JPH0936862A - Dynamic control system for communication channel - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the capacitor required for communication consumed for troubleshooting by limiting the number of machines required for troubleshooting. SOLUTION: Computers 1, 2 make periodic communication to detect a fault in a communication channel and communication devices and computers 1-4 monitors interruption of reception respectively. On the occurrence of interrupted reception, it is discriminated as occurrence of a fault in a communication channel. The designation as to which two sets among test communication mechanisms 5, 6 are to make notice transmission is decided by describing the designation in each of the computers 1-4 by the manager of the system in advance as initializing information. Then, 2×(n-1) kinds of transmission paths are enough for the notice transmission by limiting the number of the test communication devices to two sets. The reception function and the communication channels for all the computers 1-4 are monitored even when the number of the test communication mechanisms is limited to two sets as above. As a result, the capacity for network communication consumed by troubleshooting is reduced while keeping fault detection information.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dynamic control system for network paths in inter-computer communication.

[0002]

2. Description of the Related Art FIG. 12 shows an example of a conventional method for periodically confirming the reachability of a communication path in order to detect a failure in the communication path. 1, 2, 3, in the figure
Reference numeral 4 is a computer, and 52 is a test communication mechanism. The arrows in the figure indicate the test communication exchanged between the respective test communication mechanisms 52 and their directions. In this example, there are four computers, and each of them performs test transmission to the remaining three computers, so that a total of twelve types of communications are performed. When the number of computers becomes n, communication is performed through _n C ₂ routes, and the test communication amount increases as the number of computers increases.

Further, in order to detect a failure in the communication path,
There is a method for periodically confirming the reachability of a communication path by performing one-to-many communication, and a conventional example relating to this is shown in FIG.
In the figure, 1 is a computer, 9 is a communication device, and 10
Is a network communication path, 11 is communication software, 12
Is a communication control mechanism having a retransmitting function, 13 is a notification periodically communicated with another computer via the communication control mechanism 12, the communication device 9, and the network communication path 10, and 53 is a notification periodically between the computers. A mechanism for determining communication interruption during transmission / reception, and 54 is a message exchanged between computers via the communication control mechanism 12 and the communication device 9. The mechanism 53 for determining communication interruption receives a notification from the outside while issuing a notification to the mechanism mounted on another computer, and when the communication interruption is found, it is considered that a failure has occurred on the communication path. When the notification reception via the network communication path 10 is interrupted, the communication control mechanism 12 waits for the reception of the interrupted notification again. This is a process prepared for when the communication interruption is a temporary phenomenon. The communication control mechanism 12 determines that the communication path is broken and reports an error to the communication interruption determination mechanism 53 and the communication software 11 when the notification cannot be received even after a certain time has elapsed for restarting the notification. Therefore, from the failure detection to the communication interruption determination mechanism 5
There is a time lag until 3 knows the failure of the communication path,
The error will be notified at the same time as the communication software 11.

In order to distinguish the communication devices connected to the communication path and communicate with the target communication device, the communication device is provided with a number called an address corresponding to one-to-one. This number includes a physical address determined when the communication device is manufactured and a logical address used by the computer for identification. Since the physical device for communication uses only the physical address, a conversion table with the physical address is required when communicating with the logical address. FIG. 14 illustrates a procedure for converting the physical address and the logical address. In the figure, 1 is a transmitting side computer, 2 is a receiving side computer, 9 and 16 are communication devices, 10 is a network communication path, 17 is a transmission application, 18 is a communication mechanism, and 19 is a transmission application 17 to communication mechanism 1.
A flow of data transmitted through 8, and 20 is a conversion table of a physical address and a logical address. Communication device 9,
And 16 have physical addresses as identifiers for distinguishing them from others on the communication path. This is different for every communication device, and no two things are the same through the network. Here, the physical addresses assigned to the communication devices 9 and 16 will be referred to as PA9 and PA16. A clue for the transmission application 17 to identify the communication partner is a logical address, which can be changed by software designation. Here, the logical addresses assigned to the communication devices 9 and 16 will be referred to as LA9 and LA16. When the transmission application 17 first transmits to the receiving computer 2, the LA 16 is designated as the transmission address. At this point, the communication device 9 does not know where the LA 16 is. Therefore, a response is requested from the device to which the LA 16 is assigned via the network communication path 10. Then, the communication device 16 knows that it is called, and the communication mechanism 18 inside the receiving computer 2
Returns PA16 to the communication device 9. As a result, the communication device 9 finds that the destination of the LA 16 is the communication device 16 having the PA 16, and can transmit to the communication device 16. The same applies to the reverse direction. Further, the communication mechanism 18, which once knows the correspondence between the logical address LA16 and the physical address PA16, changes the correspondence relationship to the internal table 2
Temporarily store at 0. Until this memory is cleared, the PA 16 with respect to the LA 16 can be found only by checking the table 20 without issuing a response request again. Since the correspondence between the physical address and the logical address may change, the contents of the table 20 are stored only for a short period. This is the traditional Address Resolution
It is the operation of Protocol.

FIG. 15 is a diagram showing a system for performing communication using a duplicated network communication path.
In the figure, 10 and 23 are independent network communication paths, and 1 and 2 are computers. Further, 9 and 24 are from the computer 1 to the network communication path 10,
23 is a communication device for accessing the computer 23, and 16 and 25 are devices for accessing the computers 2 and 10 and 23. The communication devices 9, 24, 16 and 25 each have a unique logical address for identifying the device. Has been assigned.

