JPH06259274A - Duplex system - Google Patents

Abstract

PURPOSE:To construct the high-reliability computer system of duplex system configuration at low cost without using any special shared memory device when constructing the system. CONSTITUTION:Two pieces of an execution system computer 1-1 and a standby system computer 1-2 are connected by a data transmission line 2, and an area shared data area 3 is provided on each computer. A duplex system managing process 4 of the execution system computer writes data into the shared data area 3 on the execution system computer 101 according to a write request (a) from a process executing means 5 and issues the write request to the duplex system managing process 4 of the standby system computer. According to the request, the duplex system managing process 4 of the standby system computer performs write into the shared data area 3 on the standby system computer 1-2. At the time of recovery from any fault, the standby system computer issues the read request of data to the execution system computer and copies the data from the shared data area 3 of the execution system computer to the shared data area 3 of the standby system computer.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dual system in which when a failure occurs in a computer system, another computer system takes over the processing.

[0002]

2. Description of the Related Art FIG. 3 is a diagram showing a conventional dual system using a shared memory. Reference numeral 1-1 is an active computer, which is a computer that is currently performing processing. Reference numeral 1-2 is a standby computer, which is a computer that takes over the processing when a failure occurs in the active computer 1-1. Reference numeral 10 denotes a shared memory, which is a memory in which data required to take over the processing from the active computer 1-1 to the standby computer 1-2 is stored.

FIG. 4 shows a data protection system of a distributed system disclosed in Japanese Patent Laid-Open No. 63-236138, in which 21 is a parent computer, 22 is an external storage device, 2 is a data transmission line, and 25 is a data transmission line.
Reference numerals -1 to 25-n are slave computers, 26 is an online data storage unit, 27 is an external peripheral device, and 28 is a data protection storage unit.

FIG. 5 shows an exclusive control system for a network virtual shared memory disclosed in Japanese Patent Laid-Open No. 2-27465, in which 100 is a virtual storage area, 100a is a real storage, 5 is a process for accessing shared data, and 110 is a semaphore queue.

Next, the operation will be described. In FIG. 3, the active system computer 1-1 performs online operation, and the standby system computer 1-2 is on standby. When a failure occurs in the execution computer 1-1, an abnormality detection signal enters the standby computer 1-2,
The standby computer 1-2 issues an initialization request signal to the execution computer 1-1. The standby computer 1-2 takes over the processing of the executing computer 1-1, and the standby computer 1-2 continues the online operation. The execution computer 1-1 writes necessary data for the standby computer 1-2 to take over the processing in the shared memory device 10, and the standby computer 1-2 refers to the data to perform the processing. When the failure of the execution computer 1-1 is removed, the execution computer 1-1 becomes the standby system, and when the failure of the standby computer 1-2 occurs, it becomes the execution system again.

In FIG. 4, the data in the online storage data storage unit 26 on the slave computer 25-1 is backed up on the data protection storage unit 28 of the slave computers other than the slave computer 25-1. After the failure occurs in the child computer 25-1, after the failure is repaired, the data is restored from the data protection storage unit 28 of another small computer, and the interrupted processing is performed by the child computer 2
5-1 continues.

In FIG. 5, the data desired to be shared between the computers is held in the real memory on any one of the computers. A computer that wants to access the data is a computer having an actual storage in which the data resides and having the data transferred after performing exclusive control.

[0008]

In the system of FIG. 3, a special shared memory device is required outside for storing the data to be shared by the active system and the standby system.

In the system shown in FIG. 4, when a computer fails, the computer is restored, the data backed up on another computer is restored, and the processing is continued. Other computers do not take over the processing of.

In the system shown in FIG. 5, if a computer holding a real memory fails, shared data is lost.

The present invention has been made to solve the above problems, and an object thereof is to construct a dual system without using a special external shared memory.

[0012]

In a dual system according to the present invention, two computers are connected by a data transmission path, a shared data area is provided on each computer, and a shared data area is provided by a dual system management process. It writes the data to the computer, detects an abnormality in the active computer, and copies the data at the time of recovery from the failure.

