JP2880828B2 - Message recovery method in case of system failure - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a message recovery method in the event of a system failure in an online system using a computer.

[0002]

2. Description of the Related Art At present, online systems using computers are used in various fields. For example, in a financial business, a host computer for managing customer information at a head office is provided with a terminal device (cache) at each branch. Dispensers, etc.) are online and frequently used for checking the balance of deposits and savings, depositing cash, and refunding.

[0003] Here, in the computer on the terminal side and the host side, one element for performing accurate processing is as follows.
It is important that the input message data from the terminal or the output message data from the host be transmitted to the other computer without error. In the event of a failure such as a power failure, the message data related to uncompleted processing is accurately and quickly recovered. It needs to be given to the computer to be reached.

FIG. 3 shows a general message recovery method in an online system.
Reference numeral 1 denotes a terminal device serving as a cash dispenser or the like, and 12 denotes a business device serving as a host computer. The host computer 12 includes a database 13 for storing all kinds of customer information such as the balance of deposits and maturity of each customer.

[0005] In other words, for example, in the case of performing a refund of deposits and savings, first, customer identification information and refund amount information input at the terminal device 11 are sent to the business device 12 as input message data. Then, the business device 12
It reads customer information corresponding to the customer identification information from the database 13 and performs balance confirmation processing for the billed amount, rewriting processing of the balance data, and the like. Payment information permitted according to the business processing is output message data as terminal message data. 11 and a refund to the customer is made.

Here, the input message data (IMJ) from the terminal device 11, the output message data (OMJ) from the business device 12, and the rewriting data (AIJ) of the database 13 in the business device 12 are the history information. As time-series journal files 14
When a failure occurs, the system is restored based on the file data held in the journal file 14.

That is, the file data held in the journal file 14 is sequentially read from the head thereof, and data relating to the incomplete processing, that is, the file data immediately before the occurrence of the failure is searched. Failure recovery is performed by sequentially executing business processing, terminal processing, or data rewriting based on the last file data.

[0008]

However, in the above-described conventional failure recovery method, the input / output message data (IM) stored and retained in the journal file 14 in time series.
J, OMJ) and the rewriting data (A
The history data consisting of IJ) is read from the oldest top data, and whether or not the data is completed data is identified for each of the read data. Processing or database rewriting processing is executed, and then the next file data is read out. Therefore, if an enormous amount of file data is held, it takes a long time to read the incomplete data. However, there is a problem in that the input message data (IMJ) and the output message data (OMJ) immediately before the occurrence of the fault are recovered and a large delay occurs in returning to the normal online operation.

[0009] The present invention has been made in view of the above problems,
Even when an enormous amount of journal file data is held, a message at the time of a system failure that enables fast and efficient recovery of input / output message data without repeating reading and processing for each file data The purpose is to provide a recovery method.

[0010]

That is, a message recovery method in the event of a system failure according to the present invention provides an on-line system having a terminal computer and a host computer, in which the input message data from the terminal computer to the host computer and the host computer. And a journal file for sequentially recording the output message data to the terminal computer and the host processing data in the host computer in a time-series manner, and when a failure occurs, sequentially read the data from the journal file to perform processing prior to the occurrence of the failure. A processing / unprocessed determination unit for determining whether or not the data read from the journal file is unprocessed data; Journal data classifying means for classifying logical data as the above-mentioned input or output message data or host processing data; input message data classified by the journal data classification means to a host computer; and output message data to a terminal computer. It comprises a message restoring means and a host restoring means for providing host processing data classified by the journal data classifying means to a host computer.

[0011]

[Operation] In other words, recovery according to the classification of the journal file
The sharing of the old function becomes possible, and the message can be restored quickly and efficiently.

[0012]

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows the configuration of an online system using a message recovery system in the event of a system failure. In FIG. 1, reference numeral 21 denotes a terminal management unit, for example, as a cache dispenser, and 22, a business process as a host computer, for example. And a terminal management unit 21 and a business processing unit 2
2, a message management unit 23 that manages input / output processing of message data exchanged with each other, a recovery management unit 24 that performs recovery processing when a system failure occurs, and input / output message data (IMJ, OMJ). ) And a journal file 25 that stores and holds the rewriting data of the file system and the database provided in the business processing unit 22 in a time-series manner as the history of the system processing.

