JPH09252331A - Fault-tolerant communication control method and backup path configuration method - Google Patents

Abstract

(57) [Abstract] (Correction) [PROBLEMS] Development of dedicated hardware with shared memory that enables fault-tolerant communication control without the need for dedicated hardware with a process takeover mechanism that uses shared memory. It prevents the development period from increasing and the system price increase, and realizes porting to computer systems with different architectures because it is realized by software control without depending on dedicated hardware. A communication management function e3 and line control devices e4 and e
5 are connected to a plurality of routes r3, r4, r5 and a switch e6 that can be dynamically switched. Delivery confirmation is performed between the communication management function e3 and the line control devices e4, e5, and the backup route is dynamically changed when the working route fails. Information that has not been confirmed to be delivered, that is, information that was present on the faulty working route and has been lost is retransmitted from the backup route, and the information lost and retransmitted by the communication management function and line controller is retransmitted. Information duplication.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fault-tolerant computer system having a plurality of routes that can be switched between a communication management function and a line control device, and switches to a backup route without interrupting communication when a working route fails. The present invention relates to a communication control method and a backup route configuration method for dynamically selecting a backup route that can be used when a working route fails.

[0002]

2. Description of the Related Art A conventional fault-tolerant communication control method is disclosed in, for example, Japanese Unexamined Patent Publication (Kokai) No. 4-360242, "A system switching device for a redundant system and its method".
There is. In this system, it is composed of pre-fixed paths of the active system and the standby system and a shared memory that can be referenced by both systems, and message identification information being processed by the active system is always set in the shared memory. Normally, communication is performed while always writing the message identification information being processed in the active system to the shared memory, and when the active system fails, the standby system reads the message identification information being processed set in the shared memory. The restart point was decided and communication was continued.

[0003]

In the above-mentioned conventional method, the development period is prolonged by the development of the dedicated hardware having the shared memory, the system price is increased, and the route is the dedicated hardware. There are problems such as difficulty in porting to computer systems with different architectures. In addition, the backup route is in advance one-to-one with the working route.
The number of backup routes is the same as the number of active routes because it is fixed in the above, and when both the combined active route and backup route fail, resources cannot be used even if other backup routes exist. There was a problem that the utilization efficiency of was poor.

[0004]

In order to solve the above problems, a computer system to which the present invention is applied has a communication management function, a line control device, a plurality of routes between the communication management function and the line control device, and a plurality of routes. Is composed of a switch that can be dynamically switched, and delivery confirmation is performed between the communication management function and the line control device. Information that was not confirmed to be delivered due to a failure of the working route, that is, information that was lost on the failed working route was retransmitted from the standby route, and the information lost and retransmitted by the communication management function and line controller was retransmitted. By preventing the duplication of the provided information, the communication continues without interruption. In addition, since it does not depend on the combination of the active route and the standby route, it is possible to dynamically select and switch any spare route that can be used when the active route fails. A configuration can also be realized.

[0005]

1 is a diagram for explaining a fault-tolerant communication control method. The computer system S1 to which this fault-tolerant communication control method is applied has a communication management function e.
1, a line controller e2, a working route r1 between the communication management function e1 and the line controller e2, and a spare route r2. The line control device e2 shares a lower protocol of the communication protocol and performs protocol control asynchronously with the communication management function e1. The interface between the communication management function e1 and the line control device e2 consists of a command requested from the communication management function e1 to the line control device e2 and a completion report of the command. Is set to the sequence number of the requested command and the sequence number of the command for which completion is reported, that is, the completion sequence number. Hereinafter, the combination of the request sequence number x and the completion sequence number y will be expressed as (x, y).

The communication management function e1 sends commands cm10, cm20, cm3 to the line controller e2 via the route r1.
0 is requested continuously. At this time, assuming that the initial states of the request sequence number and the completion sequence number are (0, 0), the request sequence numbers are sequentially added, and the command cm10 is (1, 0), cm.
20 is (2,0) and cm30 is (3,0).

The line controller e2 executes the command cm10, stores the request sequence number 1 as "executed", and then executes the route r.
The completion cf10 of the command cm10 is reported to the communication management function e1 via 1. As a result, the communication management function e1
Will be 1 which is the request sequence number of the command cm10.

The line controller e2 executes the command cm20, stores the request sequence number 2 as "executed", and then executes the route r.
Although the completion cf20 of the command cm20 is reported to the communication management function e1 via 1, the failure occurred on the route r1 and the report cf20 and the unexecuted command cm30 disappeared.

Since the route r1 becomes unusable, the communication management function e1 sends commands cm20 and cm30 whose completion has not been reported to cm2 via the route r2.
The request is made again to the line control device e2 as 1, cm31. At this time, since the completion sequence number of the communication management function e1 is 1, the command cm21 is (2,1), the command cm31.
Becomes (3,1).

