JPH0553938A - System for switching duplex central processing units - Google Patents

Abstract

PURPOSE:To prevent service from being interrupted while confirming the normality of a bus state even when the number of times for copy as well as diagnosis is reduced concerning the central processing unit switching system at a system equipped with duplex central processing units and a main storage device. CONSTITUTION:In the case of switching duplex central processing units 10 and 11, it is confirmed whether the central processing units 10 and 11 can be switched or not, and a main storage device 21 of the same system as the central processing unit 11, for example, in a standby state is turned to a non-service state. Next, the normality is confirmed only for the central processing unit 11 in the non-service state, main storage device 21 and bus BD between both units. When the confirmed result is satisfactory, the central processing unit 11 in the non-service state is turned to the standby state, the main storage device 21 similarly in the non-service state is turned to a service state and afterwards, the main storage device 21 of the same system as the central processing unit 11 in the standby state is turned to the mode of instruction reading by using the bus BD confirmed by the diagnosis. Then, the central processing units 10 and 11 are switched.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a switching system of a central control unit in a system having a dual configuration for the central control unit and the main storage unit and having a bus intersection between the central control unit and the main storage unit. ..

[0002]

2. Description of the Related Art FIG. 7 is a block diagram showing a general switching system. The switching system shown in FIG. 7 has a network (call path) 101, subscriber terminals 101a and 10a.
1b, redundant central control units (CC; central controller) 10, 11, main storage devices (MM; main memory) 20, 21, input / output device channel control device 1
02, an input / output terminal device 103, an external storage device 104, and a magnetic tape 105.

In the exchange system having such a configuration, switching control of the network 101 is performed using the central control unit in service and the main storage unit on the master side. Thereby, the signal from each terminal can be sent to a predetermined terminal via the network 101.

Incidentally, the central control unit 1 having the above-mentioned duplex structure
0, 11, and the main storage devices 20, 21 are switched between the central control device in the service state and the central control device in the standby state for a periodic test or the like.

FIG. 8 is a diagram for explaining a conventional central control unit switching system. As shown in FIG. 8, the central control units 10 and 11 and the main storage units 20 and 21 are duplicated as follows. ing.

That is, these central control units 10, 1
The 1 and main storage devices 20 and 21 are provided in two systems, 0 system and 1 system, respectively. Further, a bus BA is provided between the central control device 10 and the main storage device 20, and a bus BA is provided between the central control device 10 and the main storage device 21. Bus BB, central controller 11 and main memory 2
A bus BC is connected between 0s, and a bus BD is connected between the central control device 11 and the main storage device 21, thereby forming a duplex configuration.

Here, the central control units 10 and 11 are connected to the main storage units 2 of both systems via the buses BA, BB, BC and BD.
0 and 21 are accessed, and there are a service state that means that the service is being used and a standby state that is not used for the service but can be switched to the service state at any time. Yes, when the central control unit of one system is in the service state, the other system is in the standby state.

The main storage devices 20 and 21 are accessed by the central control devices 10 and 11 as described above. In these main storage devices 20 and 21, one system is a master. It is possible to write and read by taking a state, and the other system can take only a write by taking a slave state, so that it can cope with switching at any time.

The central control units 10 and 11 and the main storage units 20 and 21 are not connected to each other in order to prevent interruption of service at the time of switching the central control units. The service status is taken, connection is made for each system, diagnosis is performed, and copying is performed to confirm the normality of each bus between the central control unit in the standby state and each of the main storage devices 20 and 21. It is supposed to return to the state of.

The central control units 10 and 11 and the main storage units 20 and 21 are a system of the central control units in a service state,
Even if the system and the main storage device in the master state intersect so as to be different from each other, writing and reading are performed, and a non-service state in which maintenance is performed to cope with a failure is taken.

As a result, as shown in FIG. 9, the central control unit 10 in the service state accesses the main memory unit 20 in the master state, such as writing and reading,
Further, the main memory 21 in the slave state can be accessed only for writing.