That is, in the dual system according to the present invention, two active computers and a standby computer are connected by a data transmission path, a shared data area is provided on each computer, and shared by a dual management process. Controls writing to the data area. On the active computer, the dual management process writes data to the shared data area on the active computer according to the write request from the process execution means, and issues a write request to the dual management process on the standby computer. . The dual system management process on the standby system computer writes to the shared data area on the standby system computer accordingly. If the standby system computer is recovering from the failure, the dual system management process on the standby system computer further issues a data read request to the dual system management process on the active system computer, and The dual system management process reads data from the shared data area on the active system computer and sends it to the dual system management process on the standby system computer.

[0014]

In the dual system according to the present invention, a part (or the whole) of the memory provided in each computer is used as a shared data area, and data is written to this shared data area by the dual management process. , The same data is held in the active and standby computers. Therefore, no special shared memory device is required externally.

When a failure occurs in one of the computers, the failure is removed, the computer is restored, and the standby system is set again.
The dual system management process on the computer which is being restored issues a read request to the active computer in order from the beginning of the shared data area while writing according to the write request from the active computer. When there is no write request from the own system process, the dual system management process on the executing system computer reads the data from the shared data area of the own system according to the read request from the recovery system and sends the data to the recovery system. Issue a write request.

[0016]

【Example】

Example 1. An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, 1-1 is an execution computer, 1-2 is a standby computer, and 3 is a shared data area for storing data to be shared by two computers on each computer. The shared data area 3 includes the active computer 1-1 and the standby computer 1.
-2 is configured by using a part (or all) of a memory provided in advance. Therefore, the shared data area 3 can be accessed by the same access as the case where the active computer 1-1 or the standby computer 1-2 makes a normal memory access, and no special memory access means is required. Reference numeral 4 is a dual system management process, and 5 is a process execution means for executing a process for reading and writing shared data. In addition, reference numeral 1-3 in FIG. 2 denotes a recovering computer which is about to become a standby system after the failure is dealt with.

Next, the operation will be described. In FIG. 1, the execution-system computer 1-1 performs the processing intended by the system by the process execution means 5. Standby computer 1
-2 stands by to take over the processing from the active computer 1-1 when a failure occurs in the active computer 1-1.
The process execution means 5 on the execution system computer 1-1 directly refers to the shared data area 3 when reading data to be shared with the standby system computer 1-2, but on the execution system computer 1-1 when writing data. The write request together with the data is issued to the dual system management process 4 of No. 1 and waits until it is completed. Dual system management process 4 on the active computer 1-1
Writes data to the shared data area 3 on the active computer 1-1 according to the write request from the process executing means 5, notifies the process executing means 5 of completion, and 2
A write request is issued together with the data to the dual system management process 4 on the standby system computer 1-2 via. When the dual system management process 4 on the standby computer 1-2 receives the write request, the shared data area 3 on the standby computer 1-2 is received.
Write the data passed to the corresponding area of.

When a failure occurs in the execution system computer 1-1,
The abnormality detection signal is transmitted to the standby system computer 1-2, the dual system management process 4 on the standby system computer 1-2 issues an initialization signal to the executing system computer 1-1, and the processing content is executed.
-1 switches to the processing performed by the dual system management process 4, and the process execution means 5 on the standby computer 1-2 refers to the contents of the shared data area 3 on the standby computer 1-2 and executes it. The process executed by the process execution means 5 of the system computer 1-1 is taken over, and thereafter the system operates as an execution system.

Next, the recovery operation will be described with reference to FIG. The recovering computer 1-3 is a computer in a recovering state due to a failure in the active computer 1-1 of FIG. Figure 2
1 is a computer in which the standby system computer 1-2 in FIG. 1 is switched from the standby system to the active system. In FIG. 2, the recovering computer 1-3 makes the contents of the shared data area 3 on the recovering computer 1-3 the same as the contents of the shared data area 3 of the executing computer 1-1, and switches to the standby system. The system is about to be ready for the failure of the active computer 1-1.

