a (first) non-volatile storage area,
Store non-volatile data. This data is to be restored and is also to be subjected to normal data processing in the data processing section 1.

The memory dump file 7 is for saving and storing recovery data in the event of a system abnormality, and is provided in the nonvolatile memory 6 for this purpose.

This is the (second) non-volatile storage area.

The system abnormality detection unit 5 monitors the system.

When a system abnormality occurs, it is detected, the evacuation mechanism section 3 is notified, and an instruction is given to evacuation of the recovered data.

The evacuation mechanism unit 3 saves the shared buffer 21 and the management information unit 2 from among the data in the r&generic memory 2 as recovery data.
2 data is saved to the memory dump file 7.

The recovery mechanism unit 4 uses the recovery data in the memory dump file 7 to restore the file 8 via the shared buffer 21.
Recover data within.

[Operation] The data processing unit 1 reads data actually used in normal data processing from the file 8 to the shared buffer 21, processes (updates) this data, and updates the management information in the management information unit 22 along with this update. is updated to indicate the latest state of the data, and then the data is written from the shared buffer 21 to the file 8.

Therefore, the contents existing in the shared buffer 21 and the management information section 22 are saved only in the event of a system error.
There is no need to save data each time during normal data processing (or data updating).

Data recovery processing is possible.

When a system abnormality occurs, the evacuation mechanism section 3 saves the data in the shared buffer 21 and the management information section 22 as recovery data to the non-volatile memory dump file 7 according to instructions from the system anomaly detection section 5 that has detected this. (1st
Dotted line ■) in the figure.

At a predetermined restart point after a system abnormality occurs.

The recovery mechanism unit 4 investigates the contents of the memory dump file 7 and performs data recovery as necessary.

That is, the recovery mechanism section 4 first reads the contents of the memory dump file 7 (recovery data) into the shared buffer 21 and the management information section 22, and uses this to recover the data into the file 8 (as shown in FIG. 1). Dotted line ■).

As described above, during normal data processing (state where no system abnormality has occurred), there is no need to save recovery data, so high-speed data processing is possible. Further, when a system abnormality occurs, the data in the state immediately before the abnormality occurs is saved, so that data recovery up to the relevant state can be performed at high speed.

〔Example〕

FIG. 2 is an explanatory diagram of an embodiment, and the volatile memory 2 of FIG.
and the configuration of the nonvolatile memory 6.

In this embodiment, the file 8 is a physical file section for storing user data (records).

It consists of an index section for searching data in the physical file section. Physical file part and index part are 2
For example, multiple pages divided into 2 kilohide units and 5
It consists of a management information area that manages the number of data items, etc. Reading and writing of the physical file section and index section are performed in page units.

Therefore, in response to this, the shared buffer 21 is also provided.

It is partitioned into pages of 2 kilobytes. Further, the management information section 22 is composed of a buffer management information section and a file management information section. The former stores buffer management information that manages the relationship between individual pages of the shared buffer 21 and pages of the physical file section and index section. The latter stores file management information that manages the states of the physical file section and the index section. These are also divided into areas of a predetermined size.

Next, an outline of the operation of the data processing system having the configuration shown in FIGS. 1 and 2 will be explained.

(1) During normal data processing The data processing section 1 reads out a plurality of pages of the physical file section and the index section to the corresponding pages of the shared buffer 21 as necessary. Then, the contents of the page are updated on the shared buffer 21 (data processing is performed).

The updated page is again written to multiple pages in the physical file section and index section.

Page reading (input from file 8) is performed as needed, and page writing (writing back) is performed asynchronously with updating in order to improve performance. It is done in batches to some extent to reduce the number of cases.

Buffer management information is updated as pages are read and written, and file management information is updated as page contents change.

Note that file management information can be read, for example, from file 8.
The writing of file management information is performed asynchronously with the closing of file 8.

(2) When a system error occurs A saving process as shown in FIG. 3(A) is performed.

A system abnormality detection unit 5 detects an abnormality in hardware or software. In response to this, the evacuation mechanism section 3 saves the shared buffer 21. All the contents of the buffer management information section and the file management information section are saved to the memory dump file 7.

After this, the system will stop.

In addition, in FIG. 1, the data is shown to be saved via the saving mechanism section 3 (dotted line ■).

Needless to say, data is transferred from the volatile memory 2, which is the main memory, to the memory dump file 7, which is the external memory, under the control of the save mechanism section 3.

(3) At the time of data recovery Restoration processing as shown in FIG. 3(B) is performed.

After restarting the system, the recovery mechanism section 4.

Based on the buffer management information of the memory dump file 7, the updated pages of the shared buffer 21 of the memory dump file 7 are written (written back) to the physical file section and the index section. As a result, the physical file section and index section (pages) are recovered. In addition, the recovery mechanism section 4
Based on the file management information of the memory dump file 7, the contents of the management information area of the physical file section and index section are updated to the latest state (before the abnormality occurs) by 0 or more,
File 8 is recovered to the state immediately before the abnormality occurred.

Note that the recovery data saved in the memory dump file 7 is actually read out to the second volatile memory 2 and then used for recovery of the file 8 by referring to the volatile memory 2.

Furthermore, in FIG. 1, data recovery is illustrated as going through the recovery mechanism section 4 (dotted line ■).

Data is transferred from the memory dump file 7 to the file 8 via the volatile memory 2 under the control of the recovery mechanism section 4.

8 is a file, 21 is a shared buffer, and 22 is a management information section.

〔Effect of the invention〕

As explained above, according to the present invention, in data recovery processing, only when a system abnormality occurs, the data in the immediately previous state is saved in the nonvolatile memory as recovery data, thereby improving the data processing performance of the system. It is possible to perform high-speed data recovery to the latest data without degrading the performance, and this greatly contributes to the increase in double-cost performance, especially in small-scale systems.

Claims (1)

[Scope of Claims] A data processing unit (1) that performs data processing; a first nonvolatile storage area (8); and a computer that stores data read from the first nonvolatile storage area (8) and In a data processing system comprising a volatile memory (2) used for data processing, an evacuation mechanism section (3) that saves data; a recovery mechanism section (4) that recovers data; The system abnormality detection unit (5) includes a system abnormality detection unit (5) that detects an abnormality, and a second nonvolatile storage area (7), and the system abnormality detection unit (5) that detects a system abnormality.
instructs the evacuation mechanism section (3) to save data, and in response to the instruction, the evacuation mechanism section (3) saves the data in the volatile memory (2) to the second nonvolatile storage area (7).
), and the recovery mechanism unit (4) restores the data to the first non-volatile storage area (8) based on the data saved to the second non-volatile storage area (7). A data recovery processing method characterized by: