JPS60101665A - Configuration switching system - Google Patents

Abstract

PURPOSE:To enable switching between CPU dynamically by providing a switching controlling device between CPU when an impediment occurs in one of plural working CPU that use plural inputting/outputting devices in common. CONSTITUTION:When an impediment occurs in CPU1 during sytem operation, the switching controlling device 3 detects the impediment and notifies CPU2 of the matter through an interface 23. Then, the device notifires CPU2 of I/O state connected to CPU1 i.e. connection of F53, F55 and CCU51 by simultaneous switching instruction. The switching controlling device 3 makes change-over switches 43, 41 and 45 switch the I/O to CPU2 through an interface 34, and after completion of switching, additionally stores F53, CCU51 and F55 in a configuration table of own device. On receiving report of completion of switching, CPU2 stores F53, CCU51 and F55 additionally in its own configuration table and starts operation.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a configuration switching method for an information processing system, and particularly to a configuration switching method suitable for a system having a plurality of processing devices.

[Background of the invention]

In information processing systems that require high reliability, such as online real-time systems, each component of the system, including the processing unit, is multiplexed and switched to backup equipment in the event of equipment failure in order to improve the system availability. In most cases, the system is designed so that processing can continue without interruption.

In addition, in such cases, there is no point in switching the equipment if it takes a long time, so a predetermined procedure may be used, such as a switching control device that controls switching based on commands from the processing device, without requiring switching operations by an operator. A method of automatically switching according to the following is also used.

In conventional information processing systems of this type, for example, two processing units are used as the active unit and a standby unit, and a monitoring device is installed to monitor the operation of each processing unit at regular intervals, so that the operating unit does not become a problem. By detecting this and contacting the standby machine, the normal standby machine is connected to the faulty working machine via a switching control device that has a separate input/output layer W for peripheral devices and terminal devices connected to the faulty working machine. There is a system that allows the backup machine to take over and continue the processing of the active machine.

In such a system, the processing equipment that serves as the active machine executes business processing, while the processing equipment that serves as the backup machine constantly monitors the operating status of the active machine so that it is ready for immediate backup in the event of a failure of the working machine. 1. In addition to being a standby standby machine, the standby machine is usually a standby machine, and in order to prevent degeneration of business processing functions, the standby machine is generally a processing device that has the same processing capacity and functions as the active machine.

Further, the monitoring device and the switching control device are integrated, and a single device often has both functions.

Also, instead of fixing the processing equipment to the active machine and the standby machine as described above, each processing equipment can perform business processing at the same time, and each processing equipment can be configured so that it mutually backs up the work of other processing equipment. - A method is also used in which when one of the processing devices becomes faulty, another processing device takes over the task being executed by the faulty processing device and continues the processing.

In all of these conventional systems, when a failure occurs in a certain processing device, the input/output device connected to that processing device can be switched to another processing device without an operator's switching operation. - It was necessary to determine in advance a fixed switching procedure for the target input/output devices. For this reason, for example, if a new file device is required during a business process due to a transaction request input from a terminal device, and the file device connected to the processing device currently processing the business becomes a failure, another It was difficult to connect and take over the processing equipment, and this could not be done without operator intervention. In addition, if there is a change in the processing task, such as a task that is not scheduled to be executed on a certain processing device to be processed on that day due to circumstances, such a change will generally change the connection of the input/output device connected to that processing device. 7) Dynamically follow changes in the system configuration when changing the connection of input/output devices. I couldn't do it.

On the other hand, instead of changing the connections of input/output devices to processing devices as described above, the same input/output device is always connected to all processing devices in terms of hardware, and connected to the necessary processing devices using software. - Although it is possible to disconnect, due to a processing device malfunction or software bug, a processing device that should not be accessing the input/output device may access the input/output device, and the contents of the file held by that input/output device may be lost. There is a limit to the construction of a highly reliable system because it may cause adverse effects such as destruction.

[Purpose of the invention]

An object of the present invention is to provide a configuration switching method that dynamically switches between a plurality of input/output devices and a plurality of processing devices that share the input/output devices.

