JPH0410137A - Fault backup system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Objective of the Invention] (Industrial Field of Application) The present invention is directed to a distributed processing system such as a process control system that utilizes a plurality of processing units, in the event that a processing unit malfunctions. However, it is related to failure backup methods to prevent complete loss of functionality.

(Prior Art) Currently, in the field of process control, digital control systems are being used that implement control logic using an arithmetic processing unit (hereinafter simply referred to as a computer) and its software.

In such a system, all control logic is implemented by a computer, so failures that occur in the computer tend to have a significant impact on the overall system functionality.

For example, if all functions are implemented with a single computer, if a failure occurs in that computer, the entire system will go down, causing problems in terms of reliability.

For this reason, conventionally, reliability has often been improved by increasing the number of computers used in control systems and making them redundant.

For example, you can configure a triple system using three computers with the same functions, have each computer perform the same processing, compare the outputs, and decide the output of the system by majority vote.
A method in which the intermediate value among the three values is set as the output of the system is a commonly used method.

In addition to these methods, systems using multiple computers are also used in which only one computer performs a function, while the others serve as backup in the unlikely event that that computer malfunctions. . In such a system, when a failure is detected in a computer, the failed computer is disconnected from the system, and a backup computer with the same functionality as the failed computer replaces it and executes the process.

Although the above-mentioned method using multiple computers improves the reliability considerably compared to a single system, it also increases the cost considerably. In particular, since a plurality of computers with the same function are used, the cost burden becomes relatively large.

For this reason, multiple computers usually run different programs, and in the unlikely event that one computer malfunctions, the other computers share the functions of that computer, and the overall A method is used to ensure functionality.

FIG. 3 shows a conventionally used method of this kind.

The computers 31 and 32 that make up the system usually execute individual functions (programs).

The computers 31 and 32 are connected to an input/output device 33 via a signal line 34, and execute control functions while receiving signals from a process to be controlled.

On the other hand, a backup storage device 35 is provided which can access data in common from each of the computers 31 and 32.

This backup storage device 35 is used for each computer 31.3.
All programs executed in 2 are centrally stored. Normally, each computer 31, 32 executes a program stored in its own internal local storage device to perform its respective functions without relying on the backup storage device 35. In the unlikely event that one of the computers 31 or 32 malfunctions and becomes unable to perform its functions, the program in the backup storage device 35 can be transferred to a healthy computer 3.
1 or 32 to ensure the functionality of the entire system.

(Problems to be Solved by the Invention) According to such conventional methods, in the event that a computer malfunctions, the mechanism for switching the program from the backup storage device to a healthy computer and loading it becomes complicated, resulting in a high cost. load increases.

Furthermore, in the unlikely event that a failure occurs in the backup storage device, the functionality of the entire system may go down, which may also pose problems in terms of reliability.

The present invention has been made in view of the above circumstances, and its purpose is to provide a low cost and highly reliable failure backup system.

[Structure of the invention] (Means for solving the problem) The failure backup method of the present invention that achieves the above object is as follows:
In a distributed processing system comprising a storage device that stores a plurality of independent programs and a plurality of arithmetic processing units that execute the programs in the storage device, the execution program in the storage device is stored in the arithmetic processing unit. two registers each storing a minimum address and a maximum address;
and an address change function for changing the value of the register to an arbitrary value, and the address change function changes the address of the register in response to a failure signal generated in the other arithmetic processing unit, and the execution program of the arithmetic processing unit changes. The present invention is characterized in that the range is changed to a range that includes the range of the execution program of the arithmetic processing device that has failed.

(Function) In the present invention, the addresses of two registers storing the minimum address and maximum address of the execution program are changed in response to a fault signal generated in any one of the plurality of processing units, and the processing unit is operated normally. The range of the execution program of the arithmetic processing unit that has failed is changed to a range that includes the range of the execution program of the failed arithmetic processing unit.

(Example) Hereinafter, an example of the present invention will be described in detail using the drawings.

FIG. 1 shows an example in which a process control system or the like is implemented as a distributed system.

This system consists of four computers 11a, llb, 11c.
It consists of lld. The four computers are coupled via signal lines 12 to input/output devices 13a, '13b, 13c and 13d.

Computers 11a, llb, 11c that make up the system
lld includes storage devices 14a to 14d and a register 15a that stores the minimum address of the program to be executed.
~15d1 and register 16 that stores the maximum address
a to 16d are provided.

The storage devices 14a to 14d in each of the computers 11a to 11d store the same program that is an integrated program executed by each of the computers 11a to 11d. Usually registers 15a to 15 in each computer 11a to 11d
The range of programs to be executed by the computer is specified by registers 16a to 16d, and each performs an individual function.

For example, as shown in the figure, in the computer 11a, the storage device 1
4a, the computer 11b stores the program in the second highest range of the storage device 14b, the computer 11c stores the program in the third highest range of the storage device 14C, and the computer 11b stores the program in the third highest range of the storage device 14d. Registers 15a to 15d are configured to execute the program in the lowest range.
, the values of registers 16a to 16d are set.