[0012]

However, in such a conventional central control unit switching system, as shown in FIG. 10A, the 0-system central control unit 10 is in the service state,
The 1-system central control unit 11 is in the standby state, the 0-system main storage device 20 is in the master state, and the 1-system main storage device 21 is in the slave state. When a fault is latent in the bus BC between the storage devices 20, the central controller switching causes the central controller 11 of the first system to enter the service state as shown in FIG. The central controller 11 of the system cannot fetch the instruction from the master side main memory 20 of the system 0 due to the bus failure.

In this case, the central control unit 20 cannot operate as the central control unit, and the system operation is stopped. As a result, the resuming process is started, and until the resuming process is completed, The service will be interrupted.

Therefore, as shown in FIGS. 11 (A) to 11 (G), the central control unit in the standby state is connected by connecting the central control unit in the standby state and the main storage device for each system and executing diagnosis. And confirm the normality of the bus between each main memory, and
It is possible to return to the original state.

That is, as shown in FIG. 11A, the central control unit 10 is in the service state, the other central control units 11 are in the standby state, the main storage unit 20 is in the master state, and the main storage unit 21 is in the master state. Assuming that it is in the slave state, first, as shown in FIG. 11B, the bus BB is disconnected, the central control unit 11 and the main storage device 21 are put into the non-service state, and the central control unit 11 and the main storage device 21 are connected. The state of the bus BD during the period is diagnosed.

After that, as shown in FIG. 11C, the bus BD is separated again and the bus BC is connected,
After putting the central controller 11 and the main storage device 21 into the standby state, the main storage device 21 is returned to the slave state as shown in FIG.

Then, as shown in FIG. 11E, the main storage device 21 is set to the master state, the bus BA is disconnected,
The central controller 11 and the main memory 2 by connecting the bus BC
The bus BC is diagnosed by setting 0 to a non-service state.

After that, as shown in FIG. 11 (F), the bus BC is disconnected, the bus BA is connected, and the central control unit 11 and the main storage unit 20 are put in a standby state.
As shown in (G), the main storage device 20 is set to the slave state.

When switching is performed in this manner, the central control units 10 and 11 and the main storage units 20 and 21 are put into a non-service state and execute a diagnosis to confirm the normality of a newly used bus. Therefore, service interruption can be prevented.

However, in such a conventional central control unit switching system, one of the main storage devices is placed in the non-service state at the time of executing the diagnosis, so that the main storage device is shifted to the slave state after the diagnosis is completed. In order to do so, it is necessary to perform copying twice between each main storage device, and further, the diagnosis must be executed twice in total for each system configuration change. Therefore, the method shown in FIG. 11 has a problem that the confirmation procedure is complicated and that it takes a lot of time to confirm the normality of the bus between the central control unit and the main storage unit in the standby state.

The present invention was devised in view of the above problems, and simplifies the confirmation procedure and shortens the confirmation time. Therefore, even if the number of times of copying and diagnosis is reduced, the normality of the bus state can be maintained. It is an object of the present invention to provide a central control unit switching system capable of preventing service interruption while checking.

[0022]

FIG. 1 is a block diagram of the principle of the present invention. In FIG. 1, 10 and 11 are central control units, 20 and 21 are main memory units, 3 is device state management means, and 4 Is a device control means, 5 is a diagnostic execution means, and the central control devices 10 and 11 and the main storage devices 20 and 21 have a dual configuration, and between the central control devices 10 and 11 and the main storage devices 20 and 21. Have a bus intersection at. That is, a bus BA is provided between the central control device 10 and the main storage device 20, a bus BB is provided between the central control device 10 and the main storage device 21, a bus BC is provided between the central control device 11 and the main storage device 20, and the central control device 1 is provided.
1 and the main storage device 21 are connected to each other by a bus BD, thereby forming a duplex configuration.

The device status management means 3 manages the device status, and the device control means 4 establishes a bus connection and disconnection between the central control devices 10 and 11 and the main storage devices 20 and 21. The central control units 10 and 11 that are controlled and have a dual structure are switched. The diagnosis execution means 5 confirms the normality of the buses between the respective devices.