When the computers 1 and 2 communicate using the communication path 10, they communicate via the communication devices 9 and 16. The software of the computers 1 and 2 determines the communication path by designating the logical address assigned to the communication devices 9 and 16. FIG. 16 shows the hierarchy of the communication system in the computer 1 in FIG. In the figure, 28 is a communication application that uses a network, 29 is communication library software, 30 is an OS kernel that operates basic functions for communication, and 55 is a communication path switching logic for avoiding a failed communication path. When a failure occurs in the network communication path 10 or the communication devices 9 and 16 which are intermediate routes while using the communication device 9, the OS kernel 30 of the computer 1 determines the error, and the error report is transmitted to the OS kernel 30 and the communication. Library software 29, communication path switching logic 5
It is transmitted in the order of 5. As described above, the communication path switching logic 55 is incorporated for each communication application 28 for the conventional communication path substitution processing. FIG. 17 shows a flowchart showing the basic operation of the communication path switching logic 55. In this method, the communication paths 9, 10 and 16 are checked before the communication is started, and when a failure is found, the bypass communication paths 24, 23 and 25 are selected to avoid the failure.

FIG. 18 shows a plurality of computers connected to a network communication path. In the figure, 1,
2 is a server computer, 3 is a client computer, and 9, 16, and 48 are computers 1, 2, and 3, respectively.
A communication device for communicating with. Further, 10 is a network communication path, 56 is a server status reporting mechanism, 57 is a client status detecting mechanism, and the communication devices 9, 16, 4
Each 8 is assigned a unique logical address for identifying the device. When the software of the client computer 3 communicates with the software of the server computer 1, communication is exchanged between the communication devices 9 and 48. In this case, in order to specify the communication partner server computer, the software on the client computer 3 specifies the host name of the server computer 1 and the logical address of the communication device 9 is obtained from the host name by searching the data table. Communication is performed via the communication device 9. The means for obtaining the address of the communication device 9 from the host name is uniquely given by the fixed table. FIG. 19 shows an operation flow of the state detecting mechanism 57, and the state reporting mechanism 56 reports the operating state of the computer in response to a request from the state detecting mechanism 57.

When a computer executes a service by responding to a request via a network, a response function is prepared for a plurality of computers, and a plurality of requests generated at the same time are distributed to the plurality of computers to concentrate the load. There are methods to avoid this and improve the efficiency of the work. The conventional load balancing logic realizes the state reporting mechanism 56 so that it can respond to the load state by making the operation flow of the state detecting mechanism 57 in FIG. 18 as shown in FIG.

[0009]

Since the conventional communication path control system is configured as described above, when the number of host computers connected to the network increases, the communication amount required for fault detection increases. However, as a result, there is a problem that the communication line capacity that the user can use is stressed.

Further, since the communication path used by the application and the communication path used for detecting the failure use the same path, there is a problem that an error cannot be detected earlier than the application.

Further, in the management method of the physical address and the logical address of the communication device, when the correspondence relationship between the addresses changes, the retention period of the previous address correspondence relationship has passed,
Until the investigation of a new correspondence relationship, there was a problem that communication with a communication device whose address has changed is impossible.

Further, in order to realize the alternative communication when a failure occurs in the duplicated network, the application needs to individually incorporate the logic for switching the communication path, which causes an increase in the program scale. At that time, since the logic that detected the failure abnormally terminated the existing communication path and restarted it, the communication that had been performed halfway had to be redone from the beginning, resulting in waste of processing. There was a problem.

Further, in the conventional communication system, since the correspondence between the logical address and the communication device is fixed, for example, the server computer 2 cannot intercept the communication to the server computer 1. Therefore, as long as the application on the client computer 3 specifies the logical address of the server computer 1 as the communication partner,
Even if the server computer 1 stops,
There is a problem that an attempt to establish a communication path connection only with the server computer 1 results in a failure.

Further, in the case of load balancing in the network, it is necessary to incorporate a load balancing logic into an application on each machine, which causes an increase in program size. Moreover, in order to avoid concentration of load by this method, it is not possible to allocate the execution of the service to an appropriate computer unless the load status is compared between compatible computers, and the load comparison work itself does not create a new load. There was a problem of producing a rise.

The present invention has been made to solve the above-mentioned problems, and limits the number of machines required to detect a failure, thereby reducing the communication capacity consumed in communication for this purpose. The purpose is to

Even when the number of machines changes due to the occurrence of a failure, the accuracy of the failure detection can be improved while suppressing the communication channel capacity consumed by substituting the notification function for the failure detection by another machine. The purpose is to maintain.

When a failure occurs in the communication path, the application is interrupted by avoiding the failure through the failure detection mechanism that has a faster detection speed than the communication means of the user before the application detects the error. The purpose is to run without.

When the correspondence between the physical address and the logical address of the communication device changes, the old information that should be held in the address correspondence table for a certain period of time is forcibly discarded and a new route based on the new address correspondence is used. The purpose is to enable communication to be started immediately.

It is another object of the present invention to provide each application with continuous communication without being aware of this by providing a common function for changing the communication path of the network without the application individually judging.

Further, even if the communication partner needs to be changed by introducing the concept of a virtual logical address that does not correspond directly and fixedly to the communication device, it is possible to continue without changing the logical address each time. The purpose is to obtain a mechanism for dynamically changing a virtual address and a logical address to enable communication.

Further, even if the load comparison is not performed by the application itself for the purpose of load distribution among the computers, the load distribution is performed among the host computers, and it is necessary to change the machine for executing the provided service. Also, the purpose is to allow the application to receive the target service without being aware of this.

[0022]

According to a first aspect of the present invention, there is provided a dynamic control system for a communication path, wherein in a plurality of computers connected to a network, the computer executes failure detection of the communication path and execution and stop of notification transmission. The test communication mechanism having the function is provided, and the communication capacity of the network required for the test communication is reduced by limiting the number of transmitting computers based on the execution and the stop function of the notification transmission held by the test communication mechanism. .