The dual system management process 4 on the recovering computer 1-3 performs the same processing as the dual system management process 4 on the standby system computer 1-2 of FIG. In order to make it the same as the shared data area 3 on the active computer 1-1, the dual system management process 4 on the active computer 1-1
A data read request is issued to.

The dual system management process 4 on the active computer 1-1 has the same processing as the dual system management process 4 on the active computer 1-1 of FIG. When receiving a read request from the above dual system management process 4, the requested data is read from the shared data area 3 on the executing system computer 1-1, and the dual system management process 4 on the recovering computer 1-3 is read. A write request is issued to it.

When the dual system management process 4 on the recovering computer 1-3 finds that it is a response to the read request issued by itself, the shared data area 3 on the recovering computer 1-3 is restored.
After writing to, the issuance of the next read request is repeated until all shared data can be copied. When all the shared data can be copied, the recovering computer 1-3 shifts to the standby system.

As described above, this embodiment uses a special external shared memory in a dual system consisting of an active system and a standby system when a failure occurs in the active system and the processing is taken over by the standby system. It is characterized by doing it without. In a dual system that does not use shared memory,
The system is characterized by having a copy function of the shared data area for restoring to the standby system after repairing the system where the failure has occurred.

Example 2. It should be noted that in the above embodiment, the data transmission path, the abnormality detection signal, and the initialization request signal are provided one by one, but they may be duplicated for higher reliability. The data transmission line may be a network or a dedicated communication line.

[0025]

As described above, according to the present invention, since the data shared by the execution system and the standby system can be placed in the memory of each computer, no special device such as a shared memory device is required. , The system can be cheap.

[Brief description of drawings]

FIG. 1 is a dual system diagram according to an embodiment of the present invention.

FIG. 2 is a diagram of data copy for recovering a system in which a failure of the dual system has occurred.

FIG. 3 is a dual system diagram using a conventional shared memory device.

FIG. 4 is a diagram of a data protection system of a distributed system disclosed in Japanese Patent Laid-Open No. 63-236138.

FIG. 5 is a diagram of an exclusive control system for a network virtual shared memory disclosed in Japanese Patent Laid-Open No. 2-27465.

[Explanation of symbols]

 1-1 Executing computer 1-2 Standby computer 1-3 Restoring computer 2 Data transmission path 3 Shared data area 4 Duplex system management process 5 Shared data read / write process 10 Shared memory device 21 Parent computer 22 External storage device 25-1 to 25-n Child computer 26 Online data storage unit 27 External peripheral device 28 Data protection storage unit 100 Virtual storage area 100a Real storage 110 Semaphore queue

Claims (2)

[Claims] 1. A dual system having the following execution system computer and standby system computer (A) Execution system computer having the following elements (a) Execution system shared data area for storing shared data,
(B) an execution system process executing means for executing a process of reading / writing shared data from / to the execution system shared data area; and (c) a process executed by the execution system process execution means to the execution system shared data area. An active system management process for transmitting a shared data write request to the standby computer, (d) an abnormality notifying means for detecting an abnormality and transmitting an abnormal notice to the standby computer, (B) a standby computer having the following elements: ) A standby shared data area for storing the same shared data as the shared data stored in the active shared data area, and (b) a standby system for executing the same process as the process executed by the active process execution means. Process execution means, (c) receiving a write request for shared data transmitted from the active management process, and based on the write request A standby system management process for writing shared data to the standby system shared data area, and (d) the standby system using the shared data stored in the standby system shared data area when an abnormality notification is received from the abnormality notification means. Execution system computer substitution means for substituting the execution system computer by starting execution of the process being executed by the execution system process execution means by the process execution means. 2. The standby system computer further comprises initialization request means for outputting a request to initialize the active system computer when an abnormality notification is received, and the active system computer comprises the initialization request means. A shared data copying means for copying the shared data from the standby shared data area of the standby computer to the active shared data area while recovering the abnormality by executing the initialization based on the request from The dual system according to claim 1, wherein the active computer is restored as a standby computer.