[Summary of the invention]

The present invention has a configuration switching means that connects an input/output device to the requesting processing device in response to a request from the processing device, and holds information about the connected input/output device for each of the processing devices, Each of the processing devices also maintains information about the input/output devices connected to its own processing device, and when any of the processing devices detects a failure in another processing device that should be backed up.

The one of the processing devices is characterized by a configuration switching method that requests the configuration switching device to switch the connection of the input/output device based on the input/output device connection information regarding the other system processing device transferred from the configuration switching device. shall be.

[Embodiments of the Invention] Embodiment 1 of the present invention will be described below with reference to FIGS. 1 and 2. FIG. 1 is an example of the system configuration. Processing units (CPU) 1 and CPU 2 are central processing units,
51 to 55 are input/output bags f (Ilo) connectable to both C1-'U1 and CPU2. Especially 51.
52 is a communication control unit (ecU), and 53 to 55 are file devices (F') such as a magnetic disk storage device or a magnetic tape device. The terminal device 6 is 51,5
2 (J) A terminal device connected to the CCU.

The interfaces 14 and 24 are input/output interfaces through which the CPU 1 and CPU 2 exchange information with each of the CPUs 51 to 55 110, respectively. The changeover switches 41 to 45 are 51 to 5, respectively.
This is a switch for connecting the 50 input/output device to the CPU 1 or CPU 2, and the CPU can be switched manually or automatically by an external signal. Note that the changeover switches 41 to 45 have a neutral state in which they are not connected to any CPU. The switching control device 3 automatically switches l/U 51 without operator intervention.
55 to Ci'U1 and CPU2, the interfaces 15 and 23 are interfaces for transferring information to and from the CPU 12, respectively, and the interface 34 is a switching switch 41. ~
This is an interface for sending a switching signal to 45. Here, the switching control device 6 performs the following operation σ).

C respectively through the lal interface 13.23
By checking whether a response is returned from CI'U1.2 within a specified time to the information sent to PQl and 2 at fixed time intervals, -CL'U
1 and C) I constantly monitor whether the '02 car is in working condition.

lbl When an abnormality is detected in any of the CPUs [J, the abnormality in the CPU of another system is notified to other normally operating CPUs.

(The C1 switching control device 6 stores a 1/(J configuration table indicating the I10 connection status for each CPU.
When a CPU abnormality is detected by I, the normal CPU is notified along with the I10 connection status of the CPU of the other system.

(Receives a command to switch each of 110 51 to 55 to CPU1 or CPU2 from CPU1 or CPU2 through the dl interface 15.25.

Note that there are two switching commands: one for switching each 170, 51 to 55 individually, and the other for switching 1/Q of each CPU connected to the other system CPU to -★.

(el Based on the above TDL, each switch 4 is connected to each 170 through the interface 54.
1 to 45 to change the connection to CPLJl or CPU2, and switch between interface 14tI' or interface 24 and ilo 51 to 55.

(fl After completing the above (el switching), update the I10 configuration table stored in the switching control device f6 to the state after switching, and then
For interface 13 or interface 23
Send a switching completion report via v.

Yx each CPU's IPI, (Initial program)
Loading) startup may be performed by the switching control device 5 or by other devices (CPU). The switching control device 6 switches the CPU by starting the LPL.
After receiving the report from σ) that the CPU is ready for operation, the CPU enters the above-mentioned (aJσ) monitoring system.

Next, CPU1. The I10 configuration table stored in the CPU 2 and the switching control device 3 will be explained. Figure 2 shows an example (al is the Ilo configuration table σ held by each CPU and switching control device 6).
1 and 2 respectively indicate the Ilo connection status of the own system. "f configuration table 1.82 f?r:", and the switching control device 6 has a configuration table 91 for CPU 1 and a configuration tape 92 for CPU 2. .

The configuration table 91 is a copy of the configuration table 71σ),
Configuration table 92 is a copy of configuration table 82.

For example, the configuration table 7111r has F5'3 and CCU3.
Two entries of 1 are shown (but this & CP
For the business being executed on Ul, that is, for CPU1, there are 53
The file device shown in FIG. 5 is connected to the CCU 51, and the other Ilo's are not connected.