In a second aspect of the invention, in the dynamic control method of the communication route according to the first aspect, the test communication mechanism reduces the test communication amount when some of the computers connected to the network are stopped. Instead, the accuracy of failure determination can be maintained.

According to a third aspect of the present invention, there is provided a dynamic control system for a communication path, in a plurality of computers connected to a network, the computer has a communication control mechanism having a communication protocol having a retransmission function and a retransmission function. By providing a communication failure detection mechanism that realizes a high-speed communication protocol function without having it, and a communication failure detection mechanism that realizes a high-speed communication protocol function notifies the failure avoidance mechanism of a failure previously found on the communication path, The alternative communication path can be provided at the timing when the communication control mechanism having the retransmission function can resume communication.

According to a fourth aspect of the present invention, there is provided a dynamic control system for a communication path, wherein a plurality of computers connected to a network correspond to a communication mechanism which realizes the dynamic path control and a logical address and a physical address. Address correspondence confirmation mechanism that dynamically checks the
An address correspondence table for temporarily storing the correspondence result between the physical address and the logical address, and an initialization mechanism for forcibly initializing the correspondence between the addresses defined in the address correspondence table are provided. A dynamic control method of a communication path characterized in that it can be used immediately when the address allocation changes by initializing without waiting for the regular update timing of the old address correspondence information.

According to a fifth aspect of the present invention, there is provided a dynamic control system for a communication path, in a plurality of computers connected to a network, the computer has a communication failure avoidance mechanism, and the communication failure avoidance mechanism communicates with a logical address. By implementing a communication path switching mechanism that switches the communication path to which the device is connected in the operating system, it is possible to switch the communication path from the application that uses the operating system function without individually incorporating the path switching logic. Is.

According to a sixth aspect of the present invention, there is provided a communication path dynamic control system, wherein in a plurality of computers connected to a network, the computer has a communication failure avoidance mechanism in which a virtual logical address independent of a logical address is allocated. Since the communication path switching mechanism that constitutes the communication failure avoidance mechanism is equipped with a virtual address receiving mechanism, even if a failure occurs in the network route, the alternate route is automatically selected by designating the communication partner by the virtual address. It was made to be done.

According to a seventh aspect of the present invention, there is provided a dynamic control system for a communication path in a system in which a plurality of computers providing the same service and a computer using the service are connected to a network, and the service providing computer is a logical system. When a computer providing a service is switched by holding a virtual address independent of the address and having a virtual address receiving mechanism for receiving the virtual address and a virtual address assignment control mechanism for assigning the virtual address. Also,
The service-using computer is designed to be able to receive services continuously without being aware of switching.

According to an eighth aspect of the present invention, in the communication path dynamic control system according to the seventh aspect, in a system in which a plurality of computers providing the same service and a computer using the service are connected to a network, The address allocation mechanism has a function of judging the load condition accompanying the service provision, and when the load fluctuation occurs, the virtual address is re-allocated to a computer with a low load, so that the load utilization can be performed without interrupting the service using computer. The adjustment is made possible.

[0030]

The dynamic control system of the communication path according to the present invention is capable of monitoring and detecting a failure on the communication path in a plurality of computers connected to the network, executing the test transmission by the communication mechanism, and using the stop designating means for the number of test transmitting machines. By limiting the above, the consumption of the communication path capacity is suppressed.

Further, in order to avoid a state in which the failure of the communication path cannot be detected due to the stop of the computer transmitting the test communication, the amount of test communication exchanged on the network is monitored and the total amount of test communication is When it is determined that the number is less than the minimum required, the limit of the computer for which the test transmission is limited is released to maintain the total amount of the test communication. On the other hand, when the test communication greatly exceeds the required amount and the consumption of the communication path capacity increases, a partial stop process is performed to dynamically suppress the test communication amount to a necessary and sufficient amount.

Further, by providing a communication mechanism for performing normal protocol communication while using the same physical communication path and a mechanism for performing test communication in a protocol format with a fast response speed, the failure on the communication path is earlier than the user application. Detect in stages.

Further, the correspondence between the logical address and the physical address of the communication partner is dynamically checked, the change of the correspondence of the address is immediately known by the communication mechanism for notifying the change and temporarily storing the change, and the correspondence of the old address is made at that timing. Destroy the table. Then, the communication mechanism restarts communication immediately after the switching process based on the new correspondence.

Further, the redundant communication paths are provided, different logical addresses are assigned to the respective paths, the address assigned to one of the paths is defined as the main communication path, and failure avoidance for the main communication path is guaranteed. First, when a failure on the main communication path is discovered, the path switching is notified to the same mechanism on all other computers. After synchronizing the computers, the addresses of the failed and unfailed routes are exchanged in all computers at once. After exchanging the addresses, since the main communication path address is assigned to the unfailed path, the communication can be continued.

Further, a network having redundant communication paths, a communication device control mechanism for dynamically relaying communication between the networks, and a logical address that can be dynamically changed independently of the communication device are used. In this way, failure avoidance on the communication path is guaranteed. When a failure occurs in this mechanism, the relay mechanism conventionally incorporated in the communication mechanism automatically sends the communication using the alternative communication path. With this mechanism, the receiving side can judge the communication addressed to itself regardless of the physical address of the received communication device and avoid the failure.

Further, the mechanism of the present invention receives the communication to the logical address attached independently of the communication device,
The logical address on the receiving side transmits information about which computer should be sent to reach the target address by using the existing relay function. When one computer performing communication wants another computer to substitute for communication, it moves the logical address used for communication to the computer for substitution. The relay function is executed in response to the movement of the logical address, and when the transmitting side attempts transmission to the target address, the relay function allows communication to the communication device of the target host computer. Before the substitution, the communication is performed to the computer that originally used the logical address, and after the substitution, to the computer that takes over. Here, it is understood that the computer that has received the communication is for itself because the logical address assigned to it independently of the communication device matches the communication destination.

