JPH0993244A - Multiple system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To select an address applicable to a protocol using a logic address by acting a computer of a slave system like a master system instead on the occurrence of a fault in the computer being the master system. SOLUTION: A computer 1 conducts the service as a master system at the operation start of a duplicate system, and a computer 2 is a slave system taking over the service on the occurrence of a fault of the computer 1 being the master system. The computer 1 is provided with a LAN adaptor 21 and a communication software 45, the adaptor 21 is provided with a ROM 31 and the ROM 31 stores specific physical addresses. Furthermore, the computer 2 acting as a standby system is provided with a LAN 22 and the communication software 45, and the adaptor 22 is provided with a register 36. The register 36 writes specific physical addresses to the adaptor 21 to take over the service when the computer 1 is failed. Furthermore, logical address setting files 41, 42 of the duplicate system are set in advance and the computers 1, 2 have the same contents independently.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a network system in a multi-system system which is connected to a network and comprises a plurality of computers which back up each other
The present invention relates to an address switching device.

[0002]

[Prior art]

Conventional example 1. Japanese Patent Laid-Open No. 59-72253 discloses an example of a system switching device for switching data input between a plurality of computers in a multi-system computer system. FIG.
A block diagram is shown in FIG. In the figure, 100 is a data communication network, and 200a, 200b and 300 are computers,
200a and 200b constitute a dual computer system that backs up each other. 500a, 500b,
500c are computers 200a, 200b, and 300, respectively.
And the data communication network 100. Each of the communication nodes 500a and 500b is provided with two address registers, and arbitrary addresses can be set in these address registers by program control from the computers 200a and 200b. Calculator 2
If both 00a and 200b are normal, the computer 20
0a executes the process A, and the computer 200b executes the process B. However, when any of the computers has an abnormality and the operation is stopped, the normally operating computers execute the process A and the process B. In the computer 300, destination addresses AD-A and AD-B are designated and sent to the data related to the processes A and B, respectively.
When both the computers 200a and 200b are operating normally, AD- is stored in the address register of the communication node 500a.
The addresses A and AD-X are set, and the addresses AD-B and AD-X are set in the address register of the communication node 500b. AD-X is an invalid address and even when the corresponding address register is in the inoperative state,
Equivalently, it is considered that an invalid address is set. In such a state, the data related to the process A is calculated by the computer 2
00a, and the data related to the process B is input to the computer 200b. At this time, when the computer 200b detects that the computer 200a is in an abnormal state and has stopped operating, the computer 200b sets the addresses AD-B and AD-A in the address register of the communication node 500b. However, the address of AD-B has already been set, and the content of the register in which the address of AD-X has been set may be changed to AD-A. The data related to the process B in which the destination address of AD-B is specified,
Data concerning the process A in which the destination address of AD-A is designated is also input to the computer 200b. Therefore, the computer 300 does not require any operation in this switching. Further, in this case, even if AD-A and AD-X are held in the address register of the communication node 500a as they are, the remaining AD-A interferes with the operation of the AD-A set in the address register of the communication node 500b. May be left as it is, or the addresses of AD-X and AD-X may be set in the communication node 500a under the control of the status signal of the computer 200a or the computer 200b.

Conventional example 2. Next, a conventional example relating to a load share type dual system that performs communication using the TCP / IP protocol will be described. Constructing a load-sharing dual system 2
One computer provides different services to other computers, and when one computer goes down, the other computer takes over the service. Conventionally, there are the following two methods for the switching processing in the load sharing type dual system. (1) Method of not taking over addresses (2) Method of taking over addresses In method (1), the requesting computer changes the service request destination that was being performed by the down computer after the system switchover. Also, in order to change, it is necessary to know the address of the change destination. Therefore, all computers that use dual services have these functions. Method (2) is further divided into the following two. (2a) Method of taking over only logical address (2b) Method of taking over logical / physical address

Next, a conventional example of the method (2a) of taking over only the logical address will be described with reference to the drawings. FIG.
2 shows the configuration diagram. In the figure, computer 1 and computer 2 are
It is a load-share type dual system that communicates using the TCP / IP protocol. The computer 1 and the computer 2 forming the dual system provide different services to other computers, and when one computer goes down, the other computer takes over the service. Computer 3 is connected to the network,
It communicates with the computer 1 and the computer 2. The computer 1 is equipped with a network interface 1a and a network interface 1b. The computer 2 is equipped with a network interface 2a and a network interface 2b. The network interface 1b is used to take over the network interface 2b, and the network interface 2a is used to take over the network interface 1a.

FIG. 23 shows the structure of an IP address. FIG.
As shown in FIG. 3, the IP address is composed of a network section and a host section. To identify multiple network interfaces, the network part must match the network address of the LAN to which it is connected. For this reason, multiple networks provided in one computer
The interface cannot be connected to one LAN. That is, as shown in FIG.
LA with multiple network interfaces
Two N are used, and an inter-LAN connection device is required. For example, in order to use the IP address 133.141.75.1, the LAN # 1 has the IP address 133.142.
LAN # 2 is used to use 75.2. Therefore, even if two LANs are provided, they cannot be used to improve the reliability of the LAN due to duplication. In addition, TCP / I
The P software holds the correspondence between the logical address and the physical address known in the past communication in the address conversion table and uses this as a cache for solving the physical address. FIG. 24 shows an address conversion table that defines the correspondence between logical addresses (IP addresses) and physical addresses. At the time of system switching, the IP address 133.141.
The physical address corresponding to 75.1 is the physical address # 1a.
From the physical address to the physical address # 2a. Therefore, all computers and L that use dual services
The address conversion table of the inter-AN connection device is updated.

Next, (2b) a method of taking over a logical / physical address will be described. The method (2b) uses a LAN adapter having a physical address changing function. By taking over the IP address and the physical address at the same time, it is not necessary to update the address conversion table. However, a plurality of LANs are required for a plurality of network interfaces provided in one computer.