The line control device e2 receives the command cm21, but since the request sequence number is 2, it judges that the command cm21 has already been executed, and does not perform the actual processing, and sends the command cm21 to the communication management function e1 via the route r2. Report completion cf21. Further, the line control device e2 determines that the communication management function e1 has received the completion report cf10 because the completion order number of the command cm21 is 1, and the command cm10
End processing is performed.

The line controller e2 receives the command cm31, executes it because the request sequence number is 3, stores the request sequence number 3 as "executed", and then sends the command to the communication management function e1 via the route r2. cm31 completed cf31
Report.

In this way, the command between the communication management function e1 and the line control device e2 is designated with the requested sequence number and the sequence number of the command confirmed to be completed, and the delivery confirmation of the command and the command completion report is made between them. By doing so, it is possible to prevent omission and duplication of command execution, and it becomes possible to perform fault-tolerant communication control independent of the route.

FIG. 2 is a diagram for explaining a method of constructing a backup route. This preparatory route configuration method is based on the assumption that the fault-tolerant communication control method described in FIG. 1 is implemented. The computer system S2 to which this method is applied includes a communication management function e3, line control devices e4 and e5, and a communication management function e.
3 and the line control device e4. Routes r3, r4, r5, and a switch e6 that dynamically connects the routes r3, r4, r5 to the line control devices e4, e5.

In the initial state, the working route from the communication management function e3 to the line controller e4 is route r3, the working route to the line controller e5 is route r4, and the backup route is route r5.

When the route r3 fails, the communication management function e3 controls the switch e6 to dynamically select the route to the line controller e4 as the route r5 to continue the communication, resulting in a failure. When the route r3 becomes available, the communication management function e3 sets the route r3 as a backup route.

Next, when the route r4 fails, the communication management function e3 controls the switch e6 to control the line controller e5.
The route to is dynamically selected as the route r3 to continue the communication.

As described above, according to the fault-tolerant communication control of the present invention, the route switching independent of the route is realized, and as a result, the spare route that can be used when the working route fails is dynamically selected, and the alternative route at the time of failure is secured. In addition, it becomes possible to configure a backup route with the number of backup routes equal to or less than the number of active routes.

[0018]

As described above, according to the present invention, fault-tolerant communication control can be performed without requiring dedicated hardware having a process takeover mechanism using a shared memory, and dedicated hardware having a shared memory is provided. Since the development period of software is prolonged and the system price is prevented from rising, and it is realized by software control without depending on dedicated hardware, it is easy to port to computer systems with different architectures. In addition, this method is a fault-tolerant communication control method that confirms delivery by the end-to-end between the communication management function and the line control device, and does not depend on the intermediate route, so the combination of the working and protection routes is not fixed and the alternate route is activated. It is possible to eliminate the influence of double failure of the route, and it is possible to realize a configuration in which the number of backup routes is less than the number of active routes without fixing the set of active route and backup route, and effective use of resources is possible. Becomes

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing an embodiment of a fault tolerance control method.

FIG. 2 is an explanatory diagram showing an example of a method of configuring a backup path.

[Explanation of symbols]

S1 ... Computer system to which fault tolerance control method is applied, e1 ... Communication management function on main memory, e2 ... Line control device, r1 ... Communication management function e1 and line control device e2
A current route between them, r2 ... a spare route between the communication management function e1 and the line controller e2, cm10 ... a command with request sequence number 1, completion sequence number 0, cm20 ... request sequence number 2,
Command of completion sequence number 0, cm30 ... Request sequence number 3, command of completion sequence number 0, cm21 ... Command of request sequence number 2, command of completion sequence number 1, command content is cm
Same as 20, cm31 ... Command with request sequence number 3, completion sequence number 1, command content is the same as cm30, cf
10 ... Command cm10 completion report, cf20 ... Command cm20 completion report, cf21 ... Command cm21 completion report, cf31 ... Command cm31 completion report, S
2 ... Computer system to which backup path configuration method is applied, e3 ... Communication management function, e4 ... Line control device, e5 ...
Line control device, e6 ... Switch for connecting the routes r3, r4, r5 to the line control devices e4, e5, r3 ... Route, r4 ... Route, r5 ... Route.

Claims (2)

[Claims] 1. A computer system having a plurality of routes that can be switched between a communication management function and a line controller,
Confirms the delivery of information between the communication management function and the line control unit, and resends information that was not confirmed due to a fault in the working route, that is, information that was lost on the faulty working route from the backup route. However, the communication management function and line control device that have the function of preventing loss of lost information and duplication of retransmitted information will continue without interruption even when the route is switched, and will not depend on the route. A fault-tolerant communication control method characterized by the above. 2. A communication management function and a line controller, which implement a fault-tolerant communication control method independent of a route, are connected by a plurality of routes and a route switcher,
By operating the route switcher when the working route fails to enable dynamic selection of the backup route, it becomes easy to secure the alternate route at the time of failure, and the working route without fixing the set of the working route and the backup route. A method for configuring a backup route, which is characterized by realizing a configuration in which the number of backup routes is equal to or less than the number.