Further, by providing the virtual address allocation mechanism of the computer providing the service with a mechanism for judging the state of the load fluctuation, the load fluctuation occurs on the service providing computer, and the response computer is dynamically operated by the optimum load distribution. Even after the new decision is made, the application designating the service providing computer by the virtual address can continue to receive the target service.

[0038]

【Example】

Embodiment 1 FIG. A first embodiment of the present invention will be described with reference to FIG. In the figure, 5 and 6 are test communication mechanisms, which monitor the interruption of the reception of notifications for network failure determination. Here, the test communication mechanisms 5 and 6 differ from the conventional test communication mechanism 52 in that a mechanism for stopping the transmission of the notification is added, and the test communication mechanism 5 transmits the notification. Is different from that it has stopped sending.

Next, the operation will be described. In order to detect a failure of a communication path or a communication device, a computer 1,
2 regularly communicates with computers 1, 2, 3,
It is assumed that each of the four monitors the reception interruption. Here, when the interruption of reception occurs, it is determined that a failure has occurred in the communication path. The administrator specifies in advance which of the test communication mechanisms 5 and 6 will perform the notification transmission as initial setting information in each computer. Conventionally, since all the machines transmitted and received the test communication between the n computers, communication on the _n C ₂ path was required. In this embodiment, by limiting the transmission mechanism to two, 2 × (n−1) types of transmission paths are sufficient. Thus, even if the number of transmitters is limited to two, the receiving function and communication paths of all computers can be monitored. As a result, it has become possible to suppress network communication capacity consumption while maintaining the accuracy of failure detection.

Example 2. A second embodiment of the present invention will be described with reference to FIG. Consider a case where only the computers 1 and 2 are transmitting in the configuration of FIG. Here, if it is assumed that the computer 1 is stopped due to the occurrence of a failure, maintenance, or the like, the number of computers that transmit a message is only 2. FIG. 2 shows such a state after the computer 1 is stopped. As a result, since no notification is sent to the computer 2 from anywhere, the failure can no longer be detected on the computer 2. In this embodiment, the computers 1, 2, 3,
4 always keeps track of the total number of notification transmitting mechanisms, and when the computer 1 is stopped, it is determined that the notification transmission amount is insufficient, and the test communication mechanisms 6 and 8 on the computers 3 and 4 start transmitting the notifications. . Further, when the test communication mechanisms 6 and 8 start transmission at the same time, the number of transmissions becomes 3 and the transmission becomes redundant.
Therefore, adjustment is made between the test communication mechanisms 6 and 8, and one of them stops the notification transmission, thereby guaranteeing the number of transmissions of two. Eventually, 6 stops the transmission and the test communication mechanism 8 continues the transmission, so that the notification is transmitted in the direction of the arrow in FIG. In this way, at least two message transmissions are secured and the amount of communication on the network is reduced. In this way, according to the present embodiment, even when the number of computers for test transmission changes due to a stop due to an abnormality or the like,
By dynamically securing the required minimum amount of notification transmission amount by the test notification mechanism, it becomes possible to suppress and adjust the communication channel capacity consumed by the notification transmission.

Example 3. A third embodiment of the present invention will be described with reference to FIG. FIG. 3 corresponds to the internal structure of each of the computers 1, 2, 3, 4 of the second embodiment, and in the figure, 14 is a communication mechanism 12 having a retransmission function.
Reference numeral 15 denotes a communication path failure detection mechanism for transmitting / receiving a test report without passing through the communication path, and reference numeral 15 denotes a flow of data transmitted / received by the communication path failure detection mechanism 14.

The communication path failure detection mechanism 14 periodically communicates with a resending communication mechanism mounted in another computer to check the operating status of the remote host computer and the presence / absence of a network failure. If communication with all computers in the network is lost, it is determined that the network has failed. If some of the plurality of computers do not respond, it is determined that the computer is out of order or stopped. The failure information determined by the communication path failure detection mechanism 14 is exchanged between other operating computers to identify the failure location. When a failure occurs in the network communication path 10 or the communication device 9, the communication mechanism 12 having the retransmission function and the communication path failure detection mechanism 14 know the occurrence of the failure almost at the same time. After that, the communication mechanism 12 having the re-sending function tries to restart the communication for a certain period of time, so that the communication software 11 is notified of the failure later. As described above, according to the present embodiment, if the communication path failure detection mechanism 14 takes measures to avoid the failure at a stage when the communication software 11 has not notified the error, the communication software 11 is considered to have a failure before the communication mechanism 12 having the retransmission function. Communication can be resumed. As a result, the failure can be avoided from the communication software 11 without interrupting the communication.

Example 4. A fourth embodiment of the present invention is shown in FIG.
This will be described with reference to FIG. In the figure, reference numeral 21 denotes an address correspondence table forced initialization mechanism provided by this embodiment, which initializes the table 20. The communication devices 9 and 16 are
A physical address is held as an identifier for distinguishing it from other devices on the communication path. This differs from communication device to communication device and is unique over the network. The physical addresses assigned to the communication devices 9 and 16 are PA9 and PA16.
, And the logical addresses are called LA9 and LA16.

Consider a case where the logical address LA16 attached to the communication device 16 is changed to LA100 under the conventional mechanism. In this case, since the LA 16 has been assigned to the communication device 16, the correspondence between the LA 16 and the PA 16 remains on the table 20. However, since the communication device 16 is no longer the LA 16, the communication device 16 no longer receives the communication addressed to the LA 16 even if the communication device 16 transmits to the PA 16. Further, even if a new device using the address LA16 appears, the transmission to the LA16 is directed to the PA16 according to the information in the table 20. This situation continues until the correspondence between LA16 and PA16 in table 20 is cleared. Here, if the mechanism that has performed the address exchange of the communication device 16 notifies the address exchange and the address correspondence table forced initialization mechanism 21 that has received the notification clears the table 20, a procedure for detecting the device corresponding to the LA 16 is performed. Will be continued. As described above, according to the present embodiment, the delay with the communication partner having the new correspondence relationship is delayed without waiting for the deadline for the old information relating to the logical and physical address correspondences stored in the table to automatically disappear. It becomes possible to communicate without using it.

Embodiment 5 FIG. The fifth embodiment of the present invention is shown in FIG.
This will be described with reference to FIG. In the figure, communication devices 9, 24, 1
6 and 25 have logical addresses LA9 and LA2, respectively.
4, LA16, LA25 are attached.
Here, when the computers 1 and 2 communicate using the network communication path 10, it is considered that communication between the logical addresses LA9 and LA16 is performed in each computer. Further, it is assumed that the mechanism corresponding to the third embodiment is used to realize early detection of a communication path failure in the computers 1 and 2.

When a failure occurs on the network communication path 10 and the communication between the communication devices 9 and 16 is interrupted, the communication failure avoidance mechanisms 26 and 27 use the logical addresses LA9 and LA24.
And LA16 and LA25 are exchanged. After the logical address exchange, the communication between the logical addresses LA9 and LA16 is
Since it is assigned to the hardware of the communication device that has not failed, communication can be resumed. In order to switch all the communication on the network communication path 10 to the communication path 23, the logical address exchanging work must be performed simultaneously by all the computers on the network. Therefore, the computer that discovers the failure of the communication path notifies the other computers of the address exchange. Here, the notification operation is repeated several times in order to prevent the notification from being missed, and the notified computer relays the notification to other computers.
The notification information also includes information for synchronizing the address exchange timing. When the notification for the address exchange is spread on the network and the logical address exchange is completed, the addresses LA9 and LA16 are assigned to the communication device 2 respectively.
4,25. After the address exchange, the communication device attempts to re-execute the communication using the addresses LA9 and LA16, and the communication device 2 through the network communication path 23.
Communication is performed between 4 and 25. In this way, the failure of the communication path 10 is avoided.

Next, FIG. 6 shows a method of using the communication failure avoidance mechanisms 26 and 27. In the figure, 28 is a communication application, 29 is a communication library, 30 is an OS kernel,
31 is a communication failure avoidance mechanism 26, 2 which is a feature of this embodiment.
7 is a communication path switching mechanism that operates based on 7. Conventionally, the communication application 28 or the communication library 29 detects an error and switches the communication path. In this embodiment, the communication path switching mechanism 31 uses the OS kernel based on the communication failure avoidance mechanisms 26 and 27. The association of communication devices viewed from 30 is automatically exchanged. With such a switching mechanism, the communication application can operate without changing the communication path without being aware of the switching of the communication path.

The inside of the communication path switching mechanism 31 is shown in FIG. In the figure, 14 is a mechanism for detecting a communication path failure,
Reference numeral 32 is communication data exchanged with the OS kernel, 33 is a communication path switching instruction signal, 34 is a switch portion for switching the data flow according to an instruction from the mechanism 14 for detecting a communication path failure, and 35 and 36 are physical communication devices, respectively. Indicates the communication path to and from. The data changeover switch portion 34 transfers the input / output data 32 with the kernel to the communication path 35,
It is connected to any one of the 36. When a failure occurs in the communication path, the failure detection mechanism 14 detects it and outputs a communication path switching instruction signal 33, and based on this, the communication paths 35, 3
One of the paths 6, which is operating normally, is connected to the kernel input / output data 32. According to the present embodiment, conventionally, it was necessary for each application to determine which of the communication paths 10 and 23 should be used to avoid a failure, but by providing the communication failure avoidance mechanisms 26 and 27,
Appropriate communication paths can be selected without the application having to be aware of it.

Example 6. A sixth embodiment of the present invention is shown in FIG.
A description will be given with reference to FIG. This embodiment is the same as the fifth embodiment with respect to FIGS. 5 and 6, but is different in the method of realizing the communication path switching mechanism of FIG. 6, and this state is shown in FIG. In the figure, 37 is a main mechanism in this embodiment, which is a virtual address receiving mechanism having a communication address independent of a physical communication device, 38 and 39 are communication data flows through the communication devices 9 and 24, and 40 is an OS kernel. And the virtual address receiving mechanism 37. Reference numerals 41 and 42 are data flows that pass through the virtual address receiving mechanism 37 to the communication devices 9 and 24. It is assumed that a mechanism for early detection of a communication path failure corresponding to the third embodiment is realized in the computers 1 and 2.

In FIG. 5, the computers 1 and 2 have virtual logical addresses VLA26 and VLA27 separately from the logical addresses LA9, LA24, LA16 and LA25 assigned to the communication devices 9, 24, 16 and 25, respectively, and have communication failures. It is assigned to the avoidance mechanisms 26 and 27. O
The S kernel has a function of exchanging route information, and can communicate from the communication devices 9 and 24 via the virtual address VLA 26. The communication devices 16 and 25 can communicate with the virtual address VLA 2
Communication is possible via 7, and the communication results are transmitted on the network. Next, the operation inside the communication path switching mechanism 31 in the computer 1 will be described with reference to FIG. The communication using the logical addresses LA9 and LA24 at the time of reception passes to the OS kernel by the flow of the data paths 38 and 39. When the virtual logical address VLA 26 is used, the data received from the communication device 9 is transmitted to the data path 41, the virtual address receiving mechanism 37, and the data path 40, and O
Cross the S kernel. The data received from the communication device 24 is transmitted through the data path 42, the virtual address receiving mechanism 37, and the data path 40, and reaches the OS kernel. The virtual address VLA 27 in the computer 2 can receive data from either the communication device 16 or 25 by the same mechanism. Here, the mechanism of receiving the virtual logical address is shown in FIG.
This will be described with reference to FIG. In the figure, reference numeral 37 is a virtual address receiving mechanism capable of allocating a plurality of logical addresses, looks like a communication device equivalent to the communication device 9 to the OS kernel 46, and sends back communication data to the OS kernel 46. Reference numeral 47 indicates the flow of data sent from the outside to the address assigned to the virtual address receiving mechanism 37. A logical address is assigned to the communication device 9 of the computer 1 in the same manner as a normal usage. This address is L
A9. Also, a virtual logical address V different from LA9
The LA 37 is prepared and assigned to the virtual address receiving mechanism 37. Further, in order to reach the virtual logical address VLA 37 as the route information to the outside connected through the network communication path 10, the fact that the communication device 9 is in charge of relay is transmitted. Virtual address VLA from outside via communication path 10
Communication data for 37 is fetched through the route of 47.
Here, first, the communication device 9 receives, and in the OS kernel 46, communication for the virtual logical address VLA 37 is first relayed to the virtual address receiving mechanism 37. Up to this point, the function of the conventional relay function is not replaced. After that, the virtual address receiving mechanism 37 that has received the communication
By returning the same content to the S kernel 46, the OS kernel 46 succeeds in receiving the communication for the virtual address VLA 37. The virtual logical address assigned to the virtual address receiving mechanism 37 can be dynamically changed, and two or more virtual addresses can be assigned at the same time. When communicating by this method, a virtual logical address V
The LA 26 and VLA 27 are used as communication addresses.
It is understood that when attempting transmission from the computers 1 to 2, it is sufficient to send the communication for the virtual address VLA 27 to the communication device 16 or 25 on the computer 1. When the computer 2 receives data addressed to the virtual address VLA 27, whichever of the communication devices 16 and 25 receives it, the data can be received through the virtual address receiving mechanism 37. When a failure occurs on the communication path 23, the communication path 23 is notified of the failure by the function of the conventional path control mechanism, and a new communication path is automatically selected while avoiding the failure point of the communication path automatically. With the new route information,
It is considered that the communication device 9 can reach the virtual logical address VLA 26 and the communication device 16 can reach the virtual logical address VLA 27, and the communication is continued between the communication devices 9 and 16. As a result, even if a failure occurs in one of the duplicated paths, the path avoidance mechanism is guaranteed because the path control mechanism works so as to use the other communication path. According to the present embodiment, even if a failure occurs in one of the duplicated paths, the path information notifies the alternative path, so that failure avoidance is guaranteed and communication for failure determination is performed. The amount can be reduced.

Example 7. A seventh embodiment of the present invention is shown in FIG.
0 and FIG. 11 will be described. In this embodiment, the operation of the server status reporting mechanism 56 of FIG. 18 in the conventional example is improved to be a communication partner dynamic switching mechanism 49, and the client status detecting mechanism 57 is omitted. Each of the communication devices 9, 16 and 48 in the figure has a logical address LA.
It is assumed that 9, LA16 and LA48 are attached.
The computer 1 operating as a server has a virtual logical address VLA1 in addition to the logical address assigned to the communication device 9. Server computer 1, 2
In the meantime, the dynamic switching mechanism 49 exchanges information with each other on a regular basis to ensure that the virtual logical address VLA1 can always be used by either one of the server computers 1 and 2. Since the OS kernel exchanges information for route control, the communication device 9 sends the virtual logical address VLA1.
Information reachable to can also be transmitted on the network. On the other hand, the application on the client computer 3 specifies the virtual logical address VLA1 as the communication partner. Here, the inside of the server computer 1 is shown in FIG.
It is shown in FIG. In the figure, 46 is an OS kernel, 43 is a flow of communication data using the logical address LA9, and 47 is a flow of communication data using the virtual logical address VLA1. Further, 37 is a virtual address receiving mechanism for receiving communication addressed to the virtual logical address VLA 1, and 50 is a virtual address receiving mechanism 37 while exchanging information between the server computers 1 and 2.
A mechanism for controlling allocation of virtual addresses to
Is a control instruction signal for the virtual address receiving mechanism 37. Here, 47, 37, 50 and 51 correspond to the dynamic switching mechanism 49 added according to this embodiment. The virtual logical address VLA is received by the virtual address receiving mechanism 37.
A data flow 47 using 1 has been realized. In addition, communication to the virtual logical address VLA1 is performed by the OS kernel 46.
Is relayed to another host computer via the route 10 by the operation of the conventional route control mechanism. The server computer 1 has a virtual logical address V
Stop using LA1 and use server computer 2 instead
Using the virtual logical address VLA1 enables dynamic switching of the communication path. When an instruction to move the VLA 1 is issued in the process of requesting a service from the server computer, the virtual address allocation control mechanisms 50 of the server computers 1 and 2 negotiate with each other, and the virtual address receiving mechanism 37 in the server computer 1 operates. Then, the use of the virtual logical address VLA1 is stopped and the virtual address receiving mechanism 3 in the server computer 2
At 7, it is decided to newly use the virtual logical address VLA1. As a result, the data flow 47 is stopped in the server computer 1, and the path 47 using the virtual logical address VLA1 can be used in the server computer 2. It means that the server computer 1 no longer uses the virtual logical address VLA1 and that the computer 2 newly acquires the virtual logical address VLA1.
It is transmitted to the external computer by the function of the S kernel 46. After that, the communication from the client computer 3 to the VLA 1 is sent to the communication device 16 and is received by the server computer 2. According to the present embodiment, as described above, it is possible to realize the dynamic switching of the communication partner without adding an extra mechanism on the computer 3 that operates as a client. Further, by adding a mechanism for the server computer 2 to automatically take over the virtual logical address VLA1 when the server computer 1 is stopped, it is possible to realize automatic switching of the communication partner by stopping the host.

Embodiment 8 FIG. An eighth embodiment of the present invention will be described with reference to FIGS. This embodiment enables the virtual address receiving mechanism 37 in the seventh embodiment to hold a plurality of virtual logical addresses, and the virtual address allocation control mechanism 50 uses the load between server computers as a trigger for switching virtual virtual addresses. A detection / comparison mechanism is added. It is assumed that the server computers 1 and 2 receive the processing request through the network and execute the service, while the client computer 3 requests the processing request from either the server computer 1 or 2. The server computers 1 and 2 are provided with a plurality of dynamically moveable virtual logical addresses shown in the seventh embodiment, and VLA1, VLA2, VLA3, VLA4.
Name it ... The application software requesting processing always communicates with the movable virtual logical address to request the request. The address allocation control mechanism 50 exchanges information between the server computers 1 and 2, and guarantees that these virtual addresses are always allocated to the address allocation mechanism 37 of either one of the server computers 1 and 2. If the load on the server computers 1 and 2 is equal, either one may undertake the processing request. Here, the client computer 3
Consider the case in which the first request occurs from the address VLA1 to the address VLA1, and the server computer 1 to which the VLA1 is assigned at the time of the request takes over. By accepting this request,
The load on the server computer 1 increases. The address assignment control mechanism 50 detects an increase in the load on the computer 1 and compares the load with the mechanism 50 in the server 2. As a result, assume that the server computer 2 has been selected as a suitable computer for accepting the second request. Then, the address assigning mechanism 50 removes the remaining VLA2, VLA3, VLA excluding the VLA1 being used by the mechanism 37 of the server 1 in the movable virtual logical address so that the server computer 2 can receive the next request.
4, ... Are assigned to the server computer 2. When the second request is generated in the client computer 3 and the communication destination thereof points to the VLA 3, the VLA 3 is assigned to the server computer 2, so the server computer 2 accepts the request. Since the load balance varies between the server computers 1 and 2 due to the second request processing, the address assignment control mechanism 50
Compares the difference in load again and reallocates unused virtual logical addresses VLA2, VLA4, ... Which can be moved to a lower load. As described above, according to the present embodiment, the address allocation mechanism determines the load state of each computer even when the load change occurs due to the end of the requested job or some change in the status in the server computers 1 and 2. Since unused virtual addresses are reassigned as appropriate, the load balance between the server computers 1 and 2 can be adjusted.

[0053]

Since the present invention is constructed as described above, it has the following effects.

According to the present invention, even if the number of transmissions is limited to two, all computers can be received and the communication path can be monitored. Therefore, while maintaining the accuracy of failure detection, the network can be used. Communication capacity consumption can be suppressed.

Further, even when the number of computers to be transmitted changes due to a stop due to an abnormality or the like, the notification mechanism dynamically secures the required minimum amount of notification transmission amount, thereby suppressing the communication channel capacity to be consumed. Can be adjusted.

Further, since the communication path failure detecting mechanism is adapted to avoid the failure at the stage when the error has not been notified to the communication software yet, the communication is restarted before the communication mechanism having the resending function judges the failure. As a result, the communication software can avoid the failure without interrupting the communication process.

Further, when the correspondence between the logical address and the physical address stored in the table is changed, the address correspondence compulsory initialization mechanism is provided so that the old address information does not have to be automatically erased. However, it is possible to communicate with the communication partner without delay based on the new correspondence.

Further, in the multiplexed network, it is necessary for each application to determine a communication path in order to avoid a failure. However, by providing a communication failure avoidance mechanism, the application is appropriate without any particular awareness. It is possible to select various communication paths.

Further, since a virtual address receiving mechanism is provided and a plurality of virtual logical addresses are dynamically assigned to the receiving mechanism, even if one of the routes fails,
The alternative route can be freely switched based on the route information, and the communication partner computer can be dynamically switched.

Since the virtual logical address takeover mechanism is provided, even if the first computer is stopped, the virtual address is automatically taken over by the second computer, and the host computer is stopped. You can automatically switch the communication partner without.

Further, since the address allocation mechanism judges the load status of each computer and reallocates an unused virtual address as appropriate, when a load change occurs due to some situation change in each computer. Also, there is an effect that load distribution and load balance can be adjusted among a plurality of computers.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a first embodiment of the present invention.

FIG. 2 is a configuration diagram showing a second embodiment of the present invention.

FIG. 3 is a configuration diagram showing a third embodiment of the present invention.

FIG. 4 is a configuration diagram showing a fourth embodiment of the present invention.

FIG. 5 is a configuration diagram showing a fifth embodiment of the present invention.

FIG. 6 is a flow chart showing the operation of the communication failure avoidance mechanism in the fifth exemplary embodiment of the present invention.

FIG. 7 is an operation explanatory diagram of a communication switching mechanism according to a fifth embodiment of the present invention.

FIG. 8 is an operation explanatory diagram of a communication switching mechanism according to a sixth embodiment of the present invention.

FIG. 9 is an explanatory diagram of a virtual logical address receiving mechanism according to the sixth embodiment of the present invention.

FIG. 10 is a diagram showing seventh and eighth embodiments of the present invention.

FIG. 11 is a diagram showing seventh and eighth embodiments of the present invention.

FIG. 12 is a conventional system configuration diagram.