Each of the computers 11a to lld has input/output devices 13a to 13d.
The controller communicates with the controller via the signal line 12, receives signals from the control target, and executes control functions.

Normally, the functions of such a control computer are realized by repeatedly executing a control processing program. In other words, control of the object is realized by repeating a program or loop infinitely.

In this system, execution starts from the instruction at the address indicated by the registers 15a to 15d of the program on the storage devices 14a to 14d, and the processing is sequentially performed to execute the instruction at the address indicated by the registers 16a to 16d. By storing an instruction to return to the address indicated by 15d, repeated processing is executed between the addresses indicated by registers 15a to 15d and registers 16a to 16d. Since which input/output device among the plurality of input/output devices 138 to 13d is to be interacted with is described in each program, normally the computers 11a to 11d and the input/output devices 13a to 13d are connected on a one-to-one basis. Processing can be done accordingly.

Let us consider a case in which a failure occurs in Toreka's computer in such a situation. For example, assume that a failure occurs in the computer 11b.

When the computer 11a is informed that a failure has occurred in the computer 11b, an interrupt is generated and the interrupt processing program runs. The interrupt processing program rewrites the address indicated in the register 16a of the storage device 14a on the computer 11a side to the jump destination address of the instruction or the value stored in the register 15b.
The value of b is stored. Further, the jump destination address of the instruction at the address indicated in the register 16b (currently the same value is stored in the register 16a) of the storage device 14a is changed to the value of the register lla.

By doing the above, the computer 11a can now execute programs for processing that was previously performed by the computer 11b, in addition to the processing that was conventionally performed by the computer 11a.
The functions of the computer 11b are backed up.

On the other hand, the computer 11b is informed that a failure has occurred in the computer 11b.
If the instruction is sent to register 15c, the interrupt processing program similarly rewrites the jump destination address of the instruction at the address indicated in register 16b of storage device 14C of computer 11c to the value stored in register 15c, and 1
5c changes the jump destination address of the instruction at the address stored in register 16c to the value of register 15b (currently, the same value is likely to be stored in register 15C).

In this way, the functions of the computer 11b can be executed by the computer 11C, so that there is no problem with the sofa 1. is pressed.

In addition, since the relationship with the number of people exiting is written in the program, the response can be accurately determined.

As mentioned above, the processing of an interrupt differs slightly depending on whether the occurrence of a failure is reported to a higher or lower level computer, but in either case, the speed at which the computer executes the functions of the failed computer is reduced. It is possible to buffer things.

According to the above embodiment, backup of a computer in which a failure has occurred is reliably performed, and costs are reduced because a complicated node switching function is not required. High reliability can be achieved because the information is distributed across multiple locations.

FIG. 2 shows another embodiment of the invention.

This embodiment is intended to further reduce costs than the previous embodiments.

That is, in this embodiment, the 1st path configuration in FIG.
d is unified and made accessible from any of the computers 21a to 21d. Further, each of the weighing machines 21a to 21d is connected to input/output devices 23a to 23d via a crossbar. In the computers 21a to 21d, which make up one memory space with the entire storage devices 24a to 24d, there are registers 25a to 25d12 indicating the minimum address and maximum address as before.
6a to 26d, thereby each computer 21a to 21
It is shown in which memory space the program d executes.

With this configuration, each function is executed by a triplet having the same alphabetic symbol, such as the computer 21a, the storage device 24a, and the input/output device 23a. In the unlikely event that a failure occurs in any of the computers 21a-21d, the process of the previously described embodiment is performed, and the combinations of the computers 21a-21d, storage devices 24a-24d, and exit devices 23a-23d are changed. Failure backup is realized.

[Effects of the Invention] As described above, according to the failure backup method of the present invention,
If a failure occurs in one of the computers that make up the system, backup using other computers can be realized at low cost, and since this mechanism is implemented individually for each computer, high reliability can be achieved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a failure backup method according to an embodiment of the present invention, FIG. 2 is a block diagram showing a failure backup method according to another embodiment of the invention, and FIG. 3 is a block diagram showing a failure backup method according to a conventional method. FIG. 11a ~ lid, 21a ~ 21d-... Calculator, 1
2...Signal line, 13a to 13d, 23a to 23d
---I/O device, 14a to 14d, 24a to 24
d...Storage device, 15a to 15d, 16a to 16
d, 25a to 25d, 26a to 26d--Registers. Applicant: Toshiba Corporation

Claims (1)

[Scope of Claims] A distributed processing system comprising a storage device that stores a plurality of independent programs and a plurality of arithmetic processing units that execute the programs in the storage device, wherein the storage device is provided in the arithmetic processing unit. It is equipped with two registers that store the minimum and maximum addresses of the execution program in the device, and an address change function that changes the value of the register to an arbitrary value. A failure backup method characterized in that the address of the register is changed by the address change function, and the range of the execution program of the arithmetic processing unit is changed to a range that includes the range of the execution program of the failed arithmetic processing unit.