JPH05314085A - System for waiting operation mutually among plural computers - Google Patents

Abstract

(57) [Summary] [Object] The present invention, in a mutual operation standby method between a plurality of computers for performing failure recovery between mutual computers, a certain computer is not always set in an idle state, and computer resources are effectively used and processed. The purpose is to obtain a high response even if the work becomes bloated. [Structure] A plurality of computer CPUs 0-3 are made to divide and perform different business processes B0-3 using independent resources respectively stored in the multi-call disk D0-3, and each computer CPU0-3 has its own system. For business processing of
It is operated as a standby system for the business processing of other systems, and when the failure occurrence of a computer is notified via the shared memory SM, the takeover information previously incorporated as operation data in each local disk C0-3. Based on the above, the takeover destination computer is recognized, and the business operation resource related to the occurrence of the fault in the takeover destination computer is selectively read from the corresponding multi-call disk D via the MCP, and the faulty business is taken over.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a mutual operation stand-by system for a plurality of computers which allows a plurality of computers to independently perform online transaction processing and recovers from a failure between the computers.

[0002]

2. Description of the Related Art Generally, in a computer that performs online transaction processing such as online business of bank terminals and online business of airline seats, one computer is used as an active system for business operation in order to avoid business suspension when a failure occurs. , The other computer is used as a standby computer for backup when a failure occurs, and a hot standby system consisting of two computers has been constructed.

FIG. 3 shows a conventional configuration of a hot standby system. In FIG. 3, 11 is an operating computer.
Reference numeral 12 is a standby computer, 13 is a database in which the resources required for online operations operated by this system are stored, 1
Reference numeral 4 is a shared memory serving as mutual communication memory between the operating computer 11 and the standby computer 12, 15 is a terminal device, and FCP
Is a file control processor, and SMI is a shared memory interface. The active computer 11 performs predetermined online business processing based on the information read from the database 13, and the standby computer 12 is always on standby as a backup computer. Is in a state.

In the database 13, a journal file showing a history of business processing and a system status log are also recorded as resources related to a predetermined online business. When a failure occurs in the active computer 11, the database file is recorded. The failure occurrence information is notified to the standby computer 12 via the shared memory 14.

Then, the standby system computer 12 takes in the business history stored in the database 12 and reproduces the business process at the time of failure occurrence, and thereafter, the standby system computer 12 takes over a predetermined online business.

[0006]

However, in the above-described conventional hot standby system, the standby system computer 12 is always in an idle state as an online system unless a business takeover due to a failure occurs, which is a viewpoint of online business operation. From the perspective, there is an inefficient drawback.

That is, for example, even if the online work in the active computer 11 becomes large and the system load becomes high and the responsiveness thereof deteriorates, the standby computer 12 is still in an idle state that does not contribute to the online work.

The present invention has been made in view of the above problems.
To provide a mutual operation standby method between multiple computers that enables efficient use of computer resources and high responsiveness even when online business becomes large, without one computer always being set to the idle state. With the goal.

[0009]

That is, the mutual operation standby system between a plurality of computers according to the present invention has a plurality of computers which perform different business processing by using independent resources, and a plurality of computers for each of the plurality of computers. Prepared independently,
A plurality of business databases storing resources necessary for individual business processing, a database selecting means for selectively switching the plurality of business databases to the plurality of computers, and mutual failure information between the plurality of computers. And a failure notification means for notifying that the failure computer is notified to another computer by the failure notification means, a business database corresponding to the failure computer is selected by the database selection means, and the business process is It is provided with a business takeover means for taking over by a computer.

[0010]

In other words, at the same time that a plurality of computers are each operated as an operating system for performing any online business,
Since they become standby systems for other computers, computer resources can be effectively used, and high responsiveness can be obtained due to the bloated division of online jobs.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a computer system using a mutual operation standby system between a plurality of computers.
In this computer system, for example, four computer CPU0
~ CPU3 is provided.

The four computers CPU0 to CPU3 each become an active system for operating different online business processes B0 to B3 by using independent resources, and at the same time, each of them stands by for business processes other than its own system. Each computer CPU0 to CPU3 has an independent main memory MEM0 to MEM3, and it corresponds to the operation data and load module of the online system stored in advance in the corresponding local disks C0 to C3. , Execute the given online business process B0 to B3.

On the other hand, the resources (database) necessary for the online business processes B0 to B3 allocated to the computers CPU0 to CPU3, and the journal file and system status log showing the business process are the business processes B0 to B3. Different multi-call disks D0 to D3
Stored in.

These four multi-call disks D0 to D
The three and four computers CPU0 to CPU3 are connected via a multi-control processor MCP, and are configured to be selectively switchable by a command from each computer CPU0 to CPU3. Furthermore, the above four computers CP
U0 to CPU3 have a shared memory SM.

The shared memory SM is composed of four computer CPUs 0
Between the CPU3 and the CPU3, the failure occurrence information and the like serving as data for diagnosing the system state of each other are stored. For example, when the failure occurrence information of the first computer CPU0 is written in the shared memory SM. The failure of the first computer CPU0 is notified to the other computers CPU1 to CPU3 via the shared memory SM.