Here, each entry in the configuration tables 71, 82, 91, 92 simply indicates whether or not the Ilo is connected.

However, if necessary, each IJ may be able to identify the unique attributes of that L/U, such as the device address, the symbolic name of the device, and information related to the task being executed, such as the volume of the storage volume mounted on that device. It may also include the number, name of the dataset in use, etc. 2 These 110 configuration tables are - In the CPU, the operating system is separate from each 01'U business program.
It is managed by a system or a predetermined monitoring program, and in the switching control device 6, it is managed by a dedicated monitoring program within the device.

Next, the operation of this embodiment will be explained based on the above configuration. First, it is assumed that under normal operating conditions,
It is assumed that the CPU 1 performs business AV processing, and the CPU 2 processes another business B.

Furthermore, in the event that any one of the CPUs fails, the other CPUs can serve as standby machines. + Sufficient processing power for CPU 2 to process jobs A and B when CPU 1 fails, and for CPU 1 to process jobs A and B when CPU 2 fails shall be equipped with the following. Also 11
Regarding 0, Business A always uses U56 and CC[J51, Business 1f3B always uses Do54 and C(,'U52, and +i"5!5 operates as necessary during business processing. Assume that work A=1: or Ilo assigned to work BK.

FIG. 2 shows CPU1. Cf'U2 and the transition of the contents of the Ilo configuration table of the switching control device B, that is, each L1
It shows the transition of the connection state to each CPU of 051 to 55.

The operation of this embodiment will be explained below with reference to FIGS. 1 and 2. First, after CPU1 starts the engineering PL,
Start. Here, the CPU 1 communicates the cP of U53 and CcU51 to the switching control device 3 through the interface 13.
Sends a switching command to UL. The switching control device 3 connects the U55 and the CCU 51 from the neutral state to the interface 14Vc through the interface 64. Once this connection is completed, the switching control device 3 connects the U55 and the CCU 51 from the neutral state to the interface 14Vc.
Register U153 and CCU51 in interface '3'z'iM (,' P U I K9J replacement completion report. Next, the monitoring program of CPU1 registers U153 and CCU51 in the configuration table 71, CP[J2
After starting the IPL, task B is started, and U54 and (:CU32 are registered in the configuration table 92 of the switching control device 3 and the configuration table 82 of the CPU 2, respectively, and both CPUs
The U and I10 configuration table of the switching control device 6 is shown in FIG. 2 tal.

Next, as described above, C and PUl transfer business A to CP.
Assume that while U2 is executing job B, job A needs to newly use file device 55 due to a request from terminal device lr6 connected to ccU51. Therefore, the CPU 1 sends a switching command for U55 to the CPU 1 to the switching control device 6 through the interface 13. Thereafter, the operation is the same as above, U55 is connected to the interface 14, and after the switching is completed, the switching control device 3 changes the configuration table 9.
CI'[
Jl's monitoring program registers 55 in the configuration table 71. Here, the Ilo configuration table of both CPUs and the switching control device 6 is as shown in FIG. 2 Fbl, and business A can use F55. Note that after using the work AfJ"-Fssv, a disconnection command may be sent to the switching control device 6 through the interface 13 to disconnect the connection to C)'U1 of the do 55. The Ilo disconnection command from the CPU is sent to the switching control device 6 When sent to the I10
is disconnected from its CPU and returns to the 211 state, resulting in configuration table 91 and configuration table 71.
U55 should be deleted from this.

Next, while the system is operating in the state shown in Figure 2(b), the CP
Assume that a failure occurs in U1. The switching control device [t6 that detected this failure communicates with the CPU 2 through the interface 23
Reported a problem with CPU1.

Next is the state of Ilo, which was connected to CPU1.

That is, the contents of the configuration table 91 are communicated to the CPU 2. The monitoring program of CPU2 is sent to CPU1 by communication from switching control device i3.
51 is connected, it sends a switching command for these Ilo to the switching control device 6. In this case, the CPU 2 sends a one-way switching command to the switching control [l] device 3, instructing it to switch all Ilo connected to CP and U1 to CIPU2 with one command.

