JPH0215093B2 - - Google Patents

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a distributed loosely coupled multiprocessor system, and in particular, to a system that facilitates processing during configuration changes such as hot/standby switching of devices. This invention relates to a device address translation control method for

[Conventional technology]

Recently, multiple processors with independent functions are shared from the standpoint of distributing functions or distributing risks.
The number of computer systems configured as distributed, loosely coupled multiprocessor systems connected via DASD (direct access storage devices) is increasing.

FIG. 2 shows the configuration of a meteorological satellite control system as an example of such a distributed loosely coupled multiprocessor system. In the figure, 1
4 are processors #1, #2, #3, respectively.
#4 and 5 are shared DASD, 6 is console, 7 is global operating system GS, 8 to 11 is application program APL corresponding to the business group, 12 is unique DASD, 13 is unique graphic display GD, 14 is unique Magnetic tape device MT, 15 is a unique card reader CR, 1
6 is a unique line printer LP, 17 is a unique DASD
It is.

The business processing functions of the entire system are divided into four systems and distributed among processors #1, #2, #3, and #4, respectively. In other words, processor #1 is in charge of centralized control of the entire system and processing of the operation management system, such as online control and operation schedule management, processor #2 is used as a backup (standby) for the operation management system, and processor #3 is used as a backup (standby) for the operation management system. Processor #4 is in charge of the image processing system, and processor #4 is in charge of data analysis and the satellite control system. Unique DASD12
Each of the individual DASDs 17 is a processor-specific I/O device connected by a solid line. Although not shown in the figure, it is also equipped with a TSS terminal and data communication interface that are shared by the entire system.

Data is exchanged between each processor via a shared DASD 5, and the entire operation is controlled by a console 6 and a global operating system.
Centrally controlled by GS7.

Each processor (for example #3) runs its application program APL (for example 10) as follows:
Processor-specific I/0 devices (e.g., unique DASD1)
Processing is performed by connecting to 2), but if another system's processor (for example, 4) goes down, the I/0 device unique to that processor (for example, unique DASD 17)
It is also possible to combine it with

[Problem that the invention seeks to solve]

Generally, in a distributed loosely coupled multiprocessor system, each processor has its own I/O device.
It has its own device address system. Therefore, when moving an I/0 device unique to one system to another system,
The device address used for that I/0 device in the previous system may overlap with other existing I/0 device addresses in the new system, or various other problems may occur, making it impossible to transfer commands and messages. Therefore,
There were times when it became unusable.

Therefore, for example, when a processor goes down due to a failure and is switched to a backup processor, or when individual I/0 devices are switched, the configuration control under the new processor It was necessary to review the device address of the /0 device, which caused problems in operation and system operation.

[Means for solving problems]

The present invention enables an operator to switch between processors and I/O devices when switching between processors and I/O devices when a failure occurs.
In order to eliminate the need to be aware of the actual physical connection state between the computer system and the 0 device, the computer system is divided into a logical system consisting of the business that each processor is responsible for or the name of its operating system and the logical name of the I/0 device used. This allows operations to be performed using only elements.

Therefore, the principle of the present invention is that one unit of the logical system defined by the business group and the I/O device used is
A table is set up that corresponds to a unit of the physical system defined by the main body system device that makes up the processor hardware, such as a processor or a channel, and when a processor is switched due to a failure, this table is rewritten to prevent the failure from occurring. Change the correspondence so that the original processor that went down, that is, the logical system consisting of the business group and I/0 devices used by that physical system, is transferred to the physical system of another processor, and then Regarding devices, we have created a table that associates the device address defined for each physical system with a logical device name that is uniquely determined regardless of the physical system.
0 device is connected, the operator inputs a command using the logical device name, and changes the logical device name to the device address in the new processor using the above table to ensure correct control. It is.

Accordingly, the configuration of the present invention is a distributed loosely coupled multiprocessor system in which different business group processing functions are distributed to multiple processors, and one processor is provided with a management function for the entire system, and each processor is connected via a shared DASD. In the system, the processors with the above management functions store logical system information that defines the correspondence between each business group distributed to each processor and the logical device name of the I/0 device used, and the main system device for each processor. For each I/0 device, the physical system information that defines the The logical system, physical system, and I/
Equipped with a management information means indicating the connection relationship with the 0 device, when a failure of a certain processor is detected, the business group and used I/0 device of the processor are transferred to another alternative processor, and the above-mentioned management is carried out accordingly. If the information updates the connection relationship between the logical system, physical system, and I/0 device, and also inputs a command that specifies the I/0 device, specify the I/0 device using the logical device name, and It is characterized in that after a command is input, the logical device name is converted into a device address using the management information.

[Action of the invention]

An overview of the operation based on the present invention will be explained using FIG. 3.

For the sake of simplicity, FIG. 3 shows the main parts related to the present invention of a system centralized control mechanism in a distributed loosely coupled multiprocessor system consisting of two processors.

In FIG. 3, reference numeral 18 denotes a system management section that has the function of monitoring and switching each system; 19 a console management section that has functions of converting logical system names and physical system names and converting device addresses; This is management information including a correspondence table regarding I/0 devices.
In the illustrated example, as shown in the management information 20, the system has two logical systems #1 and #2 indicated by 21 and 22.
#2 and the two physical systems #1 shown as 23 and 24,
#2, and three I/0 devices #1, #2, and #3 indicated by 25 to 27. Furthermore, I/
0 devices #1 and #3 are unique I/0 devices, #2 is shared I/0
It is a device.

