JPS592151A - Check point recovering system - Google Patents

Abstract

PURPOSE:To shorten the time required for obtain-processing of a check point and its recovery processing, by providing a check word before and after a control table. CONSTITUTION:In case of executing the recovery processing, first of all, one of tables on a bulk core memory 3 is read, and a check word 1 is compared with a check word 2. As a result of comparison, unless both of them are equal to each other, contents of a shunting area on the memory 3 are read, and contents which are read in a corresponding table area on the memory 3 are written. Subsequently, an address of the following table existing on the memory 3 is derived, the same processing is repeated, and the processing concerned is executed as to all the control tables.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a checkpoint recovery method using bulk core memory.

[Prior art and problems]

In conventional online systems, even if the system 9 goes down, a checkpoint is acquired in the DEAD file before updating information that requires recovery, in order to guarantee the messages that were being processed up to that point. The main memory table is restored from this checkpoint file during the recovery process after the system goes down. In this type of conventional method, it takes a considerable amount of time to access a file to obtain a checkpoint when a transaction occurs. Furthermore, since the access time is long during recovery, the entire recovery time becomes long.

[Purpose of the invention]

The present invention eliminates the drawbacks of the prior art, reduces the time required for checkpoint and recovery, and provides
Its purpose is to provide a checkpoint recovery method that facilitates program development.

[Structure of the invention]

Therefore, the checkpoint recovery method of the present invention provides a checkpoint recovery method that stores check words before and after a management table in an information processing system that includes a central processing unit, a main memory, a bulk core memory, and an input/output device. In addition, the management table is stored in the bulk core memory, and when the management table is updated in line 5, the management table on the bulk core memory is read into the main memory, and the read contents are stored in the bulk core memory. Update the part to be updated by writing to the save area in the memory, add 1 to each of the previous and following check words, and then write the updated information to the corresponding management table in the bulk core memory. When performing recovery processing, the management table in the bulk core memory is read into main memory, and the contents including whether the values of the previous and subsequent checkwords are equal or not are written to the corresponding management table area in the bulk core memory. Write, find the address of the next management table on the bulk core memory, then check whether all management tables on the bulk core memory have been updated, and if not, perform the above process. It is characterized by repeating.

[Embodiments of the invention]

Hereinafter, the present invention will be explained with reference to the drawings.

FIG. 1 is a diagram showing an example of an information processing system to which the present invention is applied, FIG. 2 is a diagram showing an example of a control table, and FIG. 3 is a diagram showing an example of a ROM (abbreviation for bulk core memory) table. FIG. 4 is a diagram explaining the recovery process.

In FIG. 1, 1 is a central processing unit, 2 is a main memory, 3 is a bulk core memory, 4 is an auxiliary storage device, 5 is a communication control device, 6 is a typewriter, and 7 is a common bus. .

The bus V core memory 3 uses 18M metal as a storage element, and can mount up to two 128-byte memory stacks.

Bulk core mesenshu, - Fast transfer speed - ・Over-two-over does not occur.

- The same software can be used to access magnetic drum devices.

- Program access is possible in 1119 (16 bits) or 1 byte units.

・Data transfer between bulk core memory and input/output devices (magnetic disk devices, etc.) is possible.

・High reliability as there are no mechanical parts.

, the memory contents are preserved in the event of a power outage or instantaneous interruption.

It has characteristics such as. Bulk core memory 3 is
It also stores received file management information of the auxiliary storage device 4. The auxiliary storage device 4 stores received files for each line, and the communication control device 5 accommodates a plurality of lines.

FIG. 2 shows an example of a table stored in the bulk core memory 3. Check word 1 is written in the first entry of the ROM table, and check 7=do 2 is written in the last entry. The multiple entries between the first check word 1 and the last check word 20 include the status, number of empty records, number of unsent messages, number of records being sent, write pointer, read pointer, received file start address, and number of received file records. etc. are entered.

FIG. 3 explains the BOM table update process.

■ Read the Todoroki table on the bulk core memory 3 into the main memory.

■ Write the read contents to the save area on bulk core memory 3.

■ Locations to be updated (e.g. write pointer in Figure 2)
Update.

■ Add 1 to each of checkword 1 and checkword 2. Note that, at the time of initial setting, check word 1 and check word 2 are both zero.

■ Write the updated information to the table on bulk core memory 3.

If a failure occurs during the transfer of updated information from main memory 2 to bulk core memory 3, checkword 1 and checkword 2 of the table in bulk core memory 3
will have different values.

FIG. 4 is a diagram illustrating recovery processing in the present invention.

Read one table on C bulk core memory 3.

■' Compare check word 1 and check word 2.

■′ Check whether they are equal or not. In the case of YO2, the process ``■'' is performed, and in the case of Nτ, the process C is performed.

■' Read the contents of the save area in bulk core memory 3.

σ The contents read by da are loaded into the corresponding table area on the bulk core memory 3 by 1.

Find the address of the next table in bulk core memory.

■′ Check whether the table is finished or not. In the case of N million, return to the process of ■.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, the time required for checkpoint acquisition processing and recovery processing can be shortened, and since the recovery program is simple, depacking time can be reduced. It can produce effects.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an example of an information processing system to which the present invention is applied, FIG. 2 is a diagram showing an example of a control table, and FIG. 3 is a diagram showing an example of a BOM (abbreviation for bulk core memory) table. FIG. 4 is a diagram explaining the recovery process. DESCRIPTION OF SYMBOLS 1... Central processing unit, 2... Main memory, 3... Bulk core memory, 4... Auxiliary storage device, 5... Communication control device, 6... Typewriter, 7...・Common bus. Patent Applicant: Fujitsu Ltd. Representative Patent Attorney: Kyotani Yotsube Saizu1

Claims (1)

[Claims] A central processing unit, a main memory, a bulk core memory,
In an information processing system equipped with an input/output device, checks are performed before and after the management table. - When the management table is stored in the bulk core memory and the management table is updated in line 5,
The contents of the management table in the bulk core memory are read into the main memory and written to the save area in the bulk core memory.
Update the parts that should be updated, add 1 to each of the previous and following check words, and then bulk update the updated information.
When performing recovery processing by writing to the large management table in the core memory, the management table in the bulk core memory is read into the main memory, it is checked whether the values of the previous checkword are equal, and both If they are not equal, the contents of the corresponding save area in the bulk core memory are read into the main memory, the read contents are written to the corresponding management table area in the bulk core memory, and the contents are stored in the bulk core memory.・It is characterized by finding the address of the next management table in memory, and then checking whether update processing has been performed for all management tables in bulk core memory, and repeating the above processing if not. checkpoint recovery method.