[0024]

In the above-mentioned central control unit switching system of the present invention,
At the time of switching the central control device, first, the device state management means 3 confirms whether or not the device state is such that the central control device can be switched, and the device control means 4 makes a main memory of the same system as the central control device 11 or 10 in the standby state. Put the device 21 or 20 in the non-service state.

Next, the diagnosis executing means 5 confirms only the normality of the non-service state central control unit 11 or 10 and the main storage unit 21 or 20 of the same system and the bus between both units.

At this time, if the confirmation result is good, the device state management means 3 is used to put the non-service state central control device 11 or 10 into a standby state. Similarly, the device state management means 3 puts the non-service state main storage device 11 or 10 into the service state, and then uses the bus confirmed by the diagnosis to establish the standby state with the central control device 11 or 10. The main memory 21 or 20 of the same system is changed to the instruction read mode, and then the central control unit is switched. As a result, it is possible to switch the central control unit without interrupting service.

[0027]

Embodiments of the present invention will be described below with reference to the drawings. Now, this embodiment is also applied to a switching system having a central control unit and a main storage device having a duplex configuration as shown in FIG. 7, and in such a system,
This is a case where the central control unit is switched in a regular test or the like.

FIG. 2 is a functional block diagram showing an embodiment of the present invention. In order to implement this central control unit switching system, the device state management means 3, the device control means 4, and the like shown in FIG. The diagnosis execution means 5, the device failure detection means 6, and the device failure processing means 7 are prepared.

The central control unit 1 having a dual structure
Buses BA, BB, BC, BD are interposed between 0, 11 and the main storage devices 20, 21 as in the conventional case. Here, the device state management means 3 manages the device state, reads out the device states of the central control devices 10 and 11 and the main storage devices 20 and 21, and rewrites them. When the system configuration is changed, the device status management table 3 provided in the device status management means 3
The device state history of 0, 3-1 is changed.

Further, the device status management means 3 reads the device status history from the device status management tables 3-1 and 3-2 before switching the central control device, and the main storage devices 20 and 2 of both systems are read.
If 1 is in the service state, the central control unit switching process recognizes it, but if it is not in the service state, the switching is not performed.

Furthermore, the device state management means 3 is
It is also confirmed that the central control unit in the opposite system to the central control unit in the service state is not in the non-service state and that the main storage device in the same system as the central control unit in the standby state is in the slave state.

Further, the device status management means 3 has the device status management tables 3-0 and 3-1 shown in FIGS. 4A and 4B as described above. Tables 3-0 and 3-1 record the current states of the central control units 10 and 11 and the main storage units 20 and 21 of the 0-system and 1-system, respectively. It has state history units 3-0a and 3-1a and main memory state history units 3-0b and 3-1b.

Here, the central controller state history unit 3-0
The main memory status history unit 3-0b records the current status of the 0-system central controller and main memory.
The central controller status history unit 3-1a and the main memory status history unit 3-1b record the current status of the 1-system central controller and the main memory.

By the way, the device control means 4 performs switching between the central control devices 10 and 11 having a duplicated configuration, and further connects and disconnects a bus between the central control devices 10 and 11 and the main storage devices 20 and 21. Is.

For example, the device control means 4 is
If the main storage device of the same system as the central control device in the standby state is not in the slave state, master / slave switching between the main storage devices 20 and 21 is performed, and thus, the central control devices 10 and 11 and the main storage device 20, When the device state of 21 becomes good, the central control unit in the standby state and the main storage unit of the same system are brought into the non-service state.

Before the switching, the device control means 4 sets the main storage device in the non-service state to the service state and the non-service state if the diagnosis result for the non-service state central control device and the main storage device is normal. The central control unit is put in a standby state, then the main storage unit in the master state and the main storage unit in the slave state are switched, and then the central control unit in the service state and the central control unit in the standby state are switched.

Further, when the device fault detecting unit 6 detects a bus fault, the device control unit 4 puts the main storage device in the slave state into the non-service state according to the instruction of the device fault processing unit 7, and sets the service system. Is to be separated from.

The diagnosis executing means 5 confirms the normality of each device and the bus between each device, that is, the diagnosis of the non-service state central control device and the main storage device 20 is performed. The normality of the bus between the central control unit in the service state, that is, the central control unit in the service state and the main storage device in the master state after the switching of the central control unit is also confirmed.

The device failure detecting means 6 is the central control unit 1.
0, 11 and the main storage devices 20, 21 are means for detecting an abnormality, that is, the diagnosis executing means 5 detects a failure of the buses BB, BC which are not diagnosed.

The device fault processing means 7 is a means for disconnecting the faulty device from the service system, that is, a device which has not been diagnosed by the newly used bus after the switching (the central control device 11 is on standby). If the state is the bus BB, or if the central control unit 10 is in the original standby state, the bus BC), when the fault is detected by the device fault detection means 6, the device control means 4 is used to deactivate the main storage device in the slave state. It is a service state and is disconnected from the service system.

The operation of the embodiment of the present invention will be described below with reference to FIGS. 3, 5 and 6 by taking the switching among the central control units 10 and 11 by the periodic test as an example. In addition,
FIG. 5 and FIG. 6 are one flow chart, but due to space limitations, they have been made into two figures.

Here, the switching factors of the central control devices 10 and 11 include the reconfiguration of the system by fault detection other than the periodical test and the input of the central control device switching command. Therefore, it is only the switching by the periodic test that needs to consider the bus failure. Therefore, the switching by the periodic test among the switching of the central control units 10 and 11 will be described as an example.

(1) First, in a highly urgent situation where the central control units 10 and 11 are switched by changing the system configuration, there is no room to perform the process of detecting a bus fault by diagnosis. (2) Further, even if the device states of the main storage devices 20 and 21 of both types are not in the service state, the service may be continued by changing the system configuration.
The switching process of 0 and 11 must be performed. (3) In the case of switching the central control unit by a command, this command is not frequently input, and a potential failure is detected by performing a diagnosis at the time of switching the central control unit by a regular test. It is not necessary to consider a bus failure when switching the central control units 10 and 11 by.

Next, the operation of the present invention in the switching by the above-mentioned periodical test will be described. Routine testing is usually performed once a day to prevent potential equipment failure in the standby state. Then, the periodic test on the central control units 10 and 11 is executed by performing periodical switching because the central control units 10 and 11 have a duplicated configuration. That is, the states of the central control unit in the service state and the central control unit in the standby state are exchanged.

First, as shown in FIG. 3A, the central control unit 10 is in the service state, the central control unit 11 is in the standby state, the main storage unit 20 is in the master state, and the main storage unit 21 is in the slave state. In this case, when switching by the regular test, first, the main memory 2
It is confirmed whether the device states of 0 and 21 are both service states (steps S1 and S2 in FIG. 5). If the main storage device 20 or the main storage device 21 is not in the service state, the regular switching is stopped (steps S1-2 and S in FIG. 5).
2-2). And if both systems are in service,
Continue regular switching.

Next, in step S3 of FIG. 5, the device state of the central controller 11 which should be in the standby state is confirmed by the device state management means 3, and if it is not in the standby state, the regular switching is stopped (see FIG. 5). Step S3-2), and if it is in the standby state, the regular switching is continued.

Then, the device state management means 3 confirms whether or not the state of the main storage device 21 of the same system as the central control device 11 in the standby state is the master state (step S4 in FIG. 5). If there is, the master / slave state between the main storage devices 20 and 21 is switched by the device control means 4 (step S5 in FIG. 5). If the main storage device 21 is in the slave state, the process directly jumps to step S5 in FIG.

After that, the central control unit 11 and the main storage unit 21 are brought into the non-service state by the unit control means 4 (steps S6, S7 in FIG. 5, see FIG. 3B), and the central control unit 10 and the main control unit 10 are operated. The bus BB between the storage devices 21 is disconnected [step S8 in FIG. 5, see FIG. 3 (B)].