[0007]

In the prior art example 1, the physical address can be switched when the system is taken over from one computer constituting the multiplex system to the other computer, but the logical address can be switched concretely. Not mentioned. Therefore, when using a logical address, there is a problem that it cannot be realized. Also, after the system switching, the communication node performs processing for two addresses for the two processes A and B. Therefore, the service of two processes may go down due to the failure of the communication node,
There was a problem of performance degradation due to concentration of communication load on communication nodes. Further, although the method of switching over the physical address and taking over the address has been described, no consideration is given to making the physical address unique in order to avoid address duplication. Therefore, duplication with the physical address of another computer connected to the LAN, or
There was a risk of duplication with the physical address of the computer newly connected to the LAN after the system operation started.

In the method of not taking over the address shown in the second conventional example, it is necessary to change the service request destination, which was performed by the down computer after the system switching, on the requesting computer. Also, in order to change, it is necessary to know the address of the change destination. Therefore, there is a problem in that all the computers that use the dual service must have these functions. Further, with the method of taking over only the logical address, it is not possible to connect a plurality of network interfaces provided in one computer to one LAN. Therefore, you have to use multiple LANs,
There is a problem that an inter-AN connection device is also required. In addition, since there is an address conversion table that defines the correspondence between logical addresses and physical addresses, it is necessary to update the address conversion tables of all computers and LAN-to-LAN connection devices that use the services of multiple systems during system switching. There was the problem of being time consuming and uncertain. Further, the method of taking over the logical / physical address has a problem that a plurality of LANs are required for a plurality of network interfaces provided in one computer.

The present invention has been made to solve the above-mentioned problems, and for example, a multiplex system capable of address switching applicable to a protocol using a logical address such as TCP / IP. The purpose is to provide. It is another object of the present invention to provide a multiplex system in which a plurality of LAN adapters are used and a plurality of services that have been taken over do not simultaneously go down due to a failure of the LAN adapter. Another object of the present invention is to provide a multiplex system which can prevent the performance from deteriorating due to the concentration of loads on the LAN adapters by using a plurality of LAN adapters.
Another object of the present invention is to provide a multi-system system that can avoid address duplication. It is another object of the present invention to provide a multiplex system capable of notifying a physical address and a logical address by intersystem communication between computers forming a multiplex system. It is another object of the present invention to provide a multi-system system in which a plurality of network interfaces provided in one computer can be connected to one LAN by utilizing the subnet function of TCP / IP protocol.

[0010]

A multiple system according to the present invention comprises a network, a main computer and a slave computer connected to the network, and the main computer is A non-volatile memory for storing a unique physical address and a storage memory for storing a unique logical address are provided, and a slave computer is provided with a register for storing a physical address and a storage memory for storing a logical address. The computer as the system uses the physical address stored in the non-volatile memory of the computer as the master system for communication, and the computer as the slave system uses the physical address of the computer as the master system as the slave system. In the register of the computer that becomes the master computer, and the logical address of the computer that becomes the master computer is set in the storage memory of the computer that becomes the slave computer, causing a failure in the master computer. If, a the slave computer is characterized in that an alternative main system computer serving as the main system.

In the multiplex system according to the present invention, the main computer is further provided with a register for storing a physical address, and the slave computer is further provided with:
With a non-volatile memory that stores a unique physical address,
The master computer sets the physical address of the slave computer in the register of the master computer,
When the logical address of the slave computer is set in the storage memory of the master computer, and one of the computers fails, the other computer takes over as the master instead of the other computer. It is characterized by

A multiple system according to the present invention comprises a network having a network address, a master computer connected to the network, and a slave computer, and the slave computer is the master computer. The physical address and logical address of the computer that becomes the system is stored, and the physical address and the logical address of the computer that becomes the slave are stored.
The master computer allocates a predetermined number of bits of the logical address to the network address and has at least one network interface in the master computer, and the slave computer has the logical address of the logical address. When more than the predetermined number of bits are assigned to the network address, the slave computer has a plurality of network interfaces, and when the master computer fails, the slave computer It is characterized in that it becomes a main system instead of the computer as the main system.

The master computer in the multiplex system according to the present invention further stores the physical address and logical address of the master computer, and stores the physical address and logical address of the slave computer. Storing, allocating more than the predetermined number of bits of the logical address to the network address, a plurality of network interfaces are provided in the main computer, and one of the computers fails. In this case, the other computer is the main system instead of the one computer.

The multiple system according to the present invention comprises N computers, one computer as the above-mentioned slave, and the computer as the above-mentioned slave has a failure in any of the remaining N-1 computers. When it occurs, it is characterized by substituting the computer in which the failure occurred.

In the multiple system according to the present invention, the subordinate computer further comprises a plurality of network
It is characterized by having an interface and being a slave to a plurality of computers.

A multiple system according to the present invention is characterized by including N computers, each of the N computers having a plurality of network interfaces so that they can be slaves of other computers.

The multiple system according to the present invention further comprises an intersystem communication device and intersystem communication for notifying the physical address and the logical address between the master computer and the slave computer. It is characterized by having a path.

The multiple system according to the present invention is further characterized in that the operation of the computer, which is the substitute computer, is stopped.

The multiple system according to the present invention is further characterized in that, when the computer operates, when the computer of the main system is operating, it becomes a slave of the computer of the main system.

A multiplex system according to the present invention comprises a network, a first computer and a second computer connected to the network, and the first computer and the second computer.
When one of the two computers fails and the other computer replaces the other computer, the other computer stops the operation of the one computer.

A multiple system according to the present invention comprises a network, a first computer and a second computer connected to the network, and the first computer and the second computer.
When one of the two computers is operating, and the other computer is operating, the operating computer is a slave of the operating computer.

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. In this embodiment, it is premised on a multiplex system including computers connected by a LAN (Ethernet). A multi-system system is composed of a main computer and a sub computer, and when a computer operating as the main system goes down, the sub computer takes over the service provided by this computer. In this embodiment, the computer that has taken over the service has an address on the same network as the computer that went down, so that another computer on the LAN that requests the service to the multiplex system can change the network configuration when the service is taken over. It is characterized by being able to continue requesting services to multiple systems without being aware of changes. The main computer also includes a LAN adapter having a ROM that stores a physical address unique to the LAN adapter. The slave computer is characterized by including a LAN adapter having a register for taking over the unique physical address of the master computer.

FIG. 1 shows an outline of a network configuration in which the dual system is connected in this embodiment. The computer 1 and the computer 2, which form a dual system, are connected to the same LAN 11. LAN11 to which the dual system is connected is L
Inter-AN connection device 13, wide area network 15, inter-LAN connection device 1
4 is connected to the LAN 12. The computer 9 requesting the service to the dual system is connected to the LAN 11, and the computer 10 is connected to the LAN 12. When a LAN is used, a unique network address is assigned to every computer that communicates, and this address is used to specify a destination / source computer during communication. In the figure, address 1 is assigned to computer 1, address 2 is assigned to computer 2, address 9 is assigned to computer 9, and so on. The address on the network is actually composed of a physical address unique to the LAN adapter and a logical address managed by the communication software. Therefore, the address 1 in FIG. 1 is the physical address 1 and the logical address 1.
Consists of

FIG. 2 shows a configuration diagram of the computer 1 and the computer 2 which compose the dual system. The computer 1 and the computer 2 are hot standby type dual systems. When the operation of the dual system is started, the computer 1 serves as the main system. Calculator 2
Is a subordinate system that takes over the service when the main computer 1 goes down, but is a standby system that does not perform its own service as the computer 2 while the computer 1 is operating as the main system.
The computer 1 operating as the main system is a LAN adapter 21,
The communication software 45 is provided. The LAN adapter 21 has a ROM 31, and the ROM 31 has the LAN adapter 21.
The physical address unique to each is stored. The computer 2 that operates as a standby system includes a LAN adapter 22 and communication software 45. LAN adapter 22 is register 3
6 is provided. The register 36 writes a physical address unique to the LAN adapter 21 in order to take over the service of the computer 1 when the computer 1 goes down. Further, the dual logical address L is set in advance in the address setting file. Both computers independently have the same contents in the address setting file. The computer 1 includes an address setting file 41 and the computer 2 includes an address setting file 42. Computers other than the dual system use only L as the logical address of the dual system.

FIG. 3 shows a processing flow of the computer 1 which is the main system. In step S10, the computer 1 sets the logical address L read from the address setting file 41 in the communication software 45. In step S12, communication processing is performed using the physical address P stored in the ROM 31 of the LAN adapter 21. FIG. 4 is a flow of processing of the computer 2 serving as a standby system. Step S14
At 2, the computer 2 sets the logical address L read from the address setting file 42 in the communication software 45. The same logical address is set in the computer 1 and the computer 2. In step S16, it is determined whether the main system is down. If YES, that is, if the main system goes down, the process branches to S18. In step S18, the physical address P is set in the register 36 of the LAN adapter 22. In this case, the operator knows the physical address P of the computer 1 in advance and sets it in the register 36. In the computer 2 of the standby system, just like the computer 1 of the main system, the logical address is set immediately after starting, but the physical address is not set. After detecting the down of the computer 1 of the main system, the physical address is written in the register and communication is started with the same address as the main system before the down. However, if a physical address is set in the register 36 of the LAN adapter 22 but is not used for communication as the physical address of the computer 2, this step may be placed before S16. In step S20, the computer 2 takes over the processing of the computer 1.

As described above, in this embodiment, the standby computer that takes over the service takes over both the physical address and the logical address of the main computer that went down. As the logical address, the same logical address is set in the communication software of both computers. Further, the main computer sets a physical address unique to the LAN adapter in the ROM and uses it for communication. When taking over the standby computer, the physical address of the main computer is taken over by setting the physical address of the main computer in the register of the LAN adapter. This enables the dual system to continue the service provided by the computer that has gone down with exactly the same address as before the downtime. Therefore, other computers requesting service to the duplex system can continue communication without incorporating the function related to the address change and without being aware of the network configuration change due to the takeover. On the other hand, transparent address takeover is realized. Further, in that case, it is possible to avoid the use of an illegal, ie, duplicate physical address.

Embodiment 2. In this embodiment, each computer forming the dual system is provided with a ROM for holding a physical address unique to the LAN adapter and a LAN adapter having a register for setting the physical address,
Both computers can be the main system hot standby type 2
The heavy system will be described.

FIG. 5 shows a block diagram of a double system in this embodiment. The difference between FIG. 5 and FIG. 2 is that the LAN adapters of the computers 1 and 2 are provided with ROM and registers. That is, the LAN adapter 21 of the computer 1 is provided with the ROM 31 and the register 35. ROM 32 and register 36 in LAN adapter 22 of computer 2
Is provided. Similar to the above embodiment, it is a hot standby type dual system composed of a computer 1 and a computer 2. Address setting file 4 for the logical address L of the duplex system
It is preset to 1, 42. Both computers independently have the same contents in the address setting file, and computers other than the dual system use only L as the logical address of the dual system. LAN in the ROM 31 of the main computer 1
A physical address P unique to the adapter 21 is set. The communication software 45 reads the physical address P from the ROM 31 and sets it in the register 35. L
The AN adapter 21 performs transmission / reception using the value P set in the register as a physical address. Stand-by computer 2
In the ROM 32, a physical address unique to the LAN adapter 22 is set. However, when taking over the computer 1, the operator sets the physical address P used in the computer 1 in the communication software 45, and the communication software 45 sets this physical address P in the register 36 of the LAN adapter 22. In this way, the standby computer takes over the primary logical address and physical address.
Although the computer 1 is the main system and the computer 2 is the standby system in FIG. 5, the LAN adapter 21 and the LAN adapter 22 are used.
Both have a ROM and a register, so both computers can be the main system.

FIG. 6 shows a processing flow of the computer 1 or the computer 2. Both the computer 1 and the computer 2 can be the main system, but as shown in FIG. 5, the computer 1 is the main system and the computer 2 is the standby system. First, the processing flow of the computer 1 will be described with reference to FIG. In step S22, the computer 1 reads the logical address L from the address setting file 41 and sets it in the communication software 45. Step S
At 24, it is determined whether or not the own machine works as the main system,
Since YES, the process branches to step S26.
In step S26, the communication software 45 of the computer 1 reads the physical address P from the ROM 31 of the LAN adapter 21, and then stores the physical address P in the register 3
Set to 5. In step S28, processing is performed as the main system.

Next, the processing flow of the computer 2 will be described with reference to FIG. In step S22, the computer 2 reads the logical address L from the address setting file 42 and sets it in the communication software 45. In step S24, it is determined that the computer 2 does not work as the main system, the result is NO, and the process branches to step S30. Step S3
At 0, it is determined whether the main system is down. YES,
That is, when the down of the main system is detected, step S
32 processing is performed. In step S32, the operator sets the physical address P of the main system in the communication software 45 of the computer 2. Then, the communication software 45 sends the physical address P of the main system to the register 3 of the LAN adapter 22.
Set to 6. In step 34, the computer 2 takes over the processing of the main system by using the physical address P and the logical address L of the computer 1 of the main system that has gone down.

As described above, in this embodiment, both the registers and the ROM are provided in the LAN adapter of the computer forming the dual system. The main system sets the logical address read from the address setting file in the communication software and
The physical address read from the ROM of the AN adapter is set in the register. In the standby system, set the logical address in the same way as the main system, write the physical address of the main system to the register after the down of the main system is detected, and start communication with the same physical and logical address as the main system before the down. . In this way, even after taking over the service, the dual system can perform the service with the same logical address and physical address as before the taking over, and the address takeover that is transparent to computers other than the dual system is realized. To be done. In addition, in that case, duplicate physical addresses are not used. Further, since the computers forming the dual system have the same configuration, either computer may be the main system.

Embodiment 3. In this embodiment, a hot standby type dual system that uses an intersystem communication device and an intersystem communication path between the main system and the standby system and does not use duplicate physical addresses between the main system and the standby system. Describe the system.

In the above embodiment, since the standby system has no means for automatically knowing the physical address of the duplex system, it is necessary to rely on operator input or the like. Therefore, there is a risk of physical address duplication in the standby system or physical address mismatch due to an input error. Also, because the contents of the address setting file differ between the main system and the standby system,
There was also the risk of a logical address mismatch. Therefore, in order to avoid these risks in this embodiment, an inter-system communication device and an inter-system communication path are provided between the main system and the standby system, and the physical address and the logical address are automatically provided from the main system to the standby system. A dual system for notifying an address will be described. FIG. 7 shows a configuration diagram of the double system. Differences from FIG. 5 are mentioned. A computer 1 and a computer 2 that form a dual system have an intersystem communication device 51 and an intersystem communication device 52, respectively. Inter-system communication device 5 for both computers
1 and 52 use the inter-system communication path 50 to notify the physical address and the logical address. Further, the address setting file 41 exists only in the main system (computer 1), and the logical address L
Is set.

FIG. 8 shows the flow of processing. In FIG.
The difference from the above embodiment is that the main system notifies the standby system of the physical / logical address. The processing flow in the figure can be applied to both the main computer and the standby computer.
First, a process when the computer 1 works as a main system will be described. In step 40, it is judged whether or not the computer 1 works as the main system, that is, YES, that is, since it works as the main system, S
Branch to 42. In step S42, the communication software 45 of the computer 1 reads the logical address L from the address setting file 41 and sets it in the communication software 45. In step S44, the communication software 4
5 reads the physical address P from the ROM 31 of the LAN adapter 21 and sets it in the register 35. Step S4
At 6, the computer 1 notifies the standby computer of the physical address P and the logical address L. In step S48, the process is started.

Next, the case where the computer 2 which works as a standby system is started will be described with reference to FIG. In step S40, it is determined whether the computer 2 works as a main system. Since the computer 2 is a standby system, the answer is NO, and the process branches to step S50. In step S50, the computer 2 receives the notification of the physical address P and the logical address L from the main system.
In step S52, the computer 2 sets the logical address L notified to the communication software 45. In step S54, it is determined whether the main system is down.
If it is determined that the main system is down, that is, if it is YES, the process proceeds to step S56. In step S56, the physical address P is set in the register 36 of the LAN adapter 22. In step S58, the main system processing is taken over by using the physical address P and the logical address L of the computer 1 which has gone down.

As described above, in this embodiment, the main system,
An intersystem communication device and an intersystem communication path are provided between the standby systems, and the main system notifies the standby system of the physical address and the logical address.
For this reason, address takeover that is transparent to computers other than the dual system is realized, and at the same time, since there is no operator intervention, duplicate physical addresses are used in the main system and the standby system, and between the main system and the standby system. Eliminates the risk of physical / logical address mismatch.

Fourth Embodiment This embodiment describes a hot standby type dual system that avoids physical / logical address duplication due to false recognition down by forcibly resetting the main computer after the standby system takes over the processing.

A conventional method such as heartbeat is used for monitoring between the standby systems to know whether the primary system is down. However, there is always a risk of misidentification down, that is, the standby system determines that the main system is down even though the main system is not down. When false positive down occurs,
Since the main system and the standby system have the same physical / logical address, the address duplication occurs on the network. In order to avoid this, in this embodiment, the standby system forcibly resets the main system.

FIG. 7 is used for the configuration diagram of the dual system. FIG. 9 shows the flow of processing. The difference from the processing flow of FIG. 8 is the forced reset of the main system by the standby system in step S60. The other processes are the same as those in FIG.

As described above, in this embodiment, address takeover that is transparent to computers other than the dual system is realized, and at the same time, the use of duplicate physical addresses in the main system and the standby system, the use of the main system and the standby system. The risk of address mismatch between Further, since the main system computer is forcibly reset after the processing is transferred to the standby system, there is no risk of address duplication between the main system and the standby system due to a false recognition down.

Embodiment 5. In this embodiment, when a computer is operating, it is determined whether another computer is already operating as the main system, and a hot standby type dual system that avoids address duplication due to two main systems operating Describe.

When the main system goes down and the standby system takes over the address and then the downed computer is started, two computers operating as the main system operate. As a result, the physical and logical addresses are duplicated and the operation of the dual system is hindered. Therefore,
When the main system restarts, the inter-communication device is used to determine whether the standby system has already taken over the service. Two
FIG. 7 is used for the configuration diagram of the heavy system. FIG. 10 shows the flow of processing. The difference from FIG. 9 is that in step S62 it is determined whether the main system already exists when the computer is started, and if the main system already exists, it operates as a standby system.

As described above, in this embodiment, when the main system is restarted, the inter-system communication device is used to determine whether the standby system has already taken over the service. for that reason,
It is transparent to computers other than the dual system, and when taking over to the standby system, address takeover without danger of address duplication is realized. At the same time, there is no risk of address duplication due to the operation of two main systems when the computer is restarted after it goes down. Further, since the two computers constituting the dual system alternately operate as the main system and the standby system, system switching at the time of restoration becomes unnecessary.

Sixth Embodiment This embodiment describes an N-tier system in which the computer N is a standby system for the computers 1 to N-1.

In the above embodiment, the dual system is used, and one standby system is required for each computer operating as the main system. When the multi-system is configured in this way, the number of required computers is twice as many as the multi-system. FIG. 11 shows a block diagram of the multiplex system. The computer N operates as a standby system for each of the computers 1 to N-1, which perform independent services. For example, when the computer 1 is down, the computer N is down
The computer 1 operates as a standby system for the computers 2 to N after restarting. Although one computer cannot back up multiple computers at the same time,
This method is sufficient because it is unlikely that multiple computers will go down at the same time.

As described above, in this embodiment, in the multiple system, when high reliability of two or more computers is required by multiplexing, the substantial function is kept the same as that of the dual system and the cost is reduced. Can be lowered.

Embodiment 7 FIG. This embodiment is intended for a load-share type dual system that uses the subnet function and communicates using the TCP / IP protocol.
Using the subnet function of the / IP software, simultaneous takeover of a logical address and a physical address is realized on one LAN. Further, in order to change the physical address, a LAN adapter having a register and a ROM for holding a unique physical address is used as in the above-described embodiment. The two computers that make up the load-sharing dual system provide different services to other computers,
And if one computer goes down, the other computer takes over the service.

FIG. 12 shows a block diagram of the double system in this embodiment. The computer 1 and the computer 2 are load sharing type dual systems. The computers 9 and 10 connected to the LAN 11 are computers that request a service with respect to the computers of the dual system. The network address of LAN 11 is 133.141. Calculators 1, 2,
9 and 10 each include TCP / IP software 60 and a LAN adapter. The computer 1 has two IP addresses # 1a and IP in the TCP / IP software 60.
It has address # 1b. 13 for IP address # 1a
3.141.75.1 but 13 for IP address # 1b
3.141.203.2 is set. Also, two L
L equipped with AN adapter 21a and LAN adapter 21b
The physical address # 1 is used for the AN adapter 21a, and the physical address # 2 is used for the LAN adapter 21b. IP above
The network interface 61a is configured by the address # 1a and the LAN adapter 21a, and the IP address #
A network interface 61b is configured by 1b and the LAN adapter 21b. The computer 1 uses the network interface 61a for its own service. When the computer 2 goes down, the computer 1 continues its own service through the network interface 61a,
Network interface 6 for services of computer 2
Take over using 1b. Similarly, the computer 2 uses TCP / I
The P software 60 has two IP addresses # 2a and # 2b. For IP address # 2a,
133.141.5.1, and IP address # 2b is set to 133.141.203.2. In addition, two LAN adapters 22a and 22
b, and the LAN adapter 22a has a physical address # 1.
And the physical address # 2 is assigned to the LAN adapter 22b. IP address # 2a and LAN adapter 22
A configures the network interface 62a, and the IP address # 2b and the LAN adapter 22b configure the network interface 62b.
The computer 9 is equipped with a LAN adapter 29 and PCP / IP software 60. The TCP / IP software 60 has an IP address # 3. An IP address of 133.1411.75.3 is set to the IP address # 3. The LAN adapter 29 has a physical address # 3.
Assign. Similarly, the computer 10 has an IP address #
133.141.203.4 is set as 4, and LA
The physical address # 4 is set in the N adapter 30. Calculator2, when Calculator1 is not down,
The network interface 62b is used. When the computer 1 is down, the computer 2 continues the service of the computer 1 through the network interface 62b and takes over the service of the computer 1 through the network interface 62a. In this way, only the computers 1 and 2 forming the dual system are provided with two network interfaces.
Each of the computers 9 and 10 requesting a service from the heavy system has one network interface.

TCP / IP software uses the subnet mask to learn the network address from the IP address. The subnet mask specifies how many upper bits of the IP address are considered to be the network address. Usually one LAN
All computers connected to will set the same subnet mask. However, in this embodiment, the subnet mask is set to be 1 bit longer than that set by other computers only for the computers forming the dual system. As a result, the computers 1 and 2 can each have two network interfaces, and one network interface can take over the service of its own computer and another network interface can take over the service of the other computer. Two IP addresses used by the dual system 13
The subnet mask will be described with reference to FIG. 13 by taking 3.1411.75.1 and 133.141.203.2 as an example. In a computer other than the dual system, as shown in FIG. 13A, the upper 16 bits of the IP address are used as the network section indicating the network address. In computers other than the dual system, the upper 16 bits are used for the above two IP addresses as well, so the network part is set to 1
33.141 and the same network address.

In the computers 1 and 2 forming the double system, FIG.
As shown in (b), the upper 17 bits of the IP address are used as the network section. The dual network part has two IP addresses, 133.1411.75.1 and 13
For each of the 17th bit of 3.141.203.2 "
Since it is 0 "or" 1 ", different network addresses are used. For example, taking computer 1 as an example, which of network interfaces 61a and 61b is used inside computer 1 due to the difference in network address. However, from the computers other than the dual system, the two network interfaces of computer 1 are LA
Considered to have the same network address leading to N11. By setting in this way, the computers 1 and 2 constituting the dual system can each have two network interfaces for one LAN,
One can be used for services provided by its own system, and the other can be used for taking over when the other system goes down. For example, Calculator 1 goes down, Calculator 2 is Calculator 1 and Calculator 2
In order to perform the above service, it is possible to specify which network interface, that is, the LAN adapter is to be used for transmission, by different network addresses. It should be noted that this embodiment is a case where they are backups of other computers. However, when only the computer 2 has the backup function, the computer 1
The subnet mask of the network is 16 bits
It has one interface and the subnet of computer 2
The mask is 17 bits and has two network interfaces.

For the physical address, the method described in the above embodiment is used. Although not shown in FIG. 12, the LAN adapter includes a register and a ROM. Physical address # 1
Uses the value stored in the ROM of the LAN adapter 21a and sets it in the register of the LAN adapter 21a. LAN adapter 22a of computer 2 after computer 1 goes down
By setting the physical address # 1 in the register, it is possible to take over the physical address while avoiding duplication.

FIG. 14 shows a processing flow of the computer 2. Hereafter, the service performed by the computer 1 will be referred to as service A, and the service performed by the computer 2 will be referred to as service B. In step S70, the IP address 133.
141.75.1 to the network interface 62
a and set the IP address 133.141.203.2
Is set in the network interface 62b. In step S72, the ROM of the LAN adapter 22b
Then, the physical address # 2 is read and set in the register. In step S74, the computer 2 uses service B
To start. In step S76, it is checked whether or not the computer 1 has gone down. If it is detected that the computer 1 has gone down, that is, if YES, then step S7.
The following processing is performed. In step S78, the computer 2 sets the physical address # 1 in the register of the LAN adapter 22a. The physical address # 1 is the physical address used in the communication of the computer 1. In step S80, the computer 2 takes over the service A of the computer 1 and starts it.

As described above, according to this embodiment, TCP /
One L using the subnet function of IP software
The AN has two network interfaces. Further, the LAN adapter is provided with a register and a ROM for storing a physical address unique to the LAN adapter, so that the takeover can be realized while avoiding the duplication of the physical address. Therefore, in the load-sharing dual system, address takeover can be realized with one LAN, which is transparent to other computers and does not require any change in network settings. This method can be applied to any protocol whose logical address is composed of network part and host part.
It can be applied by using the setting function of the network unit corresponding to the subnet function in TCP / IP.

Embodiment 8. In this embodiment, a multi-system system having a plurality of computers and an inter-system communication path and characterized in that one computer is replaced when any one of the other computers goes down will be described.

FIG. 15 is a block diagram showing an N-layer system consisting of N-1 computers and one computer having a backup function, where N = 4, for example. The multiple system is composed of a total of four computers 1, 2, 3 and 4. The computer 4 is a computer having a backup function. The computers 1, 2 and 3 have the same subnet mask as the computers other than the multi-system, and each have one network interface. The computer 4 has a subnet mask having one bit more than that of the other computers, and has a network interface 64a used for its own service and a backup network interface 64b.
When one of the computers 1, 2 and 3 goes down, the network interface 64b is used to take over by the method described in the above embodiment. The computers 1 to 4 are connected by an inter-system communication path 50. Then, the computers 1 to 4 respectively include an inter-system communication device 5
1 to 54 are provided. The inter-system communication path 50 allows the computer 4, which is a backup machine, to know the addresses of the other computers 1 to 3. Further, when any of the computers 1 to 3 is down, a notification is received.

FIG. 16 shows a processing flow of the computer 4. In step S82, the IP address 133.141.
203.4 is set to the network interface 64a. In step S84, the LAN adapter 24
The physical address # 4a is read from the ROM a and set in the register. In step S86, the inter-system communication path 50
By this, the addresses of the computers 1 to 3 are known. In this case, the address is an IP address and a physical address. In step S88, the computer 4 starts its own service. In step S90, it is checked whether the computer n, that is, any of the computers 1 to 3 has gone down. YE
If S, that is, one of the computers 1 to 3 goes down, the process proceeds to step S92. When any of the computers 1 to 3 is down, the computer 4 is notified by the inter-system communication path 50. In step S92, IP address 1
33.141.75. n is set in the network interface 64b. After that, n is any one of 1 to 3. In step S94, the physical address #n of the failed computer is set in the register of the LAN adapter 24b. In step S96, the computer 4 takes over the service of the computer n and starts processing.

As described above, in this embodiment, one computer in a multiplex system substitutes when any one of the other computers goes down. For example, N
In the case of a multi-system of two units, if N computers each have two LAN adapters as described in the seventh embodiment, 2N LAN adapters are required. On the other hand, in this embodiment, N + 1 LAN adapters may be provided. In a multi-system system, it is rare that a plurality of computers go down at the same time, and it is sufficient to back up a plurality of computers with one computer. Therefore, it is possible to keep the cost low by minimizing the unused LAN adapters.

Embodiment 9 FIG. In this embodiment, one computer with backup function
This paper describes a load-share type multi-system that has an interface and can take over the services of multiple computers.

In the above-mentioned embodiment, if a computer having a backup function takes over the services of other computers, it is impossible to take over the services. In the multiple system according to this embodiment, a computer having a backup function has a plurality of LAN adapters whose physical addresses can be changed, and a subnet mask having a larger number of bits than other computers, so that a plurality of network It has an interface. This allows the services of multiple computers to be taken over at the same time.

FIG. 17 shows a block diagram of a quadruple system.
In the figure, the quadruple system is composed of computers 1, 2, 3, and 4. Only the computer 4 has a backup function. The backup computer 4 includes a network interface 64a for its own service and backup network interfaces 64b, 64c, 64d. The three backup network interfaces 64b, 64c, 64d are used to take over the service when the computers 1 to 3 are down. The network interfaces 64a to 64d include LAN adapters 24a to 24d and IP addresses # 4a to #.
4d are provided respectively. The IP addresses of the computers 1 to 3 are set in the IP addresses # 4b to # 4d, respectively, and are used when the corresponding computers are down. The method of taking over the address is the same as that of the above-mentioned embodiment, but the method of using the IP address is characteristic and will be described below. A multi-system computer uses 18 bits as a network unit. As shown in FIG. 18, the subnet mask of each computer forming the multiplex system uses 18 bits so that four computers can be identified. The subnet mask of the computer other than the multiple system is 16 bits. The IP addresses of the computers 1 to 4 are 133.141.11.1 and 133.14, respectively.
1.75.2, 133.141.139.3, 133.
It is set to 141.203.4. Therefore, 17 bits and 18 bits are "00", "01", "10", and "1".
It becomes 1'and four computers can be identified. This gives 4
It is possible to recognize four network addresses corresponding to one computer. Generally, in the case of N-ary system, it can be identified by using log2N bits extra as a network part.

As described above, in this embodiment, the backup machine is provided with a plurality of network interfaces, so that the services of a plurality of computers can be simultaneously taken over. Therefore, the availability of the system is high.

Embodiment 10. In this embodiment, a load share type multiple system in which all computers have a backup function will be described. In the above embodiment, backup of a plurality of computers is possible, but if the computer having the backup function goes down, the service is stopped. Especially when the service of another computer has already been taken over, a plurality of services will be stopped at the same time.
Also, in a system in which two computers are paired so that a plurality of dual systems exist, if both computers constituting a certain dual system go down, another dual system will have a network interface for takeover. The two services that were running on the computer that went down despite being left unused will both stop. In the multi-system according to this embodiment, each computer has two network interfaces. During normal operation, one of the network interfaces is used to provide a service for itself, and a network interface for backup is also provided. The feature is that all computers can back up any other computer.

FIG. 19 shows a block diagram of a quadruple system.
The computer 1 includes an intersystem communication device 51, two LAN adapters 21a and 21b, and two IP addresses. The network interface 61a is the LAN adapter 21.
a and IP address # 1 (133.141.11.1), which is used for own service. The network interface 61b is for backup and LA
Although it consists of the N adapter 21b and an IP address, the IP address is undecided until the service of the computer that has gone down is taken over. The computers 2 to 4 have the same configuration as the computer 1. The computers 1 to 4 are connected by an intersystem communication path 50. In addition, the subnet mask of each computer constituting the multiplex system uses 18 bits so that four computers can be identified. The subnet mask of the computer other than the multiple system is 16 bits. The IP addresses of the computers 1 to 4 are 133.141.11.1, 13 respectively.
3.141.75.2, 133.141.139.3,
133.141.203.4, four computers can be identified by 17, 18 bits as shown in the above embodiment. With the above configuration, all computers can back up any other computer.

FIG. 20 shows the flow of processing when attention is paid to the computer n (n is one of 1 to 4). This processing flow is such that, when the computer n detects that another computer is down, the service n-1 (n-1 is the number of the other computer) is requested by another computer to take over the service K. In case of failure, the service K is taken over. If the service of another computer has already been taken over, the other computer is requested to take over the service to be taken over. The flow of processing will be described below with reference to the drawings. In step S100, the network interface 6n
An IP address and a physical address are set in a. In step S102, service n is started. Step S
At 104, it is determined whether another computer is down. When no other computer is down, ie N
If O, the process branches to step S106. Step S10
At 6, it is determined whether another computer has requested the service K. If another computer has not requested the service, that is, if NO, the process returns to step S104.
When another computer requests the service K in step S106, that is, when YES, step S110
Branch to. In step S110, S = K. Next, it goes to step S112. Step S104
In, if it is determined that the other computer is down, that is, YES, the process branches to step S108. In step S108, S = n-1. Here, n-1 is the number of the down computer. next,
Go to step S112. In step S112,
The computer n determines whether the service can be taken over. If service can be taken over, that is, YE
If S, the process of step S114 is performed. Step S1
At 14, the address is taken over. Address takeover is a physical address and a logical address of a computer that wants to take over the service transmitted through the inter-system communication path 50. In step S116, the service S is started. If it is determined in step S112 that service transfer is not possible, that is, if NO, the process proceeds to step S120. Step S120
At, the service S is requested to another computer. Then, the process returns to step S104.

As described above, this embodiment has described the multi-system in which all the computers have the backup function and all the computers can back up any other computer. Therefore, since the service takeover is not fixed to a specific computer, the service is continued as long as a backable computer exists in the multiple system. It is possible to take over even if half of the computers that make up the multiple system are down.

[Brief description of drawings]

FIG. 1 is a diagram showing an outline of a network configuration according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of a double system according to an embodiment of the present invention.

FIG. 3 is a diagram showing a processing flow of a main system according to the embodiment of the present invention.

FIG. 4 is a diagram showing a processing flow of a standby system according to the embodiment of the present invention.

FIG. 5 is a configuration diagram of a double system according to an embodiment of the present invention.

FIG. 6 is a diagram showing a processing flow of a dual computer according to an embodiment of the present invention.

FIG. 7 is a configuration diagram of a double system according to an embodiment of the present invention.

FIG. 8 is a diagram showing a processing flow of a computer forming a dual system according to an embodiment of the present invention.

FIG. 9 is a diagram showing a flow of processing of a dual system in the embodiment of the present invention.

FIG. 10 is a diagram showing a flow of processing of a dual system in the embodiment of the present invention.

FIG. 11 is a block diagram of a multiplex system according to an embodiment of the present invention.

FIG. 12 is a diagram illustrating TCP / in one embodiment of the present invention.
The block diagram of a double system when IP software is used.

FIG. 13 is a diagram illustrating a subnet mask according to the embodiment of the present invention.

FIG. 14 is a computer 2 according to an embodiment of the present invention.
Of the processing flow of FIG.

FIG. 15 is a diagram illustrating TCP / in one embodiment of the present invention.
The block diagram of a quadruple system when IP software is used.

FIG. 16 is a computer 4 according to an embodiment of the present invention.
Of the processing flow of FIG.

FIG. 17 is a diagram of TCP / in one embodiment of the present invention.
The block diagram of a quadruple system when IP software is used.

FIG. 18 is a diagram for explaining a subnet mask in the embodiment of the present invention.

FIG. 19 is a diagram illustrating TCP / in one embodiment of the present invention.
The block diagram of a quadruple system when IP software is used.

FIG. 20 is a computer n according to an embodiment of the present invention.
Of the processing flow of FIG.

FIG. 21 is a block diagram of Conventional Example 1.

FIG. 22 is a block diagram of Conventional Example 2.

FIG. 23 is a diagram illustrating an IP address in Conventional Example 2.

FIG. 24 is a diagram showing a correspondence between a logical address and a physical address in Conventional Example 2.

[Explanation of symbols]

Computers 1, 2, 3, 4 (constituting multiple systems), 9, 1
0 (requesting services from multiple systems) computer, 11, 1
2 LAN, 13 and 14 LAN connection device, 15 wide area network 21, 21a, 21b, 22, 22a, 22b,
23, 23a, 23b, 24, 24a, 24b, 24
c, 24d, 29, 30 LAN adapter, 31, 32
ROM, 35, 36 registers, 41, 42 address setting file, 45 communication software, 50 inter-system communication path, 51, 52, 53, 54 inter-system communication device, 60
TCP / IP software, 61a, 61b, 62a,
62b, 63a, 63b, 64a, 64b, 64c, 6
4d network interface.

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁶ Identification number Internal reference number FI Technical indication location H04L 12/26 (72) Inventor Yoshihiko Hanazaki 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Sanryo Denki Within the corporation

Claims (12)

[Claims] 1. In a multi-system system including a network and a computer serving as a main system and a computer serving as a slave connected to the network, the computer serving as the master system is non-volatile for storing a unique physical address. A computer having a memory and a storage memory storing a unique logical address, a slave computer having a register storing a physical address and a storage memory storing a logical address, and the computer serving as the master system is the master system. The physical address stored in the non-volatile memory of the computer to be used for communication, the slave computer is set the physical address of the master computer to the register of the slave computer,
When the logical address of the master computer is set in the storage memory of the slave computer and a failure occurs in the master computer, the slave computer becomes the master computer. A multi-system system characterized by becoming the main system instead. 2. The main computer is further provided with a register for storing a physical address, and the slave computer is further provided with a non-volatile memory for storing a unique physical address. The computer to be set the physical address of the computer to be the slave in the register of the computer to be the master,
When the logical address of the slave computer is set in the memory of the master computer and one of the computers fails, the other computer takes over as the master instead of the other computer. The multi-system according to claim 1, wherein 3. A multi-system system comprising a network having a network address, a master computer connected to the network and a slave computer connected to the network, wherein the slave computer is the master system. The physical address and the logical address of the computer are stored, the physical address and the logical address of the slave computer are stored, and the master computer allocates a predetermined number of bits of the logical address to the network address. At least one network interface is provided in the master computer, and the slave computer allocates more than the predetermined number of bits of the logical address to the network address, and a plurality of slave computers are provided in the slave computer. With a network interface of If you, multiple systems to be the slave computer is characterized in that an alternative main system computer serving as the main system. 4. The master computer further stores the physical address and logical address of the master computer, the physical address and logical address of the slave computer, and the logical address. If more than one of the above specified number of bits are assigned to the network address and a plurality of network interfaces are provided in the main computer, and one of the above computers fails, the other computer 4. The multi-system according to claim 3, wherein one of the computers serves as a master system. 5. The multiple system comprises N computers, one computer as the slave, and the slave computer has a failure in any of the remaining N-1 computers. The multi-system according to any one of claims 1 to 4, wherein a computer in which a failure has occurred is substituted for the case. 6. The multi-system according to claim 5, wherein the slave computer is further provided with a plurality of network interfaces and serves as a slave of a plurality of computers. 7. The multi-system includes N computers, each of the N computers being a plurality of networks.
5. The multi-system according to claim 3, wherein the multi-system includes an interface and can be a slave to other computers. 8. The inter-system communication device and inter-system communication path for notifying the physical address and the logical address between the master system computer and the slave system computer. The multi-system according to any one of claims 1 to 7, further comprising: 9. The multiplex system according to claim 1, wherein the multiplex system further suspends the operation of the computer to be replaced by the slave computer. 10. The multiple system according to claim 1, further comprising a slave of the main computer when the main computer is in operation when the computer is operating. The multi-system system described in. 11. In a multi-system including a network and a first computer and a second computer connected to the network, one of the first computer and the second computer has a failure. Occurs, and when the other computer replaces one computer, the other computer stops the operation of the one computer. 12. In a multi-system including a network and a first computer and a second computer connected to the network, one of the first computer and the second computer operates. When the other computer is in operation, the operating computer becomes a slave of the operating computer.