JPH09185659A - Data succeeding method for cluster type server system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the efficiency of transmission by storing transaction log data, which are generated by an account system so that an end user can collate the result of transaction on the day, into a customer database provided at an adjacent outside system in real time. SOLUTION: Transaction log data 1 sent to the adjacent outside system in real time are dispatched to a transaction log reception control part 2. At the transaction log reception control part 2, a distribution destination deciding part 3 decides distribution destination store group with a 'store number' in the transaction log data as a key. A transaction log stack part 5 writes the transaction log data into a transaction log stack file 6 prepared (classified) for each distribution destination store group. A transaction log transmission control part 8 is parallelly operated for each distribution destination store group and a transaction log record 7 stored in each transaction log stack file is extracted, blocked and succeeded to a correspondent job node.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an external connection system of an electronic banking system, and when a large amount of transaction log data generated in an accounting system is received by an external system which is a cluster type server system, the data is transferred to another node. Processing performance guarantee method.

[0002]

2. Description of the Related Art Conventionally, when a transaction occurs in an accounting system, transaction log data is written in a queue held by a receiving node for each record, and after extraction, a distribution destination node is determined and a customer database is held. The method was taken over.

[0003]

The above-mentioned conventional data transfer method for other nodes requires the process activation to perform RPC (remote procedure call) to transfer data to other nodes by the number of received transaction log data. , C
It was a method in which the load on the PU was a concern. In addition, since data is transmitted to other nodes on a record-by-record basis, there is a problem that the line utilization rate between nodes is high and the transmission efficiency is poor.

[0004]

[Means for Solving the Problems] In order to solve the above-mentioned problems, a stack file for each store group is prepared in the receiving node, the generated transaction log records are saved, and one of the stack files has a specific number of records. When it reached, all transaction log records were sent to other nodes all at once.
In this way, by stacking the stack file and the transmission control unit and the node (store group) that should be taken over, the number of process activations is limited at the time of distribution to one transaction log data group, and at the same time. Evenness of service will be guaranteed.

[0005]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described with reference to the drawings.

[0006] Transaction log data 1 randomly generated in the accounting system is sent to the external system in real time using a data transmission means by a line and passed to the transaction log reception control unit 2.

In the transaction log reception control unit 2, the distribution destination determination unit 3 refers to the distribution destination determination table 4 by using the "store number" in the transaction log data as a key, and the distribution destination store group (nodes where data should be handed over) ) Is determined.

The transaction log stack unit 5 writes the transaction log data to the transaction log stack file 6 prepared (classified) for each distribution destination store group.

The transaction log transmission control unit 8 is configured such that the transaction log stack unit selects one of the transaction log stack files.
When the number of stacks in each node is less than the total number of input transaction log records
Will be started. If the stack file fullness is not detected, the stack is repeated. The transaction log transmission control unit operates in parallel for each distribution destination store group, extracts the transaction log record 7 retained in each transaction log stack file, and blocks and responds when there are multiple transaction log records. Take over to the business node to be used.

[0010]

As described above, according to the present invention, the transaction log data sent from the accounting system in real time is temporarily held in the transaction log stack file classified for each store group, and the transaction log stack file is stored for each store group. It is effective in assuring the evenness of the customer database update timing by activating the transmission processes in parallel, and in accumulating processing efficiency at the transmission destination by transferring a plurality of transaction logs to another node in one transmission process.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is a flow chart showing an example of an operation in a data takeover process to another node in the present embodiment.

[Explanation of symbols]

1 ... Transaction log data, 2 ... Transaction log reception control unit, 3
... Distribution destination determination unit, 4 ... Distribution destination table, 5 ... Transaction log stack unit, 6 ... Transaction log stack file,
7 ... Transaction log record, 8 ... Transaction log transmission control unit.

Claims (1)

[Claims] 1. Information on a node to be transmitted from the transaction log data in a data distribution processing method in a cluster type server system when a large amount of transaction log data generated in an accounting system is taken over to a node holding a customer database. A distribution destination determining unit that acquires (store group information); a transaction log stack unit that accumulates the transaction log data in a transaction log stack file prepared for each node (store group) and monitors the accumulation status; When the transaction log stack unit detects that the stack file is full, it is activated for each node (store group), extracts the transaction log record from the transaction log stack file, blocks it, and transmits it to the corresponding transaction log transmission control. Other node data distribution processing method characterized by possessing a department.