The recovery manager 24 includes a system state manager 26, a system recovery manager 27, a message recovery manager 28, and a message recovery file 28 which are sequentially connected between the journal file 25 and the message manager 23. A management unit 29 is provided. Further, the system recovery management unit 27 includes a file system recovery management unit 30 and a database recovery management unit 3.
1, a system status log file 32 is connected to the system status management unit 26, and a message recovery file 33 is connected to the message recovery file management unit 29.

The system recovery management unit 27 reads out file data stored and held in the journal file 25 in advance when a system failure occurs, for example, when the business processing unit 22 is temporarily stopped due to a power failure, and the read out file is read. It is determined whether the data is processed completed data or uncompleted data. If the data is incomplete data, whether the type of the file data is file system rewriting data, database rewriting data, or input / output message data (IMJ, OMJ) The file rewrite data is stored in the file system recovery management unit 30, the database rewrite data is stored in the database recovery management unit 31, and the input / output message data (IMJ, OM)
J) is given to the message recovery management unit 28.

The message recovery management unit 28 selects whether the message data provided from the system recovery management unit 27 is input message data (IMJ) or output message data (OMJ). The output message data (OMJ) is provided to the processing unit 22 and the message recovery file management unit 29 is provided.

Message recovery file management unit 29
Monitors whether there is an input request for output message data (OMJ) obtained from the terminal management unit 21 in response to the failure recovery. If there is no input request for this message data, the message recovery management unit The output message data (OMJ) given from 28 is temporarily saved and held in the message recovery file 33.
When there is the message input request, the output message data (OMJ) is given to the terminal management unit 21 through the message management unit 23.

On the other hand, the file system recovery management section 30 recovers the file system in accordance with the file rewrite data provided from the system recovery management section 27. Data is business processing unit 2
2 to the file system.

The database recovery management unit 3
Numeral 1 is for recovering the database in accordance with the database rewrite data provided from the system recovery management unit 27. The database recovery data from the database recovery management unit 31 is provided to the database through the business processing unit 22.

Further, the system status management unit 26 manages the operation status of each unit of the online system by sequentially storing it in a system status log file 32. The system status data stored in the system status log file 32 is: The data is read out in response to the recovery from the failure, and is selectively given to each part of the system as state recovery data immediately before the failure. Next, the operation of the online system including the recovery management unit 24 having the above configuration will be described.

First, when the customer gives a refund request to the terminal management unit 21 in a normal state of the system, customer identification information pre-recorded on the cash card,
And the amount requested for reimbursement are input message data (IM
J) as the business processing unit 2 through the message management unit 23
Sent to 2.

Then, in accordance with the input message data (IMJ) given, the business processing unit 22 reads out customer account information corresponding to the customer identification information from the database, executes balance confirmation processing for the billed amount, and rewrites the balance data. The payment information permitted in accordance with this business process is sent to the terminal management unit 21 through the message management unit 23 as output message data (OMJ). As a result, the terminal management unit 21 repays the customer according to the given output message data (OMJ).

In this case, the input message data (IMJ) from the terminal management unit 21, the output message data (OMJ) from the business processing unit 22, and the rewriting data of the file system or the database in the business processing unit 22 are respectively All are written in the journal file 25 in chronological order. Further, the system operation state according to the customer transaction is sequentially recorded in the system state log file 32 through the system state management unit 26.

Here, when the above-mentioned online system is normal, the system recovery manager 27, message recovery manager 28, message recovery file manager 29, file system recovery manager 30, and database recovery manager 31 all function. Will not do. FIG. 2 is a flowchart showing a message data recovery process when a failure occurs in the online system.

When a system failure occurs, for example, when the business processing unit 22 is temporarily stopped due to a power failure, the file data stored and held in the journal file 25 in advance is read out by the system recovery management unit 27. It is determined whether the read file data is processed completed data or uncompleted data (steps S1 and S2).

If the file data read from the journal file 25 is determined to be processed completed data before the time of the system failure in the system recovery management unit 27, new file data is repeatedly read out. If the data is determined to be incomplete data, the type of the file data is classified as file system rewrite data, database rewrite data, or input / output message data (IMJ, OMJ) (step S3).

Here, the file data read from the journal file 25 is incomplete data, and the file data is input or output message data (IMJ
Or OMJ), the message data (IMJ or OMJ) is given to the message recovery management unit 28 (step S3 → S4).

Then, the message data provided from the system recovery management unit 27 to the message recovery management unit 28 is selected from input message data (IMJ) and output message data (OMJ). , If the input message data (IMJ) is selected, the input message data (IMJ) is given to the business processing unit 22 (steps S5 → S6). As a result, the job processing unit 22 executes an uncompleted job process corresponding to the time of occurrence of a failure, thereby achieving recovery.

The message recovery management unit 28
In the case where the message data provided from the system recovery management unit 27 is selected as the output message data (OMJ), the output message data (OMJ)
J) is temporarily saved in the message recovery file 33 from the message recovery file management unit 29, and is then given to the terminal management unit 21 through the message management unit 23 (steps S5 → S7). Thereby, the terminal management unit 21
In this case, an uncompleted customer response process corresponding to the time of occurrence of a failure is executed and recovery is achieved.

On the other hand, if the file data read from the journal file 25 is classified as incomplete data by the system recovery management unit 27 and the file data is classified as rewrite data of the file system or the database, the rewrite is performed. The data is given to the file system recovery management unit 30 or the database recovery management unit 31 (step S3 → S8 or S
9).

Then, the file system recovery management section 30 recovers the file system in accordance with the file rewrite data given from the system recovery management section 27.
File system recovery data from 0
To the file system, and a file corresponding to the time of the system failure is restored.

The database recovery management unit 3
Numeral 1 is for recovering the database in accordance with the database rewrite data provided from the system recovery management unit 27. The database recovery data from the database recovery management unit 31 is provided to the database through the business processing unit 22, Recovery of data corresponding to the time of the failure is achieved.

Therefore, the recovery management unit 2 having the above configuration
According to the online system provided with the system file 4, in the event of a system failure, the file data previously recorded and held in the journal file 25 is read out to the system recovery management unit 27, it is determined whether or not the data is processed data. (IMJ, OMJ) and the rewriting data of the file system or the database are separated, and the separated message data is further input to the input message data (IM
J) and output message data (OMJ)
The input message data (IMJ) is sent to the business processing unit 22 to execute the recovery of the business process corresponding to the occurrence of the failure, and the output message data (OMJ) is transmitted to the terminal management unit 21 to perform the recovery of the terminal processing corresponding to the occurrence of the failure. With this configuration, it is not necessary to sequentially perform a serial recovery process for each piece of read data from the journal file 25, and it is possible to share a recovery function according to the classification of the journal file. You will be able to recover.

[0034]

As described above, according to the present invention, in an online system including a terminal computer and a host computer, input message data from the terminal computer to the host computer, output message data from the host computer to the terminal computer, and A journal file that records host-processed data in a host computer in a time-series manner, and when a failure occurs, data from the journal file is sequentially read to determine whether the data is processed data related to processing before the failure occurred Processing / non-processing determining means, and when the processing / non-processing determining means determines that the data read from the journal file is unprocessed data, the unprocessed data is the input or output message data. Journal data classification means for classifying the data as input data or host processing data; message recovery means for providing input message data classified by the journal data classification means to a host computer; and output message data to a terminal computer; And a host recovery means for providing host processing data classified by the host computer to the host computer, even if an enormous amount of journal file data is held, without repeating reading and recovery processing for each file data. , Journal file classification
, And the input / output message data can be restored quickly and efficiently.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an online system using a message recovery method at the time of a system failure according to an embodiment of the present invention.

FIG. 2 is a flowchart showing message data recovery processing when a failure occurs in the online system.

FIG. 3 is a diagram showing a general message recovery method in an online system.

[Explanation of symbols]

21 terminal management unit, 22 business operation unit, 23 message management unit, 24 recovery management unit, 25 journal file, 26 system status management unit, 27 system recovery management unit, 28 message recovery management unit 29
... Message recovery file management unit, 30 ... File system recovery management unit, 31 ... Database recovery management unit, 32 ... System status log file, 33 ... Message recovery file.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G06F 13/00 G06F 15/00 320

Claims (1)

(57) [Claims] 1. A system for recovering a message in the event of a system failure in an online system comprising a terminal computer and a host computer, comprising: input message data from the terminal computer to the host computer; output message data from the host computer to the terminal computer; A journal file for sequentially recording the host processing data in time series, and when a failure occurs, a process of sequentially reading data from the journal file and determining whether or not the data is processed data related to processing prior to the occurrence of the failure If the data read from the journal file is determined to be unprocessed data by the processed / unprocessed determining means, the unprocessed data is determined to be the input or output data. Journal data classification means for classifying message data or host processing data; message recovery means for providing input message data classified by the journal data classification means to a host computer and providing output message data to a terminal computer; message recovery method in the event of a system failure, characterized in that the host processing data classified by the classification means and and a host recovery means for providing to the host computer.