Here, in each of the computers CPU0 to CPU3, information indicating the business takeover destination computer at the time of failure is stored as system operation data in the local disks C0 to C3 of the individual computers CPU0 to CPU3. .. Next, the operation of the computer system using the mutual operation standby system between a plurality of computers having the above configuration will be described.

FIG. 2 is a flow chart showing a failure recovery process in the computer system described above.
U0 to CPU3 are corresponding local disks C
It operates based on the system operation data stored in 0 to C3, and the first computer CPU0 performs the online business process B0 using the database stored in the first multi-call disk D0 as a resource and the second computer CPU0. CP
U1 performs online business B1 using the database stored in the second multi-call disk D1 as a resource, and third computer CPU2 performs online business B2 using the database stored in the third multi-call disk D2 as a resource. In addition, the fourth computer CPU3
Then, the online business B3 using the database stored in the fourth multi-call disk D3 as a resource is performed (step S1).

In this case, each computer CPU0 to CPU3
Are operating systems for the respective assigned online services B0 to B3, and at the same time are standby systems for computers that perform other online services.

Here, for example, when a failure occurs in the first computer CPU0 and the online business B0 is interrupted, the failure occurrence information is written in the shared memory SM,
~ The fourth computers CPU1 to CPU3 are notified that the first computer CPU0 has failed (steps S2, S3).

Then, the second to second notifications of the failure information are given.
The fourth computers CPU1 to CPU3 recognize whether or not the business takeover destination is their own computer based on the system operation data read from the respective local disks C1 to C3. In the system operation data, if the first computer CPU0 failure, the second computer CPU1; if the second computer CPU1 failure, the third computer CPU2; if the third computer CPU2 failure, the fourth computer CPU3, the fourth computer CPU3 failure In this case, if the takeover information of the first computer CPU0 has been assembled in advance, the second computer CPU1 recognizes that it is the work takeover destination (step S4).

Then, the first multi-call disk D0 is selected for the second computer CPU1 via the multi-control processor MCP, and the operating resources of the online business B0 up to the point of failure of the first computer CPU0 are incorporated. Then, along with the recovery processing of the operation state, the recovery processing of the database and terminal message using the journal file is executed.
The online business process B0 performed by the first computer CPU0 is subsequently resumed by the second computer CPU1 (steps S5 and S6).

In this case, the second to fourth computer CPUs 1 to
Business processes B1 to B previously performed in the CPU3
3 is operated in parallel without being affected by the above failure occurrence takeover processing.

Therefore, according to the computer system using the mutual operation standby system between a plurality of computers having the above-mentioned configuration,
~ Different online business processing B using independent resources stored in the first to fourth multi-call disks D0 to D3 respectively for the fourth computers CPU0 to CPU3
0 to B3 are divided and executed, and each computer CPU
0 to CPU3 are all operating systems for their own business processing,
It is operated as a standby system for the business processing of other systems, and, for example, when a failure occurrence of the first computer CPU0 is notified via the shared memory SM, system operation data is previously stored in the individual local disks C0 to C3. The takeover destination computer is recognized based on the takeover information incorporated as, and the operating resources of the business process B0 related to the occurrence of a failure in the takeover destination computer are selectively read from the first multi-call disk D0 through the multi-control processor MCP. Was
All computers C are configured to take over the faulty work.
It is possible to effectively use computer resources with PU0 to CPU3 in an operating state, and to obtain a high responsiveness by dividing a plurality of independent work processes.

[0025]

As described above, according to the present invention, a plurality of computers that respectively perform different business processing using independent resources, and a plurality of computers that are independently provided for each of the plurality of computers and are individually processed. A plurality of business databases storing necessary resources, a database selecting means for selectively switching the plurality of business databases to the plurality of computers, and a failure for mutually notifying failure information between the plurality of computers. Notifying means and, when this fault notifying means notifies another computer of a faulty computer, a business database corresponding to the faulty computer is selected by the database selecting means, and the business process is taken over by another computer. Since it is configured with a takeover means, one computer is not always set to the idle state, the computer resources are effectively used, Inline business becomes possible to obtain a high response even bloated.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a computer system according to an embodiment of a mutual operation standby system between a plurality of computers according to the present invention.

FIG. 2 is a flowchart showing a failure recovery process in the computer system.

FIG. 3 is a block diagram showing a conventional configuration of a hot standby system.

[Explanation of symbols]

CPU0 to CPU3 ... Computer, B0 to B3 ... Online work, C0 to C3 ... Local disk, D0 to D3 ... Multicall disk, MCP ... Multicontrol processor, SM ... Shared memory, MEM0 to MEM3 ... Main memory.

Claims (1)

[Claims] 1. A plurality of computers, each of which performs different business processing by using independent resources, and a plurality of businesses, each of which is independently provided and stores resources necessary for each business processing. Database, a database selecting means for selectively switching the plurality of business databases to the plurality of computers, a failure notifying means for mutually notifying failure information between the plurality of computers, and a failure by the failure notifying means. When the occurrence computer is notified to another computer, a business database corresponding to the faulty computer is selected by the database selection means, and the business processing is taken over by another computer, and the business takeover means is provided. A mutual operation standby system between multiple computers, in which the computers are operated as an active system and a standby system, respectively.