Logical systems #1 and #2 are each defined by a combination of a business group distributed to two processors and a logical device name (for example, I/0 name and device address) of the I/0 device used.

Further, physical systems #1 and #2 are each defined by a combination of main body system devices such as a CPU, a channel processor CHP, and a channel CH that constitute each processor.

The I/0 device used is represented by a combination of a logical device name uniquely defined within the system and a device address for each physical system.

The management information 20 includes the correspondence between each of these multiple logical systems, physical systems, and used I/0 devices, for example, logical system #1 - physical system #1 - represented by a solid line.
It holds information representing the combination of I/0 devices #1 and #2 and logical system #2-physical system #2-I/0 devices #2 and #3 in a table format.

By the way, when one processor goes down due to a failure, the system management unit 18 detects this and performs a switching process to transfer the function to the other normal processor, and also modifies the management information 20 accordingly. For example, if a processor containing logical system #1 and physical system #1 becomes unusable, logical system #1 is separated from physical system #1 and connected to physical system #2 of the other processor, and I/
0 device #1 is connected to physical system #2. Therefore, physical system #2 includes logical systems #1 and #2,
I/0 devices #1, #2, and #3 are all coupled, and the management information 20 comes to display this coupled state.

As described above, when the logical system is switched between processors, the device address of the I/0 device used may change. Therefore, for commands such as I/0 configuration control commands that require the operator to issue with the device address of the I/0 device in mind, specify the logical device name instead of the device address, and The management unit 19 refers to the management information 20, converts the logical system to the physical system, and the logical device name to the device address, and sends a command specifying the device address to the corresponding physical system.
[Example] The details of the present invention will be described below with reference to Examples.

FIG. 1 is a block diagram of a system according to an embodiment of the present invention. In the figure, 1, 2, 3, and 4 are respectively processors #1, #2, #3, #4, and 5 are shared.
DASD, 6 is console, 7 is global operating system GS, 18 is system management department, 19
is the console management section, 28 is the system switching processing section,
29 is a switching device, 30 is a management information memory, 3
1 is a logical system/physical system correspondence table, 32 is a logical device name/device address correspondence table, 33 is a logical system/physical system correspondence table search processing section, and 34 is a logical device name/device address conversion processing section. In addition,
Elements 1 to 7 are the same as in Figure 2, and 1
Elements 8 and 19 are the same as in FIG.

When the system management unit 18 detects that a processor in the system is down, it first activates the system switching processing unit 28 to start switching processing. The system switching processing unit 28 executes switching of logical systems, physical systems, and I/0 devices between related processors via a switching device 29.

At the same time, the system management unit 18 instructs the logical/physical system/correspondence table search processing unit 33 to rewrite the logical/physical system correspondence table 31.

The logical system/physical system correspondence table 31 displays definition contents of all logical systems, physical systems, and used I/0 devices in the system and their connection relationships. Note that #1, #2, #3, and #4 represent the numbers of each of the four logical systems and the physical system. The table is rewritten by replacing the arrows (pointers) indicating the above-mentioned connection relationships.

Also, logical device name/device address correspondence table 3
2 displays the logical device name of the I/0 device and the corresponding device address in each logical system #1, #2, #3, #4. For example, the device address of one direct access device (DASD) with the logical device name "DP00" is
It is set like 030, 030, 040, 040.

If the operator inputs the configuration control command V _L DP00, OFFLINE to instruct the direct access device DP00 to be disconnected from the system for offline operation, the logical device/device address translation unit 34 Find the logical system number and physical system number corresponding to the logical device name DP00 using the correspondence table 31, and then use the logical device name/device address correspondence table 32 to find the logical system device address to which the correspondence between the physical system number and the logical device name corresponds. and sends it to the processor corresponding to the physical system number.

〔Effect of the invention〕

As described above, according to the present invention, even if the device address of an I/0 device changes due to processor switching, the operator can input a command using the logical device name without having to be aware of the change. This improves operability, and reduces errors and improves reliability compared to the conventional case where the device address for each processor is considered.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram of a system according to an embodiment of the present invention.
FIG. 2 is a diagram showing a general configuration example of a distributed loosely coupled multiprocessor system, and FIG. 3 is a schematic explanatory diagram of the present invention. In the figure, 1 to 4 are processors #1 to #4,
5 is a shared DASD, 6 is a console, 7 is a global operating system GS, 18 is a system management unit, 19 is a console management unit, 28 is a system switching processing unit, 29 is a switching device, 30 is a management information memory, and 31 is a logic unit. System/physical system correspondence table, 32
is a logical device name/device address correspondence table, 33
is a logical system/physical system corresponding table search processing unit, 34
represents the logical device name/device address conversion processing section.

Claims (1)

[Scope of Claims] 1. A DASD in which different business processing functions are distributed to multiple processors, and one processor is provided with a management function for the entire system, and each processor is shared.
In a distributed loosely coupled multiprocessor system connected via information, physical system information that defines the main system device for each processor, and for each I/0 device, a logical device name that is uniquely determined regardless of the physical system, and a device address that is determined for each logical system. A management information means is provided for each processor that indicates the coupling relationship between the logical system, the physical system, and the I/0 device using the I/0 device information that defines the correspondence between the processor and the I/0 device. , transfer the business group and I/0 device used by the processor to another alternative processor, update the connection relationship between the logical system, physical system, and I/0 device indicated by the management information above, and When inputting a command that specifies the /0 device, specify the I/0 device using the logical device name, and after inputting the command, convert the logical device name into a device address using the above management information. A distributed loosely coupled multiprocessor system control method.