The switching control device 3 instructs the changeover switches 43, 41 and 45Vc through the interface 34,
Switch these Ilo to CPU2.

When the switching is completed, the switching control device 6 adds F5M, CCU51, and F55 to the configuration table 92, and deletes all the entries from the configuration table 91.

In addition, the monitoring program of CPU 2 that received the switching completion report writes F56° CCU31 and F5 in the configuration table 82.
Add 5. Here, the CPU 2 starts a new task A, and the I10 configuration table and the 110 configuration table of the switching control device 3 become as shown in FIG. 2 (C1).

As described above, according to this embodiment, the Il used in each business
It is not necessary to set a fixed switching procedure for the The system configuration can be changed automatically and in a short time in response to unplanned business interruptions, resulting in an extremely flexible system.

In addition, since each I10 is connected only to the CPU that uses that Ilo, it is protected from hardware malfunctions of other systems C1-'U, software bugs, and accidental or intentional file destruction, resulting in high reliability. system can be realized.

Also, when a failure occurs in C1-'U1 while CPU1 and CPU2 are executing their respective tasks.

CPU2 is CPU1's preset (1mffl and rjll CP
I added during processing to the task being processed in U1
10, that is, all I10s connected to CPU1, including F55, are reliably and automatically switched to CPU2, and processing of business A can be continued with CPU2. In this way, the system configuration can be changed without operator intervention, so the changeover time is extremely short (
A system with a high operating rate can be obtained, and in particular, a system that operates continuously for 24 hours can be realized economically.

Note that it is easy to extend the above embodiment to a system with three or more CPUs. In this case, when one CPU backs up all other CPUs, the switching control device f5 has an I10 configuration table as shown in Fig. 2 for the number of CPUs, and a spare CPU You can decide the order in advance. Chobe-C (
7) Since the I10 configuration table of the CPU can be managed, each CPU does not need to be aware of other CPUs when the system is normal, and processing is simplified and overhead is small. Further, in such a case, if it is decided that one CPU is to be backed up from another specific CPU, the amount of processing capacity required as a standby machine can be reduced.

By the way, in the description of the above embodiment, for example, business A
No mention was made of the operation of allocating the F55 to service A by software when the file device F55 is connected to the CP U1 due to the need for this. Also CPU1
No mention was made of the start of job A in CPU 2 when a failure occurs in CPU 2 and CPU 2 takes over the processing of job A. For the former, depending on the operating system used in the system, press F5 during execution of a task.
5 may be newly assigned to the job A, and F'55 may be deallocated from the job A before F55 is separated from the CPU 1. Regarding the latter, it is sufficient to use the checkpoint/restart function provided by the operating system and the program, so a detailed explanation thereof will be omitted.

〔Effect of the invention〕

According to the present invention, there is an effect that switching between a plurality of processing devices and input/output devices, that is, configuration switching can be performed dynamically.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the system configuration in an embodiment of the present invention, and FIG. 2 shows 110f! of each processing device and switching control device. FIG. 2 is a diagram showing a configuration table and changes in its contents. 1.2... Processing device, 6... Switching control device, 41~
45... Selector switch, 51-55... Input/output device, 601. Terminal device-13, 1423, 24, 3'4
...Interface, 71.82191192...
configuration table. Representative Patent Attorney Akio Takahashi 1st Menbori

Claims (1)

[Scope of Claims] In a configuration switching method for controlling switching between a plurality of input/output devices and a processing device for a number of bales that share the input/output device, the input/output device is switched according to a request from the processing device. It has a configuration switching means that is connected to a request source processing device and holds information about the input/output device connected to each of the processing devices, and each of the processing devices is connected to its own processing device. retains information about the input/output device that is currently in use, and when any of the processing devices becomes aware of a failure in another processing device that should be backed up, the processing device is transferred from the configuration switching means. A configuration switching method, characterized in that the configuration switching means is requested to switch the connection of the input/output device based on the input/output device connection information regarding the other system processing device.