FIG. 13 is a device configuration diagram of a conventional system.

FIG. 14 is a diagram showing an address conversion method in a conventional system.

FIG. 15 is a diagram showing a duplicated communication path in a conventional system.

FIG. 16 is a diagram showing a hierarchical structure of a conventional communication system.

FIG. 17 is a flowchart of communication path switching in the conventional system.

FIG. 18 is a diagram showing one form of network connection in a conventional system.

FIG. 19 is a flowchart showing the operation of the state detection mechanism in the conventional system.

FIG. 20 is an operation flowchart for realizing load distribution in the conventional system.

[Explanation of symbols]

1, 2, 3, 4 Computer, 5 Test communication mechanism for transmitting / receiving, 6 Test communication mechanism for stopping transmission only, 7
Test communication mechanism that stopped operation, 8 Test communication mechanism that started transmission, 9 Communication device for using communication path 10 from computer 1 10 Network communication path, 11
Software for communication, 12 a communication device control mechanism having a retransmission function, 13 data passing through the communication device 9 from the software 11 via the mechanism 12, 14 communication path failure detection mechanism, 15 communication device 9 without passing through the mechanism 12 And mechanism 1
Data exchanged between 1; 16 communication device for using the communication path 10 from the computer 2; 17 application to send; 18 communication mechanism; 19 17 to 18
Data transmitted via the, 20 destination physical address, logical address correspondence table, 21 mechanism for forcibly initializing physical address, logical address correspondence table, 22 operation for initializing 20 from mechanism 21, 23 Network communication path, 24, 25 communication device, 26 communication failure avoidance mechanism, 27 communication failure avoidance mechanism, 28 communication application, 29 communication library, 30 OS kernel, 3
1 communication path switching mechanism, 32 flow of input / output data with OS, 33 communication path switching instruction, 34 communication data switching switch, 35, 36 data path with communication device, 37 virtual address receiving mechanism, 38, 39 Data flow between kernel and communication device, 40 Data flow between kernel and virtual address receiving mechanism 37, 41, 42
Data flow between the mechanism 37 and the communication device, 43 communication data, 44 virtual communication mechanism for returning communication to itself,
45 communication data returning to itself, 46 OS kernel, 47 flow of communication data using virtual address, 48 communication device, 49 dynamic switching mechanism of communication partner, 50 virtual address allocation to virtual address receiving mechanism Control mechanism, 51 control of mechanism 37 by mechanism 50, 52 test communication mechanism, 53 conventional test communication mechanism,
54 data from the test communication mechanism 53 through the communication device 9 via the mechanism 12, 55 communication path switching logic, 56
Server status reporting mechanism, 57 Client status detecting mechanism.

Claims (8)

[Claims] 1. A plurality of computers connected to a network, wherein the computer includes a test communication mechanism having a communication path failure detection and notification transmission execution / stop function, and the notification transmission held by the test communication mechanism. A communication path dynamic control method characterized in that the communication capacity of the network required for test communication is reduced by limiting the number of transmitting computers based on the execution and stop functions of. 2. The test communication mechanism is capable of maintaining the failure determination accuracy without reducing the test communication amount when the part of the computers connected to the network is stopped. Dynamic control method of the communication path described. 3. A plurality of computers connected to a network, wherein the computer has a communication control mechanism having a communication protocol having a retransmission function, and a communication failure realizing a high-speed communication protocol function without the retransmission function. The communication failure detection mechanism, which has a detection mechanism and realizes the above-mentioned high-speed communication protocol function, notifies the failure avoidance mechanism of a failure previously detected on the communication path, so that the communication control mechanism having the retransmission function can communicate. A dynamic communication path control method characterized in that an alternative communication path can be provided at a timing at which it can be restarted. 4. A plurality of computers connected to a network, wherein said computer has a communication mechanism that realizes dynamic route control, and address correspondence that dynamically checks the correspondence between logical addresses and physical addresses and notifies the change in correspondence. A confirmation mechanism, an address correspondence table for temporarily storing the correspondence result of the physical address and the logical address, and an initialization mechanism for forcibly initializing the route information defined in the address correspondence table are provided, and the communication route has changed. In this case, the communication path dynamic control method is characterized in that the old address correspondence information is initialized without waiting for the regular update timing so that it can be used immediately when the path information changes. 5. A plurality of computers connected to a network, wherein the computer includes a communication failure avoidance mechanism, and the communication failure avoidance mechanism switches a communication path for connecting a communication device corresponding to a logical address. A dynamic control system of a communication path, which realizes the mechanism in an operating system, thereby making it possible to switch the communication path from an application utilizing the operating system function without individually incorporating the path switching logic. 6. A plurality of computers connected to a network, wherein the computer includes a communication failure avoidance mechanism to which a virtual logical address independent of a logical address is assigned, and communication path switching constituting the communication failure avoidance mechanism. The mechanism has a virtual address receiving mechanism, so that even if a failure occurs in the network route, the alternate route can be automatically selected by designating the communication partner by the virtual address. Dynamic control system. 7. In a system in which a plurality of computers that provide the same service and a computer that uses the service are connected to a network, the service providing computer holds a virtual address independent of a logical address, A virtual address receiving mechanism for receiving virtual addresses and a virtual address allocation control mechanism for allocating the above virtual addresses are provided, and even if the computer providing the service is switched, the computer using the service continues without being aware of the switching. A communication path dynamic control method characterized in that the service can be provided through the communication. 8. In a system in which a plurality of computers that provide the same service and a computer that uses the service are connected to a network, the address allocation mechanism has a function of determining a load state associated with the provision of the service. The load balance can be adjusted without adding a function to the service using computer by reallocating the virtual address to a computer having a low load when a load change occurs. 7. A communication path dynamic control method described in 7.