JPH0895932A - System switching control method for distributed processing system - Google Patents

Abstract

PURPOSE: To provide a technique with which the service provided by an executing computer can be speedily succeeded to a reserve computer without loading any heavy burden on other computers on a communication network when any fault is generated at the executing computer while utilizing the reserve computer disconnected from the communication network when the executing computer is normal. CONSTITUTION: Concerning a distributed processing system provided with an executing computer 200 and a reserve computer 300 to be the reserve system of the executing computer 200 to be connected by a communication network 400 for a client computer 100 for executing the distributed processing with the reserve computer 300 in place of the executing computer 200 when any fault is generated at the executing computer 200, the client computer 100 previously registers address information specifying the executing computer 200 and the reserve computer 300 and attribute information distinguishing whether it is the executing system or the reserve system and when any fault is generated at the executing computer 200 of which the attribute information shows the executing system, after the erasure of address information specifying the executing computer 200 where the fault is generated, the required distributed processing is executed by the reserve computer 300 of which the attribute information shows a standby state.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is a distributed system which has an execution computer and a standby computer among a plurality of computers connected by a communication network, and when a failure occurs in the execution computer, the processing is taken over and executed by the standby computer. The present invention relates to a system switching control method for a processing system, and particularly to a system switching control method for effectively using a standby computer.

[0002]

2. Description of the Related Art Conventionally, as a system switching control method of this kind, for example, IBM (International Business
Machines Corporation HACMP / 600
0 (High Availability Clus
ter Multi-processing / 600
0), TCP / IP (Transmission)
Control Protocol / Internet
IP (InternetPr) of the execution computer connected by the communication network by the Protocol)
auto) address is taken over by the IP address of the standby computer when the execution computer fails.

FIG. 5 is a block diagram of a distributed processing system using this system switching control method. In the figure, 70
0 is a client computer in the distributed processing system, 7
40 is a client program of the client computer 700, 730 is an inquiry about the address information of the server process required by the client program 740,
And a name service that stores the newly acquired address information.

Reference numeral 800 is an execution computer for performing various online services, 810 is a switching monitor for detecting a failure of the execution computer 800 and instructing computer switching, 820 is an online control program of the execution computer 800, 82
1 is a server process for performing online service, 830
Is a name service having address information of the execution computer 800.

Reference numeral 900 denotes a standby computer that provides an online service in place of the executing computer 800 when a failure occurs in the executing computer 800, 910 a switching monitor that detects a failure of the standby computer 900 and instructs computer switching, and 920 executes An online control program that is the same as the online control program 820 of the computer 800, 9
21 is a server process that provides the same online service as the server process 821 of the execution computer 800;
0 is a name service that inherits the address information of the name service 830 of the execution computer 800 when the execution computer 800 fails.

Reference numeral 1100 is a switching communication line for performing switching communication between computers between the executing computer 800 and the standby computer 900, and a switching monitor 810 of the executing computer 800.
Is directly connected to the switching monitor 910 of the standby computer 900.

Finally, 1000 is a T connecting the computers.
It is a communication network line based on the CP / IP protocol.

FIG. 6 shows an execution computer 800 and a standby computer 9.
Address information of the name services 830 and 930 of 00, 1210 is a server name field in which the computer name in the network is entered, 1220 is an IP address field in which the computer address in the network is entered, 123
0 is an in-computer server process address field in which address information of each server process is entered.

In this case, the in-computer server address field 1230 of the address information of the name service is uniquely set in the communication network.

Next, the operation of the conventional distributed processing system will be described with reference to FIGS.

During normal operation of the execution computer 800,
Since no IP address is set in the standby computer 900, the client computer 700 cannot communicate with the standby computer 900 via the communication network 1000. That is, in FIG. 5, the communication network 1
Only the execution computer 800 is connected to the client computer 700 by 000.

In this state, a failure occurs in the executing computer 800, and when the switching monitor 810 detects this failure, it issues a computer switching request to the switching monitor 910 of the standby computer 900 through the system switching communication line 1100. By this request, the IP address field 1220 of the standby computer 900 is set to the IP address of the executing computer 800.
The address is set.

On the other hand, the IP address in the execution computer 800 is deleted, and the execution computer 800 becomes the communication network 10.
Separated from 00.

From the point of view of the client computer 700 until a failure occurs in the execution computer 800 and the system is switched to the standby computer 900, the execution computer 80 in which the failure has occurred
During the period from 0 to the transfer of the address information to the standby computer 900, all services including the inquiry of the address information of the server process 821 using the server name as a key from the name service 730 via the communication network 1000 It will stop.

However, when the deletion of the address information of the execution computer 800 and the setting of the address of the standby computer 900 are completed, all the address information of the standby computer 900 is inherited from the execution calculation. When a service request is issued to the target server process based on the address information acquired from 800, the server process of the standby computer 900 replaces the executing computer 800 and provides the target service.

As described above, according to the conventional system switching method, since the IP address is inherited when the execution computer 800 and the standby computer 900 are switched, the client program 740 can continue the processing without being aware of the system switching. Is possible.

[0017]

The present inventor has found the following problems as a result of examining the above-mentioned prior art.

In the above-mentioned conventional technique, when the executing computer 800 is operating normally, the IP address is not set in the standby computer 900, and the communication network 100
Since it was separated from 0, it was impossible to effectively utilize the standby computer 900. That is, for example, the load of the execution computer 800 becomes temporarily heavy, and there is a problem that it is impossible to distribute this load to the standby computers 900.

It is an object of the present invention to provide a system switching control method for a distributed processing system that enables effective utilization as needed while maintaining the character of a standby computer.

The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

[0021]

Of the inventions disclosed in the present application, a representative one will be briefly described below.
It is as follows.

The client computer is provided with an execution computer connected by a communication network and a standby computer serving as a standby system for the execution computer. When a failure occurs in the execution computer, the execution computer is distributed to the standby computer instead of the execution computer. In a distributed processing system for executing processing, address information specifying the executing computer and standby computer and attribute information for distinguishing between an active system and a standby system are registered in advance in the client computer, and the attribute information is executed. When a failure occurs in the execution computer indicating the system, the address information for specifying the failed execution computer is deleted, and then the standby computer having the attribute information in the standby state executes the necessary distributed processing.

[0023]

According to the above means, the standby computer can be communicated by registering in advance the address information for identifying the executing computer and the standby computer and the attribute information for distinguishing between the active system and the standby system in the client computer. It connects to the network and makes the standby computer available to other computers on the communication network.

When the client computer inquires the address information of the server process using the computer name as a key, the address information is returned from both the executing computer and the standby computer to the client computer, but the standby attribute field of the address information is set. By referring, the client computer issues a service request to the execution computer in the "execution state".

On the other hand, when the client computer causes the standby computer to execute necessary processing as necessary, it becomes possible after recognizing that the client computer is in the "standby state" based on the address information. Effective use of computer resources. In this case, a heavy-load process or a long-time process that impairs the original purpose of the standby computer can be solved by, for example, not accepting the service request from the client computer.

When a failure occurs in the execution computer, all computers on the communication network are made to erase the address information regarding the execution computer and the standby computer, while
By setting the standby attribute of the execution computer to "standby state" and the standby attribute of the standby computer to "execution state", the standby computer executes the service request issued by the client computer.

With this, it is possible to realize the system switching between the executing computer and the standby computer while the standby computer remains in the communication network.

[0028]

EXAMPLES The structure of the present invention will be described below with reference to examples.

In all the drawings for explaining the embodiments, those having the same function are designated by the same reference numerals, and the repeated description thereof will be omitted.

FIG. 1 is a block diagram showing an embodiment of a distributed processing system to which the present invention is applied. 100 is a client computer, 140 is a client program, and 130 is a name service of the client computer 100.

Reference numeral 200 is an execution computer, 210 is a switching monitor which detects a failure and outputs a computer switching signal, 220 is an online control program for the execution computer 200, 221
Is a server process under the online control program 220, and 230 is a name service in which the address information of the server process 221 is registered.

Reference numeral 300 is a standby computer, 310 is a switching monitor which detects a failure and outputs a computer switching signal, 320 is an online control program for the standby computer 300, 321
And 322 are server processes under the control of the online control program 320, and 330 is a name service in which the address information of the server process 321 is registered.

Reference numeral 500 is a system switching communication line connecting the switching monitor 210 of the execution calculation 200 and the switching monitor 300 of the standby computer 300, and 400 is a communication network using the TCP / IP protocol.

FIG. 2 shows address information possessed by the name service in the network. 610 is a server name (computer name) field, 620 is an IP address field, 630 is a server process address in computer, and 640 is a standby attribute field. is there.

Next, the operation of the distributed processing system of this embodiment will be described with reference to the operation sequence diagrams shown in FIGS. 3 and 4, and when the execution computer 200 is operating normally, it is shown in FIG. As described above, the client process 140 uses the server name as a key for the name service 130 to provide the requested service to the server process 221.
Is inquired about the address information (S301).

Then, the name service 130 checks whether or not the address information of the server process 221 is included in the previously stored address information, and if there is no corresponding address information, another computer connected by the communication network 400, Inquiry is made about the address information of the server process 221 using the server name as a key (S302, S303).

At this time, since the same server name is set in the executing computer 200 and the standby computer 300,
As shown in FIG. 3, the execution computer 200 and the standby computer 3
The name services 230 and 330 of 00 each return address information (S304, S305).

Therefore, the name service 130 uses the standby attribute field 64 of the two pieces of returned address information.
The contents of 0 are checked, and the address information of the server process 221 of the execution computer 200 in the "execution state" is acquired.

Here, the name service 130 returns the address information to the client program 140, and stores the address information in case the same request comes from the client program 140 next time (S306).

The client program 140 issues a service request to the server process 221 of the execution computer 200 based on the address information of the name service 130 (S307).

Then, next time the same request comes in,
The name service 130 causes the client computer 140 to execute the execution computer 200 based on the address information just stored.
Address information of the server process 221 of
308, S309, S310).

On the other hand, the standby computer 300 is always available.

The operation for making a processing request to the standby computer 300 during normal operation of the execution computer 200 will now be described with reference to C in FIG.

First, the client program 140 of the client computer 100 inquires of the name service 130 about the address information of the server process 322 which provides the requested service, using the server name as a key (S31).
1).

Then, the name service 130 checks whether or not the address information of the server process 322 is included in the previously stored address information, and if there is no corresponding address information, another computer connected via the communication network 400 Inquiry is made about the address information of the server process 322 using the server name as a key (S312, S313).

At this time, since the same server name is set in the executing computer 200 and the standby computer 300,
The name services 230 and 330 of the executing computer 200 and the standby computer 300 respectively return the address information (S314, S315).

Therefore, the name service 130 uses the standby attribute field 64 of the two pieces of returned address information.
By checking the content of 0, the address information of the server process 322 of the standby computer 300 in the "standby state" is obtained.
Here, the name service 130 returns the address information to the client program 140 (S316).

The client program 140 issues a service request to the server process 322 of the standby computer 300 based on the address information of the name service 130 (S317).

As a result, even if the standby computer 300 is in the standby state, the client computer 100 can cause the standby computer 300 to execute necessary processing as needed, and the computer resources of the standby computer 300 can be effectively utilized. it can.

In this case, since the standby computer 300 basically needs to maintain the character of "standby",
The client computer 100 requests processing, for example, by limiting the time so that the character “waiting” can be maintained.

As a result, the standby computer 300 has a "standby" status.
While maintaining this character, it is possible to temporarily execute the necessary processing.

Next, the case where a failure occurs in the execution computer 200 will be described with reference to FIG.
When the switching monitor circuit 210 of 00 detects a failure,
Broadcast to all computers (here, the client computer 100) in the communication network 400 so as to erase the address information regarding the executing computer 200 (S401, S402). As a result, the address information of the execution computer 200 stored in the name service 130 is deleted.

Next, the standby attribute field of the address table of the name service 230 of the execution computer 200 is rewritten from "execution state" to "standby state" (S40).
3).