After disconnecting the bus BB, the bus BD between the central control unit 11 in the non-service state and the main storage unit 21 is connected [step S9 in FIG. 5, see FIG. 3 (B)], and again the non-service state. Diagnosis is executed by the diagnosis executing means 5 on the central control unit 11 and the main storage unit 21 in the state, whereby the central control unit 11, the main storage unit 21,
The normality of the bus BD is confirmed, and if the diagnosis result is normal, the regular switching process is continued [step S10 in FIG.
See Steps S12 and S13 in FIG. 6 and FIG. 3B], or if there is an abnormality, stop (Step S11-2 in FIG. 6,
S13-2).

Then, as shown in FIG. 3C, when the diagnosis result is normal, the device control means 4 causes the non-service state between the central control device 11 and the non-service state main storage device 21. The bus BD of is disconnected.

After that, by the device control means 4, the main storage device 21 in the non-service state, the central control device 1
1 is also placed in a standby state (steps S14 to S1 in FIG. 6).
6) After that, the bus BB between the central control units 10 is connected [step S17 of FIG. 6, see FIG. 3 (C)].

Next, the contents of the main storage device 20 are copied to the main storage device 21 [step S18 in FIG. 5, FIG.
(See (C)), and thereafter, the main storage device 21 is brought into the slave state out of the service state by the device control means 4 [step S19 in FIG. 6, see FIG. 3 (C)].

Although the normality of the bus BC between the central control unit 11 and the main storage unit 20 has not been confirmed by the above diagnosis, the bus BD between the central control unit 11 and the main storage unit 21 cannot be confirmed. Since the normality has been confirmed, the central control unit 11
Of the main storage devices 20 and 21 in advance so that the main storage device 21 will be in the master state when the storage device shifts to the service state.
The master / slave state is switched [see step S20 in FIG. 6 and FIG. 3 (D)]. Originally, since the bus BB between the central controller 10 and the main memory 21 is used, no problem occurs even if the main memory 21 is shifted to the master state.

Next, the central control unit 10 in the service state and the central control unit 11 in the standby state are switched. By this switching, the central control unit 11 enters the service state and the central control unit 10 enters the standby state [step S21 in FIG. 6, see FIG. 3 (E)].

In the previous diagnosis, since the bus BC between the central control unit 11 and the main memory unit 20 is not diagnosed, if the fault is latent, the central control unit 11 causes an abnormality in the main memory unit 20. Is detected [Step S2 in FIG. 6]
2, see FIG. 3 (E)].

If there is no abnormality, the switching of the central control unit is completed (step S22-2 in FIG. 6).
However, even if there is an abnormality, in this case, since the failure is in the main storage device 20 in the slave state, the failure processing program can be executed without any problems, so that the central controller 11 in the service state
The bus BC between the main memory device 20 in the slave state and the main memory device 20 in the slave state is disconnected [step S23 in FIG. 6]. After that, the main storage device 20 in the slave state is brought into the non-service state [step S24 in FIG. 6]. This completes the switching of the central controller.

The device status management means 3 for managing the device status in this way, the central control units 10 and 11, and the main storage unit 2
Device control means 4 for controlling the bus connection and disconnection with 0 and 21 and switching between the central control devices 10 and 11 having a dual configuration, and a diagnosis for confirming the normality of the bus between each device. Execution means 5 and central controller 10,
11 is provided with a device failure detecting means 6 for detecting an abnormality in the main storage devices 20 and 21, and a device failure processing means 7 for disconnecting the failed device from the service system to the device state managing means 3 to provide a central control device. 11 and the main memory 20,
In the case where a fault is latent in the buses BC and BD between 21 and the fault is brought to the surface by the switching of the central control units 10 and 11, the central processing unit 10 is simply executed by the fault processing program. , 11 and the central control unit 11 which has been confirmed by diagnosis before switching,
Since the main storage device 21 and the bus BD between the central control device 11 in the service state after switching and the main storage device 21 in the master state are used, the number of times of copying and diagnosis is reduced by one, and the confirmation procedure is Even if it is simplified and the confirmation time is shortened, the normality of the bus state can be confirmed.

Further, the time-consuming restart processing eliminates the need to restore the service of the system, and the operation between the central control unit 11 and the main storage units 20 and 21 can be performed without interrupting the operation of the central control unit 10 in the service state. Can handle bus failures.

[0059]

As described above in detail, according to the central control unit switching system of the present invention, when the central control unit is switched, it is confirmed by the device state management means whether or not the central control unit can be switched. , The device control means,
The main memory of the same system as the central control unit in the standby state is put into the non-service state, and then the diagnostic execution means confirms only the normality of the non-service state central control unit and the main storage unit and the bus between them. If the confirmation result is good, the device status management unit is used to put the non-service state central control unit in the standby state, and the device status management unit similarly services the non-service state main storage device. After changing to the state, using the bus confirmed by the diagnosis, the main memory in the same system as the standby central control unit is changed to the instruction read mode, and then the central control unit is switched. Therefore, in switching the central control unit, the confirmation procedure is simplified, and the normality of the bus state is confirmed even if the number of times of copying and diagnosis is reduced by one to shorten the confirmation time. Bets can be, This eliminates the need to perform restart processing takes time, resulting in an advantage of preventing interruption of service.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a functional block diagram showing an embodiment of the present invention.

FIG. 3 is a diagram for explaining the operation of one embodiment of the present invention.

FIG. 4 is a diagram showing a device state management table used in an embodiment of the present invention.

FIG. 5 is a flowchart illustrating the operation of the embodiment of the present invention.

FIG. 6 is a flowchart illustrating the operation of the embodiment of the present invention.

FIG. 7 is a block diagram showing a switching system.

FIG. 8 is a block diagram showing a conventional example.

FIG. 9 is a diagram illustrating an operation of a conventional example.

FIG. 10 is a diagram illustrating an operation of a conventional example.

FIG. 11 is a diagram for explaining the operation of another conventional example.

[Explanation of symbols]

 1, 10 and 11 Central control unit (CC) 2, 20 and 21 Main storage unit (MM) 3 Device state management means 3-0 and 3-1 Device state management table 3-0a and 3-1a Central control unit state history Part 3-0b, 3-1b Main memory device status history part 4 Device control means 5 Diagnostic execution means 6 Device failure detection means 7 Device failure processing means 101 Network 101a, 101b Terminal 102 Input / output device channel control device 103 Input / output terminal Device 104 External storage device 105 Magnetic tape BA, BB, BC, BD bus

Front Page Continuation (72) Inventor Yasuhiro Anezaki 3-9-18 Shin-Yokohama, Kohoku-ku, Yokohama-shi, Kanagawa Within Fujitsu Comunication Systems, Inc.

Claims (1)

[Claims] 1. The central control unit (10, 11) and the main storage unit (20, 21) have a dual configuration, and
The central controller (10, 11) and the main memory (20,
21) in a system having a bus intersection with the device state management means (3) for managing the device state, the central control unit (10, 11) and the main memory unit (20,
21) control the bus connection to and disconnection from, and
A device control means (4) for switching between the central control devices (10, 11) having a duplex configuration and a diagnostic execution means (5) for confirming the normality of the above-mentioned devices and the bus between the devices are provided. At the time of switching the central control device, the device state management means (3) confirms whether or not the device state is such that the central control device can be switched, and the device control means (4) checks the main system of the same system as the central control device in the standby state. The storage device is put into the non-service state, and then the diagnostic execution means (5) confirms only the normality of the non-service state central control unit, the main storage unit, and the bus between the two units. If there is, the device state management means (3) is used to put the non-service state central control unit in the standby state, and similarly, the device state management means (3) also sets the non-service state main storage device. Service letter After changing to the state, use the bus confirmed by the diagnosis, change the main memory of the same system as the standby central control unit to the instruction read mode, and then switch the central control unit. The feature is a dual control central control unit switching system.