On the other hand, the switching monitor circuit 210 uses the computer switching signal dedicated line 500 to connect the standby computer 300.
Communicate the occurrence of failure to. In response to this, the standby computer 300
Name service 330 rewrites the standby attribute field of the address table from “standby state” to “execution state”. Thus, the execution computer 200 to the standby computer 300
The system switching to is completed (S404, S405).

When the client program 140 of the client computer 100 inquires of the name service 130 about the address information of the server process 221 using the server name as a key after the above processing is completed, the stored address information is deleted. Therefore, another computer (execution computer 2) connected to the communication network 400 again.
00 and the standby computer 300) for the address information of the server process 221 using the server name as a key (S
406, S407, S408).

At this time, if the execution computer 200 is operating with a failure, the execution computer 200 is operated as described above.
And the name services 230 and 33 of the standby computer 300
0 returns address information (S409, S4)
10).

Then, again, the name service 130 checks the standby attribute table 640 in the address information to obtain the address information in the "execution state" among the returned address information, and checks the "execution state". The address information of the server process 321 of the standby computer 300 that is The name service 130 is the standby computer 3
The address information of the server process 321 of 00 is returned to the client program 140 and stored for the next request (S411).

Then, the client program 140
Issues a service request to the standby computer 300 based on this address information. From the next time, as shown in E of FIG. 4, the name service 130 of the client computer 100
The client program 140 directly makes a service request to the server process 321 of the standby computer 300 based on the address information stored by the client computer 140 (S412).

As described above, the standby computer 300 becomes the running state, the running computer 200 can be inspected and repaired, and the restored running computer 200 now backs up the running computer 300 in the running state. The roles of 200 and standby computer 300 will change.

However, it goes without saying that at the time when the execution computer 200 is restored, the network administrator can put the execution computer 200 into the execution state again and return the standby computer 300 to the standby state.

As is clear from the above operation, the computer resources of the standby computer 300 can be effectively utilized from the client computer 100 as needed until a failure occurs in the execution computer 200.

As described above, the invention made by the present inventor is
Although the present invention has been specifically described based on the above-mentioned embodiments, the present invention is not limited to the above-mentioned embodiments, and it goes without saying that various modifications can be made without departing from the scope of the invention.

[0063]

The effects obtained by the typical ones of the inventions disclosed in the present application will be briefly described as follows.

The standby computer is given an address for specifying the computer on the communication network, and the execution computer and the standby computer are switched by the client computer based on the address information, so that while the execution computer is operating normally, With another computer connected to the communication network,
It is possible to use it effectively as needed while maintaining the character as a standby computer.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a distributed processing system using a system switching method according to an embodiment of the present invention.

FIG. 2 is a diagram showing an address table of a name service of a computer according to the embodiment.

FIG. 3 is a diagram showing an operation sequence of the distributed processing system according to the embodiment at a normal time.

FIG. 4 is a diagram showing an operation sequence when a failure occurs in the distributed processing system according to the embodiment.

FIG. 5 is a configuration diagram of a distributed processing system using a conventional system switching method.

FIG. 6 is a diagram showing an address table included in a name service of a computer of a conventional example.

[Explanation of symbols]

100 ... Client computer, 130 ... Name service, 140 ... Client program, 200 ... Execution computer, 210 ... Switching monitor, 220 ... Online control program, 221 ... Server process, 230 ... Name service, 300 ... Standby computer, 310 ... Switching monitor , 320 ... Online control program, 321, 3
22 ... Server process, 330 ... Name service, 40
0 ... communication network, 500 ... system switching communication line,
610 ... Server name field, 620 ... IP address field, 630 ... In-computer server process address field, 640 ... Standby attribute field.

Claims (1)

[Claims] 1. A client computer, comprising: an execution computer connected by a communication network; and a standby computer serving as a standby system for the execution computer. When a failure occurs in the execution computer, the standby computer is replaced with the execution computer. In the distributed processing system for executing distributed processing, the client computer is preregistered with address information for specifying the executing computer and the standby computer and attribute information for distinguishing between the active system and the standby system. Indicates the active system.When a failure occurs in the execution computer, the address information that identifies the execution computer where the failure occurred is deleted, and then the distributed processing required for the standby computer whose attribute information is in the standby state is executed. A system switching control method for a distributed processing system, comprising: